From nemo-bounces@ietf.org Mon Apr 03 05:58:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQLpK-0004iG-Dt; Mon, 03 Apr 2006 05:58:26 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQLpJ-0004iB-C8 for nemo@ietf.org; Mon, 03 Apr 2006 05:58:25 -0400 Received: from tama5.ecl.ntt.co.jp ([129.60.39.102]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQLpI-0001WJ-II for nemo@ietf.org; Mon, 03 Apr 2006 05:58:25 -0400 Received: from vcs3.rdh.ecl.ntt.co.jp (vcs3.rdh.ecl.ntt.co.jp [129.60.39.110]) by tama5.ecl.ntt.co.jp (8.13.6/8.13.6) with ESMTP id k339w59x026901; Mon, 3 Apr 2006 18:58:05 +0900 (JST) Received: from mfs3.rdh.ecl.ntt.co.jp (mfs3.rdh.ecl.ntt.co.jp [129.60.39.112]) by vcs3.rdh.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k339w4aK015360; Mon, 3 Apr 2006 18:58:04 +0900 (JST) Received: from nttmail3.ecl.ntt.co.jp ([129.60.39.100]) by mfs3.rdh.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k339w3Pi009574; Mon, 3 Apr 2006 18:58:03 +0900 (JST) Received: from eclscan3.m.ecl.ntt.co.jp (eclscan3.m.ecl.ntt.co.jp [129.60.5.69]) by nttmail3.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k339w3Bu013753; Mon, 3 Apr 2006 18:58:03 +0900 (JST) Received: from eclscan3.m.ecl.ntt.co.jp (localhost [127.0.0.1]) by eclscan3.m.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k339w3KL016456; Mon, 3 Apr 2006 18:58:03 +0900 (JST) Received: from imj.m.ecl.ntt.co.jp (imj.m.ecl.ntt.co.jp [129.60.5.234]) by eclscan3.m.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k339w2Cw016453; Mon, 3 Apr 2006 18:58:02 +0900 (JST) Received: from [129.60.10.37] ([129.60.10.37]) by imj.m.ecl.ntt.co.jp (8.13.4/8.13.4) with ESMTP id k339w11O007893; Mon, 3 Apr 2006 18:58:02 +0900 (JST) Date: Mon, 03 Apr 2006 18:58:16 +0900 From: Takaaki Moriya To: "Vijay Devarapalli" , Brian.Haley@hp.com, Subject: Re: [nemo] Fragmentation at mobile router In-Reply-To: References: X-Mailer-Plugin: BkASPil for Becky!2 Ver.2.063 Message-Id: <20060403185523.4FB3.MORIYA.TAKAAKI@lab.ntt.co.jp> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver. 2.24.02 [ja] X-Spam-Score: 0.0 (/) X-Scan-Signature: 244a2fd369eaf00ce6820a760a3de2e8 Cc: X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Vijay, Brian and all, Thanks for comments. Since RFC3963 does not refer to this issue, it seems to be outside of the scope of NEMO specification. Can we conclude it's implementation-dependent ? If so, TAHI interoperability guideline will not specify fragmentation at mobile router as well as sending ICMP packet too big by mobile router. Best Regards Takaaki Moriya moriya.takaaki@lab.ntt.co.jp On Fri, 31 Mar 2006 10:40:25 -0800 "Vijay Devarapalli" wrote: > > I have a question about fragmentation at mobile router. > > When mobile router receives a long packet from a node in the mobile > > network, should mobile router fragment this packet? > > I would prefer the mobile router to tell the > MNN to reduce MTU. > > > Of course, normally it is end node to manage fragmentation after > > receiving ICMP. However, if so, when the mobile network > > moves, MTU size > > toward tunneling path may be changed and then end nodes in the mobile > > network receive ICMP packet too big. > > (e.g. when mobile router moves under another mobile router, > > MTU size is > > changed.) > > It means that movement forses end node to retransmit the > > packet. It goes > > against the rule of hiding movement from end node. > > receiving ICMP packet too is not the same as > the MNN being aware of the movement. > > Vijay > > > > > > > The problem in mobile router is the same in home agent. > > > > > > Thanks in advance. > > > > --- > > Best Regards > > Takaaki Moriya > > > > > > > > --- From nemo-bounces@ietf.org Mon Apr 03 13:59:00 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQTKL-0003ea-RP; Mon, 03 Apr 2006 13:58:57 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQTKJ-0003bD-PS for nemo@ietf.org; Mon, 03 Apr 2006 13:58:55 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQTKI-0005Qh-Et for nemo@ietf.org; Mon, 03 Apr 2006 13:58:55 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Fragmentation at mobile router Date: Mon, 3 Apr 2006 10:58:51 -0700 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Fragmentation at mobile router Thread-Index: AcZXBSWBJJToB+XKTlaA8K/aUqc36wAQxIMg From: "Vijay Devarapalli" To: "Takaaki Moriya" , , X-Spam-Score: 0.0 (/) X-Scan-Signature: 4b800b1eab964a31702fa68f1ff0e955 Cc: X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > Thanks for comments. >=20 >=20 > Since RFC3963 does not refer to this issue, it seems to be outside of > the scope of NEMO specification. > Can we conclude it's implementation-dependent ? > If so, TAHI interoperability guideline will not specify=20 > fragmentation at > mobile router as well as sending ICMP packet too big by mobile router. sounds good to me. Vijay >=20 >=20 >=20 > Best Regards > Takaaki Moriya > moriya.takaaki@lab.ntt.co.jp >=20 >=20 >=20 > On Fri, 31 Mar 2006 10:40:25 -0800 > "Vijay Devarapalli" wrote: >=20 > > > I have a question about fragmentation at mobile router. > > > When mobile router receives a long packet from a node in=20 > the mobile > > > network, should mobile router fragment this packet? > >=20 > > I would prefer the mobile router to tell the > > MNN to reduce MTU. > >=20 > > > Of course, normally it is end node to manage fragmentation after > > > receiving ICMP. However, if so, when the mobile network=20 > > > moves, MTU size > > > toward tunneling path may be changed and then end nodes=20 > in the mobile > > > network receive ICMP packet too big. > > > (e.g. when mobile router moves under another mobile router,=20 > > > MTU size is > > > changed.) > > > It means that movement forses end node to retransmit the=20 > > > packet. It goes > > > against the rule of hiding movement from end node.=20 > >=20 > > receiving ICMP packet too is not the same as > > the MNN being aware of the movement. > >=20 > > Vijay > >=20 > > >=20 > > >=20 > > > The problem in mobile router is the same in home agent. > > >=20 > > >=20 > > > Thanks in advance. > > >=20 > > > --- > > > Best Regards > > > Takaaki Moriya > > > > > >=20 > > >=20 > > >=20 >=20 > --- >=20 >=20 From nemo-bounces@ietf.org Mon Apr 03 20:49:36 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQZjg-0002Yh-Er; Mon, 03 Apr 2006 20:49:32 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQZjf-0002Yc-RH for nemo@ietf.org; Mon, 03 Apr 2006 20:49:31 -0400 Received: from omgo.iij.ad.jp ([202.232.30.157]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQZjc-0007YS-A8 for nemo@ietf.org; Mon, 03 Apr 2006 20:49:31 -0400 Received: OTM-MO id k340nJk3018526; Tue, 4 Apr 2006 09:49:19 +0900 (JST) Received: OTM-MIX(otm-mix01) id k340nIFA009102; Tue, 4 Apr 2006 09:49:19 +0900 (JST) Received: from localhost (keiichi00.osaka.iij.ad.jp [192.168.64.45]) by jc-smtp.iij.ad.jp (JC-SMTP/jc-smtp) id k340nHi9013234; Tue, 4 Apr 2006 09:49:18 +0900 (JST) Date: Tue, 04 Apr 2006 09:50:59 +0900 (JST) Message-Id: <20060404.095059.24529705.keiichi@iijlab.net> To: moriya.takaaki@lab.ntt.co.jp Subject: Re: [nemo] Fragmentation at mobile router From: Keiichi SHIMA In-Reply-To: <20060403185523.4FB3.MORIYA.TAKAAKI@lab.ntt.co.jp> References: <20060403185523.4FB3.MORIYA.TAKAAKI@lab.ntt.co.jp> X-Mailer: Mew version 4.1.50 on Emacs 21.3 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Spam-Score: 0.0 (/) X-Scan-Signature: b19722fc8d3865b147c75ae2495625f2 Cc: nemo@ietf.org, Vijay.Devarapalli@AzaireNet.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi, From: Takaaki Moriya Subject: Re: [nemo] Fragmentation at mobile router Date: Mon, 03 Apr 2006 18:58:16 +0900 > Since RFC3963 does not refer to this issue, it seems to be outside of > the scope of NEMO specification. > Can we conclude it's implementation-dependent ? > If so, TAHI interoperability guideline will not specify fragmentation at > mobile router as well as sending ICMP packet too big by mobile router. Let me ask a clarification question. A mobile router have to send an ICMP too big, if the forwarded packet exceeds the MTU of the tunnel interface between the MR and its HA. Otherwise, we lose connectivity. A mobile router may not send a too big message if the packet is bigger than the physical path MTU (not the MTU for the tunnel i/f). It is implementation dependent. Do I understand correctly? BTW, is there any word about the MTU between an MN/MR and an HA? Our implementation uses the IPv6 minimum MTU (1280) for the interface, but I know some implementations use bigger MTU size, which may cause a Path MTU discovery problem. --- Keiichi SHIMA IIJ Research Laboratory WIDE Project From nemo-bounces@ietf.org Tue Apr 04 04:42:15 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQh73-000411-TY; Tue, 04 Apr 2006 04:42:09 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQh72-00040I-U3 for nemo@ietf.org; Tue, 04 Apr 2006 04:42:08 -0400 Received: from tama5.ecl.ntt.co.jp ([129.60.39.102]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQh71-00031p-3N for nemo@ietf.org; Tue, 04 Apr 2006 04:42:08 -0400 Received: from vcs3.rdh.ecl.ntt.co.jp (vcs3.rdh.ecl.ntt.co.jp [129.60.39.110]) by tama5.ecl.ntt.co.jp (8.13.6/8.13.6) with ESMTP id k348faWN006267; Tue, 4 Apr 2006 17:41:36 +0900 (JST) Received: from mfs3.rdh.ecl.ntt.co.jp (mfs3.rdh.ecl.ntt.co.jp [129.60.39.112]) by vcs3.rdh.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k348fZ2v011976; Tue, 4 Apr 2006 17:41:35 +0900 (JST) Received: from nttmail3.ecl.ntt.co.jp ([129.60.39.100]) by mfs3.rdh.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k348fYCp007440; Tue, 4 Apr 2006 17:41:34 +0900 (JST) Received: from eclscan3.m.ecl.ntt.co.jp (eclscan3.m.ecl.ntt.co.jp [129.60.5.69]) by nttmail3.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k348fYSP021744; Tue, 4 Apr 2006 17:41:34 +0900 (JST) Received: from eclscan3.m.ecl.ntt.co.jp (localhost [127.0.0.1]) by eclscan3.m.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k348fXQF004418; Tue, 4 Apr 2006 17:41:33 +0900 (JST) Received: from imj.m.ecl.ntt.co.jp (imj.m.ecl.ntt.co.jp [129.60.5.234]) by eclscan3.m.ecl.ntt.co.jp (8.12.11/8.12.11) with ESMTP id k348fXWj004413; Tue, 4 Apr 2006 17:41:33 +0900 (JST) Received: from [129.60.10.37] ([129.60.10.37]) by imj.m.ecl.ntt.co.jp (8.13.4/8.13.4) with ESMTP id k348fWt5013614; Tue, 4 Apr 2006 17:41:33 +0900 (JST) Date: Tue, 04 Apr 2006 17:41:46 +0900 From: Takaaki Moriya To: Keiichi SHIMA Subject: Re: [nemo] Fragmentation at mobile router In-Reply-To: <20060404.095059.24529705.keiichi@iijlab.net> References: <20060403185523.4FB3.MORIYA.TAKAAKI@lab.ntt.co.jp> <20060404.095059.24529705.keiichi@iijlab.net> X-Mailer-Plugin: BkASPil for Becky!2 Ver.2.063 Message-Id: <20060404172209.33D0.MORIYA.TAKAAKI@lab.ntt.co.jp> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver. 2.24.02 [ja] X-Spam-Score: 0.0 (/) X-Scan-Signature: c1c65599517f9ac32519d043c37c5336 Cc: nemo@ietf.org, Vijay.Devarapalli@AzaireNet.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Shima-san, Thanks for comments. > A mobile router have to send an ICMP too big, if the forwarded packet > exceeds the MTU of the tunnel interface between the MR and its HA. > Otherwise, we lose connectivity. It seems to be good to send ICMP too big. But I guess some MR implementations may fragment the packet instead of sending ICMP too big. Isn't it what Brian says ? On Thu, 30 Mar 2006 23:23:23 -0500 Brian Haley wrote: > The MR will encapsulate it to traverse the MR-HA > tunnel which will cause the outer IPv6 packet to be fragmented. > A mobile router may not send a too big message if the packet is bigger > than the physical path MTU (not the MTU for the tunnel i/f). It is > implementation dependent. I agree about it. > > Do I understand correctly? > > BTW, is there any word about the MTU between an MN/MR and an HA? Our > implementation uses the IPv6 minimum MTU (1280) for the interface, but > I know some implementations use bigger MTU size, which may cause a > Path MTU discovery problem. RFC does not refer to this issue including MTU. It seems to be outside of the scope of NEMO specification. --- Best Regards Takaaki Moriya From nemo-bounces@ietf.org Tue Apr 04 05:02:02 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQhQ1-0001ys-Ls; Tue, 04 Apr 2006 05:01:45 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQhQ0-0001yf-R8 for nemo@ietf.org; Tue, 04 Apr 2006 05:01:44 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQhQ0-0003Zp-Bu for nemo@ietf.org; Tue, 04 Apr 2006 05:01:44 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 04 Apr 2006 11:01:44 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3491b3K025270; Tue, 4 Apr 2006 11:01:41 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Tue, 4 Apr 2006 11:01:37 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Fragmentation at mobile router Date: Tue, 4 Apr 2006 11:01:25 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC02043A55@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Fragmentation at mobile router Thread-Index: AcZXga0aztW9ya7zRCupu3D7qoMMUgAQlyRQ From: "Pascal Thubert \(pthubert\)" To: "Keiichi SHIMA" , X-OriginalArrivalTime: 04 Apr 2006 09:01:37.0479 (UTC) FILETIME=[6155C570:01C657C6] X-Spam-Score: 0.0 (/) X-Scan-Signature: 244a2fd369eaf00ce6820a760a3de2e8 Cc: nemo@ietf.org, Vijay.Devarapalli@AzaireNet.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Keiichi >> Since RFC3963 does not refer to this issue, it seems to be outside of >> the scope of NEMO specification. >> Can we conclude it's implementation-dependent ? >> If so, TAHI interoperability guideline will not specify fragmentation at >> mobile router as well as sending ICMP packet too big by mobile router. > >Let me ask a clarification question. [Pascal] I'll describe what our implementation does. Now, I think it is more a matter of good sense than something that could vary a lot with implementations. In any case, the MR must do something so that the packets go through, don't you think? > >A mobile router have to send an ICMP too big, if the forwarded packet >exceeds the MTU of the tunnel interface between the MR and its HA. >Otherwise, we lose connectivity. > [Pascal] The MR can not fragment the packet that it will tunnel when it forwards it over the tunnel interface; but it can fragment the packet once it has been tunneled, because at this point, the MR IS the source of the MR to HA packet, so the rules are not broken. The HA reassembles the tunneled packet, receives it, and then it extracts the inner packet (MNN to CN) which does not have a fragment header. >A mobile router may not send a too big message if the packet is bigger >than the physical path MTU (not the MTU for the tunnel i/f). It is >implementation dependent. > [Pascal] For the MR, the tunnel is just another interface, with an MTU that it can compute based on the encapsulation overhead and packet too big it would receive because of recent history on MR to HA packets. The MR needs to reassess that MTU every so often and when it moves. If the MR does not send packet too big to the MNN, then it will have to fragment the encapsulated packets forever.=20 It makes sense to notify the MNN, unless the tunnel MTU gets below the min MTU of 1280 bytes. In a heavy case of nested NEMO, such a thing could happen. >Do I understand correctly? > >BTW, is there any word about the MTU between an MN/MR and an HA? Our >implementation uses the IPv6 minimum MTU (1280) for the interface, but >I know some implementations use bigger MTU size, which may cause a >Path MTU discovery problem. > [Pascal] Initially, we set the tunnel MTU to (egressMTU-overhead), and then react on packet too big that we might receive because of MR to HA packets. By default, egressMTU will be 1500 on Ethernet. If you force the tunnel MTU to 1280 from the start, then nested NEMO might be a problem for you. >--- >Keiichi SHIMA >IIJ Research Laboratory >WIDE Project From nemo-bounces@ietf.org Tue Apr 04 06:44:39 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQj1T-0006eZ-SL; Tue, 04 Apr 2006 06:44:31 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQj1S-0006eU-JA for nemo@ietf.org; Tue, 04 Apr 2006 06:44:30 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQj1S-0007RA-3i for nemo@ietf.org; Tue, 04 Apr 2006 06:44:30 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id E137B212C4E; Tue, 4 Apr 2006 13:44:27 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id A9D2B212C46; Tue, 4 Apr 2006 13:44:27 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 6B9C5BDC40; Tue, 4 Apr 2006 13:44:27 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 349C1BDC3D; Tue, 4 Apr 2006 13:44:27 +0300 (EEST) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Tue, 4 Apr 2006 13:44:33 +0300 To: "Vijay Devarapalli" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 25620135586de10c627e3628c432b04a Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "Davis, Terry L" , "T.J. Kniveton" , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi all, global HAHA draft states in section 5.1.1 External routing In return, each site advertises the aggregation that encompasses the Distributed Home Network aggregation back into the service provider network, to the outside world (the internet). So, the (distributed) home network prefix is announced through BGP and=20= this implies that each Home Network prefix will impact with an=20 additional route in the global BGP table, all agree? The question at this point is whether this is a reasonable model Let's see what the current practise in IPv6 is. As today, the reccommendation is that only ISPs can announce their=20 blocks in BGP. In particular, end sites are discouraged to do that, in particular,=20 multihomed sites have to find an alternative solution rather than rely=20= on BGP. Moreover, as today, there is no RIR policy that allows an end site to=20 get their own address space. Only Provider aggregatable address is=20 being handed out by the RIRs. Now, as is read this proposal, it implies that each distributed home=20 netwrok will need a prefix that is separately announced in the global=20 BGP table. Now, the question is if this is in line with current=20 tendency to consider global routing scalability as a top priority. As i see it, there must be a very important reason to push a solution=20 that relies on this approach, and i really would like to understand why=20= the support of globally distributed HAs requires such approach. I mean one could think of an approach to support gloablly distributed=20 HAs that do not require having the home network prefix injected=20 separately in the global BGP table. In particular, one could envision a solution where each HA has its own=20= prefix block, belonging to the ISPs aggregate and that the mobile node=20= prefers to use the address of the closest HA to initiate new=20 communications. This with a route optimization solution should provide=20= similar benefits. Regards, marcelo El 30/03/2006, a las 22:07, Vijay Devarapalli escribi=F3: > >> That's /8 or less, from less than 10 points. Here we talk >> /24 or more >> (there are 255 /8 prefixes, while there are 16777216 /24s). >> And mobiles >> can be attached to much more than just 10 points. No one could ever >> advertise that large number of routes from that many parts of >> Internet, >> not even two parts. >> >> Worse with IPv6, where we talk /32 (usual operator boundary) vs /64 >> prefixes (Ethernet end-user boundary). /64 prefixes are much more >> numerous than IPv4 /24 prefixes. > > you shouldnt be advertising /64 mobile network > prefixes. only aggregated home prefixes. but yes, > it could still be a problem. > > Vijay > From nemo-bounces@ietf.org Tue Apr 04 07:03:25 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQjJV-0006eq-0m; Tue, 04 Apr 2006 07:03:09 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQjJT-0006eE-Tb for nemo@ietf.org; Tue, 04 Apr 2006 07:03:07 -0400 Received: from omgo.iij.ad.jp ([202.232.30.157]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQjIq-00084r-RM for nemo@ietf.org; Tue, 04 Apr 2006 07:02:31 -0400 Received: OTM-MO id k34B2Mj7002690; Tue, 4 Apr 2006 20:02:22 +0900 (JST) Received: OTM-MIX(otm-mix00) id k34B2LRN018533; Tue, 4 Apr 2006 20:02:21 +0900 (JST) Received: from localhost (keiichi00.osaka.iij.ad.jp [192.168.64.45]) by jc-smtp.iij.ad.jp (JC-SMTP/jc-smtp) id k34B2LsD021354; Tue, 4 Apr 2006 20:02:21 +0900 (JST) Date: Tue, 04 Apr 2006 20:04:05 +0900 (JST) Message-Id: <20060404.200405.100484694.keiichi@iijlab.net> To: pthubert@cisco.com Subject: Re: [nemo] Fragmentation at mobile router From: Keiichi SHIMA In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC02043A55@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC02043A55@xmb-ams-337.emea.cisco.com> X-Mailer: Mew version 4.1.50 on Emacs 21.3 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Spam-Score: 0.0 (/) X-Scan-Signature: f66b12316365a3fe519e75911daf28a8 Cc: moriya.takaaki@lab.ntt.co.jp, nemo@ietf.org, Vijay.Devarapalli@AzaireNet.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Pascal, From: "Pascal Thubert \(pthubert\)" Subject: RE: [nemo] Fragmentation at mobile router Date: Tue, 4 Apr 2006 11:01:25 +0200 > >> Since RFC3963 does not refer to this issue, it seems to be outside of > >> the scope of NEMO specification. > >> Can we conclude it's implementation-dependent ? > >> If so, TAHI interoperability guideline will not specify fragmentation > at > >> mobile router as well as sending ICMP packet too big by mobile > router. > > > >Let me ask a clarification question. > > [Pascal] I'll describe what our implementation does. Now, I think it is > more a matter of good sense than something that could vary a lot with > implementations. In any case, the MR must do something so that the > packets go through, don't you think? > > > > >A mobile router have to send an ICMP too big, if the forwarded packet > >exceeds the MTU of the tunnel interface between the MR and its HA. > >Otherwise, we lose connectivity. > > > [Pascal] The MR can not fragment the packet that it will tunnel when it > forwards it over the tunnel interface; but it can fragment the packet > once it has been tunneled, because at this point, the MR IS the source > of the MR to HA packet, so the rules are not broken. The HA reassembles > the tunneled packet, receives it, and then it extracts the inner packet > (MNN to CN) which does not have a fragment header. OK, our implementation does the same as yours. > >A mobile router may not send a too big message if the packet is bigger > >than the physical path MTU (not the MTU for the tunnel i/f). It is > >implementation dependent. > > > [Pascal] For the MR, the tunnel is just another interface, with an MTU > that it can compute based on the encapsulation overhead and packet too > big it would receive because of recent history on MR to HA packets. The > MR needs to reassess that MTU every so often and when it moves. If the > MR does not send packet too big to the MNN, then it will have to > fragment the encapsulated packets forever. I understand. But it is a nature of a tunnel interface, isn't it? Usually, a node which has an IPv6-over-IPv6 tunnel interface just notify the MTU size of the tunnel interface itself when forwarding packets. The node doesn't notify the MTU value of the inner path. > It makes sense to notify the MNN, unless the tunnel MTU gets below the > min MTU of 1280 bytes. In a heavy case of nested NEMO, such a thing > could happen. > > > >Do I understand correctly? > > > >BTW, is there any word about the MTU between an MN/MR and an HA? Our > >implementation uses the IPv6 minimum MTU (1280) for the interface, but > >I know some implementations use bigger MTU size, which may cause a > >Path MTU discovery problem. > > > [Pascal] Initially, we set the tunnel MTU to (egressMTU-overhead), and > then react on packet too big that we might receive because of MR to HA > packets. By default, egressMTU will be 1500 on Ethernet. If you force > the tunnel MTU to 1280 from the start, then nested NEMO might be a > problem for you. I think the problem is there is no specification how to decide the tunnel MTU size between an MN and an HA. I guess your HA may have interoperability problem with an MN which assumes the tunnel MTU as 1280. Since you may send a packet which size is 1500, but the tunnel i/f of the MN may drop it because it exceeds the MTU size. In this case a packet too big will not be generated, since it is generated when packet is being sent to an interface, not being received from an interface. If I remember correctly, the very old Mobile IPv6 specification was talking the MTU size of the tunnel interface, I think. But the latest specification doesn't say anything. Do we need to clarify the point in another document? Regards, --- Keiichi SHIMA IIJ Research Laboratory WIDE Project From nemo-bounces@ietf.org Tue Apr 04 07:03:25 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQjJk-0006n9-Tu; Tue, 04 Apr 2006 07:03:24 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQjJU-0006eE-Kv for nemo@ietf.org; Tue, 04 Apr 2006 07:03:08 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQj9q-0007gp-3d for nemo@ietf.org; Tue, 04 Apr 2006 06:53:10 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 4DF66212C4E; Tue, 4 Apr 2006 13:53:09 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id DC5C3212C46; Tue, 4 Apr 2006 13:53:08 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 47054BDC40; Tue, 4 Apr 2006 13:53:08 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 137DBBDC3D; Tue, 4 Apr 2006 13:53:08 +0300 (EEST) In-Reply-To: <675DDBDD-84C1-4A43-98AF-25CC6164F94A@kniveton.com> References: <675DDBDD-84C1-4A43-98AF-25CC6164F94A@kniveton.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: ingress filtering for (n,n,n) (was Re: [nemo] Charter proposal Date: Tue, 4 Apr 2006 13:53:14 +0300 To: "T.J. Kniveton" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 4adaf050708fb13be3316a9eee889caa Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi all, In the proposed nemo charter, there is a bullet stating: Aug 2006 Submit -00 draft on (n,n,n) ingress filtering I am not sure i understand this point. I don't think that there is much nemo specific in the ingress=20 filtering problem itself OTOH, i think that the interesting problem to deal with in nemo=20 multihoming is how to provide an optimization for the provision of=20 fault tolerance in the link between the MR and the AR (or the HA) I mean, an important different between the nemo multihoming and the=20 general site multihoming scenario, is that failures are much more=20 likely to occur in the link between the MR and the AR (or the HA). So=20 haivng special tools to deal with this case would be a useful=20 optimization imho The nemo multihoming analysis draft presents one of such solutions in=20 the Appendix B and as i have mentioned already, imho the presented=20 soilution has enough merits to be desribed in a separate draft.=20 However, i don't see this as an ingress filtering solution, but rather=20= an optimization for the fault tolerance support for the most common=20 case of failure. It should be noted that the solution presented in the=20= appendix does not requires new protocol extensions, but it would be=20 more a informational work afaict so, not sure what you have in mind with this bullet but i would suggest=20= to reword it into something like Nemo multihoming support Optimizations for MR-AR link fault tolerance=20 or something more readable but with this ideas. Regards, marcelo El 24/03/2006, a las 22:27, T.J. Kniveton escribi=F3: > Here is a starting point for charter update. Please send your = comments. > > TJ > = From nemo-bounces@ietf.org Tue Apr 04 09:01:23 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQl9r-0003qn-4O; Tue, 04 Apr 2006 09:01:19 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQl9p-0003qi-Q3 for nemo@ietf.org; Tue, 04 Apr 2006 09:01:17 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQl9n-0004E7-GA for nemo@ietf.org; Tue, 04 Apr 2006 09:01:17 -0400 Received: from il06exr03.mot.com (il06exr03.mot.com [129.188.137.133]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k34DHGjD002442; Tue, 4 Apr 2006 06:17:20 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr03.mot.com (8.13.1/8.13.0) with ESMTP id k34DHaus026356; Tue, 4 Apr 2006 08:17:36 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id A5194865980; Tue, 4 Apr 2006 15:01:03 +0200 (CEST) Message-ID: <44326E0F.4050100@motorola.com> Date: Tue, 04 Apr 2006 15:01:03 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> In-Reply-To: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: d185fa790257f526fedfd5d01ed9c976 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , Vijay Devarapalli , "Davis, Terry L" , "T.J. Kniveton" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: > Hi all, > > global HAHA draft states in section 5.1.1 External routing > > In return, each site advertises the aggregation that encompasses the > Distributed Home Network aggregation back into the service provider > network, to the outside world (the internet). > > So, the (distributed) home network prefix is announced through BGP > and this implies that each Home Network prefix will impact with an > additional route in the global BGP table, all agree? > > The question at this point is whether this is a reasonable model > > Let's see what the current practise in IPv6 is. As today, the > reccommendation is that only ISPs can announce their blocks in BGP. > In particular, end sites are discouraged to do that, in particular, > multihomed sites have to find an alternative solution rather than > rely on BGP. Additionally, I think that even if two routes are announced in the BGP for same prefix, it's insured that the final destination of these two routes are in the _same_ physical place (same end user). Or, global HAHA claims its strengths in having a HA close to the point wherever the mobile may be, with HAs placed in various parts of the Internet. I.e. global HAHA requires the use of a multiply-announced BGP prefix that eventually ends in _different_ places. > Moreover, as today, there is no RIR policy that allows an end site to > get their own address space. Only Provider aggregatable address is > being handed out by the RIRs. > > Now, as is read this proposal, it implies that each distributed home > netwrok will need a prefix that is separately announced in the > global BGP table. Now, the question is if this is in line with > current tendency to consider global routing scalability as a top > priority. As i see it, there must be a very important reason to push > a solution that relies on this approach, and i really would like to > understand why the support of globally distributed HAs requires such > approach. > > I mean one could think of an approach to support gloablly distributed > HAs that do not require having the home network prefix injected > separately in the global BGP table. In particular, one could envision > a solution where each HA has its own prefix block, belonging to the > ISPs aggregate Would you elaborate on this perspective? How would an "ISP aggregate" be able to be present anywhere in the world where the mobile may roam? As a side note, do you think an "ISP aggregate" could outnumber all the ISPs present out there? If the media talks about some merger between two large ISPs and then calls that "the largest combined ISP in the world" do you really think that ISP covers the entire Internet? I am sure not. "Large" ISP in terms of number of subscribers is different than "large" in terms of square kilometers covered, because population density varies so much. A single cellular ISP in China may be bigger than all biggest western ISPs if we talk subscriber count, but of course much smaller than Iridium if we talk square kilometer count. > and that the mobile node prefers to use the address of the closest HA > to initiate new communications. How would a mobile _know_ which HA is "closest"? IP "distance" is an unreliable variable, nowhere defined. Maybe you meant that the BGP network will direct the BU to the HA that the network thinks it's closer, without MN involvement in the choice? > This with a route optimization solution should provide similar > benefits. An RO solution gets its first packets through HA and then straight to CN, thus eliminating HA from the path. Doesn't need global HAHA. Alex From nemo-bounces@ietf.org Tue Apr 04 09:43:13 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQloN-0002JW-48; Tue, 04 Apr 2006 09:43:11 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQloL-0002JR-Jm for nemo@ietf.org; Tue, 04 Apr 2006 09:43:09 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQloK-00061c-S3 for nemo@ietf.org; Tue, 04 Apr 2006 09:43:09 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id D0D59212C4E; Tue, 4 Apr 2006 16:43:06 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 69906212C46; Tue, 4 Apr 2006 16:43:06 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id E9DCBBDC40; Tue, 4 Apr 2006 16:43:05 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 7F6A5BDC3D; Tue, 4 Apr 2006 16:43:05 +0300 (EEST) In-Reply-To: <44326E0F.4050100@motorola.com> References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Tue, 4 Apr 2006 16:43:03 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 88b11fc64c1bfdb4425294ef5374ca07 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Alex, El 04/04/2006, a las 16:01, Alexandru Petrescu escribi=F3: > marcelo bagnulo braun wrote: >> Hi all, >> global HAHA draft states in section 5.1.1 External routing >> In return, each site advertises the aggregation that encompasses the=20= >> Distributed Home Network aggregation back into the service provider=20= >> network, to the outside world (the internet). >> So, the (distributed) home network prefix is announced through BGP=20 >> and this implies that each Home Network prefix will impact with an=20 >> additional route in the global BGP table, all agree? >> The question at this point is whether this is a reasonable model >> Let's see what the current practise in IPv6 is. As today, the=20 >> reccommendation is that only ISPs can announce their blocks in BGP.=20= >> In particular, end sites are discouraged to do that, in particular, >> multihomed sites have to find an alternative solution rather than=20 >> rely on BGP. > > Additionally, I think that even if two routes are announced in the BGP > for same prefix, it's insured that the final destination of these two > routes are in the _same_ physical place (same end user). > well, there are some configurations that use this, basically anycast=20 services, like DNS infrastucutre > Or, global HAHA claims its strengths in having a HA close to the point > wherever the mobile may be, with HAs placed in various parts of the=20 > Internet. I.e. global HAHA requires the use of a multiply-announced=20= > BGP > prefix that eventually ends in _different_ places. exactly, but i am not sure if this is a problem w.r.t. the scalability=20= of the global routing system. I mean, do you think that there is a difference w.r.t. routing system=20 scalability in the case that it is a single multihomed site that=20 announces the same prefix through multiple links to the internet, or a=20= multiple seperate sites announcing the same prefix? > >> Moreover, as today, there is no RIR policy that allows an end site to >> get their own address space. Only Provider aggregatable address is=20= >> being handed out by the RIRs. >> Now, as is read this proposal, it implies that each distributed home >> netwrok will need a prefix that is separately announced in the >> global BGP table. Now, the question is if this is in line with >> current tendency to consider global routing scalability as a top >> priority. As i see it, there must be a very important reason to push >> a solution that relies on this approach, and i really would like to >> understand why the support of globally distributed HAs requires such >> approach. >> I mean one could think of an approach to support gloablly distributed >> HAs that do not require having the home network prefix injected=20 >> separately in the global BGP table. In particular, one could envision >> a solution where each HA has its own prefix block, belonging to the >> ISPs aggregate > > Would you elaborate on this perspective? How would an "ISP aggregate" > be able to be present anywhere in the world where the mobile may roam? > see below... > As a side note, do you think an "ISP aggregate" could outnumber all = the > ISPs present out there? If the media talks about some merger between > two large ISPs and then calls that "the largest combined ISP in the > world" do you really think that ISP covers the entire Internet? I am > sure not. > actually by ISP aggregate i meant the ISP prefix (which aggregates all=20= the prefixes of the ISP's customers), so i guess we only have a=20 terminology misunderstanding here.. > "Large" ISP in terms of number of subscribers is different than = "large" > in terms of square kilometers covered, because population density=20 > varies > so much. A single cellular ISP in China may be bigger than all=20 > biggest western ISPs if we talk subscriber count, but of course much=20= > smaller than Iridium if we talk square kilometer count. > >> and that the mobile node prefers to use the address of the closest HA >> to initiate new communications. > I haven't thought this approach completelly but i was thinking=20 something in these lines. Suppose that a mobile network wants to use 4 HA that are geographically=20= distributed. Each of these 4 home networks obtain a topologically meaningful prefix=20= from the ISP prefix (which does not need to be announced seprately=20 since it is part of the ISP prefix) Suppose that the monet gets a home network prefix from each of the Home=20= networks Now, the goal at this point would be to use an address from the home=20 network prefix that is closer to the monet in a certain moment in time=20= to initiate new communications. If the node in the monet is able to=20 choose the right address, then it will be using the HA that is closer=20 to the monet achieving similar benefits to the ones obtained by the=20 global HAHA in terms of round trip time, agree? So in order to do that, the gaol would be for the MR to measure the=20 distance with the different HA and inform the mobile network nodes to=20 preffer the prefix that is associated to the HA that is closer. In order to measure the distance with the HA, there are several=20 possibilities as you mention below. One option would be to listen to=20 BGP and use the AS-path length. this is probably not the best measure.=20= Other option is to send probes and measure the RTT, which is likely to=20= obtain better results than the BGP based technique. In order to inform the hosts about which address to preffer, it could=20 easily do that by deprecating the prefixes of the HAs that are far away=20= and only leave the one of the closets HA as preferred. This would make=20= that all the new communications to be initiated with the address=20 associated to the closest HA There is an underlying assumption here that the movement is slow,=20 comparing distance and HAs. I mean, that in most of the cases, the=20 address preferred when the communication is initiated will be good=20 enough during all the lifetime of the communication. So the solution=20 works better in this case. I think this is the case for the airplane=20 case presented, since even while the airplane moves very fast, it=20 doesn't changes that often of attachment point to the internet, soi it=20= doesn't need to change the HA that often (compared with the lifetime of=20= the communication) > How would a mobile _know_ which HA is "closest"? IP "distance" is > an unreliable variable, nowhere defined. > > Maybe you meant that the BGP network will direct the BU to the HA that=20= > the network thinks it's closer, without MN involvement in the choice? > >> This with a route optimization solution should provide similar >> benefits. > > An RO solution gets its first packets through HA and then straight to=20= > CN, thus eliminating HA from the path. Doesn't need global HAHA. > Well, i hope it is clearer now what i had in mind. So there are two cases that additional work is needed: - one case is for communication that lasts for long and it was=20 initiated with the address associated with a previous HA. In this case,=20= this communication would not benefit from the closest HA. So in this=20 case we need an additional RO to optimize this situation. But this is=20 only for a reduced set of communications, so we can use something like=20= host based RO, like MIPv6 based RO for this particular nodes. - another case that needs to be taken into account is incoming=20 communications. I guess that most of the application cases the mobile=20 network nodes are clients, rather than servers, but in any case, it=20 would be worthy to consider this case. The problem here is that the CN=20= is the one selecting the address, so it may well choose not the =20 address corresponding to the closest HA. For this cases, we also need=20 some form of host based RO. So, i guess this would work fine for outgoing communications that are=20 not very long lived. But my feeling is that these are the most of the=20 traffic to be expected in many scenarios, like planes and buses and=20 trains. For the other communications a RO optimization mechanism is needed So i see this as an approach that is complementary with a RO solution,=20= makes any sense to you? Regards, marcelo > Alex > From nemo-bounces@ietf.org Tue Apr 04 13:28:01 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQpJs-0000DK-3X; Tue, 04 Apr 2006 13:27:56 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQpJq-0000DF-IH for nemo@ietf.org; Tue, 04 Apr 2006 13:27:54 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQpJq-0000ws-76 for nemo@ietf.org; Tue, 04 Apr 2006 13:27:54 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Fragmentation at mobile router Date: Tue, 4 Apr 2006 10:27:52 -0700 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Fragmentation at mobile router Thread-Index: AcZXga0aztW9ya7zRCupu3D7qoMMUgAQlyRQABJAPCA= From: "Vijay Devarapalli" To: "Pascal Thubert \(pthubert\)" , "Keiichi SHIMA" , X-Spam-Score: 0.0 (/) X-Scan-Signature: 73734d43604d52d23b3eba644a169745 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org what Pascal wrote makes sense to me. couldn't have described it better. Vijay=20 > -----Original Message----- > From: Pascal Thubert (pthubert) [mailto:pthubert@cisco.com]=20 > Sent: Tuesday, April 04, 2006 2:01 AM > To: Keiichi SHIMA; moriya.takaaki@lab.ntt.co.jp > Cc: nemo@ietf.org; Vijay Devarapalli > Subject: RE: [nemo] Fragmentation at mobile router >=20 > Hi Keiichi >=20 >=20 > >> Since RFC3963 does not refer to this issue, it seems to be=20 > outside of > >> the scope of NEMO specification. > >> Can we conclude it's implementation-dependent ? > >> If so, TAHI interoperability guideline will not specify=20 > fragmentation > at > >> mobile router as well as sending ICMP packet too big by mobile > router. > > > >Let me ask a clarification question. >=20 > [Pascal] I'll describe what our implementation does. Now, I=20 > think it is > more a matter of good sense than something that could vary a lot with > implementations. In any case, the MR must do something so that the > packets go through, don't you think? >=20 > > > >A mobile router have to send an ICMP too big, if the forwarded packet > >exceeds the MTU of the tunnel interface between the MR and its HA. > >Otherwise, we lose connectivity. > > > [Pascal] The MR can not fragment the packet that it will=20 > tunnel when it > forwards it over the tunnel interface; but it can fragment the packet > once it has been tunneled, because at this point, the MR IS the source > of the MR to HA packet, so the rules are not broken. The HA=20 > reassembles > the tunneled packet, receives it, and then it extracts the=20 > inner packet > (MNN to CN) which does not have a fragment header. >=20 > >A mobile router may not send a too big message if the packet=20 > is bigger > >than the physical path MTU (not the MTU for the tunnel i/f). It is > >implementation dependent. > > > [Pascal] For the MR, the tunnel is just another interface, with an MTU > that it can compute based on the encapsulation overhead and packet too > big it would receive because of recent history on MR to HA=20 > packets. The > MR needs to reassess that MTU every so often and when it moves. If the > MR does not send packet too big to the MNN, then it will have to > fragment the encapsulated packets forever.=20 >=20 > It makes sense to notify the MNN, unless the tunnel MTU gets below the > min MTU of 1280 bytes. In a heavy case of nested NEMO, such a thing > could happen. >=20 >=20 > >Do I understand correctly? > > > >BTW, is there any word about the MTU between an MN/MR and an HA? Our > >implementation uses the IPv6 minimum MTU (1280) for the=20 > interface, but > >I know some implementations use bigger MTU size, which may cause a > >Path MTU discovery problem. > > > [Pascal] Initially, we set the tunnel MTU to (egressMTU-overhead), and > then react on packet too big that we might receive because of MR to HA > packets. By default, egressMTU will be 1500 on Ethernet. If you force > the tunnel MTU to 1280 from the start, then nested NEMO might be a > problem for you. >=20 >=20 > >--- > >Keiichi SHIMA > >IIJ Research Laboratory > >WIDE Project >=20 >=20 From nemo-bounces@ietf.org Tue Apr 04 13:35:31 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQpRD-0005go-3N; Tue, 04 Apr 2006 13:35:31 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQpRC-0005gg-4I for nemo@ietf.org; Tue, 04 Apr 2006 13:35:30 -0400 Received: from motgate5.mot.com ([144.189.100.105]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQpRB-0001BL-Id for nemo@ietf.org; Tue, 04 Apr 2006 13:35:30 -0400 Received: from az33exr02.mot.com (az33exr02.mot.com [10.64.251.232]) by motgate5.mot.com (8.12.11/Motgate5) with ESMTP id k34HkmTJ026966; Tue, 4 Apr 2006 10:46:52 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by az33exr02.mot.com (8.13.1/8.13.0) with ESMTP id k34Hm3bB006587; Tue, 4 Apr 2006 12:48:04 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 3705F865980; Tue, 4 Apr 2006 19:35:18 +0200 (CEST) Message-ID: <4432AE55.7010805@motorola.com> Date: Tue, 04 Apr 2006 19:35:17 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: d008c19e97860b8641c1851f84665a75 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: > well, there are some configurations that use this, basically anycast > services, like DNS infrastucutre yes, right, but infrastructure not edge fringes. >> Or, global HAHA claims its strengths in having a HA close to the >> point wherever the mobile may be, with HAs placed in various parts >> of the Internet. I.e. global HAHA requires the use of a >> multiply-announced BGP prefix that eventually ends in _different_ >> places. > > exactly, but i am not sure if this is a problem w.r.t. the > scalability of the global routing system. I mean, do you think that > there is a difference w.r.t. routing system scalability in the case > that it is a single multihomed site that announces the same prefix > through multiple links to the internet, or a multiple seperate sites > announcing the same prefix? Yes, while it's feasible to multi home a few yearly-changing short prefixes it's infeasible to multi home several minutely-changing long prefixes. >>> Moreover, as today, there is no RIR policy that allows an end >>> site to get their own address space. Only Provider aggregatable >>> address is being handed out by the RIRs. Now, as is read this >>> proposal, it implies that each distributed home netwrok will need >>> a prefix that is separately announced in the global BGP table. >>> Now, the question is if this is in line with current tendency to >>> consider global routing scalability as a top priority. As i see >>> it, there must be a very important reason to push a solution >>> that relies on this approach, and i really would like to >>> understand why the support of globally distributed HAs requires >>> such approach. I mean one could think of an approach to support >>> gloablly distributed HAs that do not require having the home >>> network prefix injected separately in the global BGP table. In >>> particular, one could envision a solution where each HA has its >>> own prefix block, belonging to the ISPs aggregate >> >> Would you elaborate on this perspective? How would an "ISP >> aggregate" be able to be present anywhere in the world where the >> mobile may roam? >> > > see below... > >> As a side note, do you think an "ISP aggregate" could outnumber all >> the ISPs present out there? If the media talks about some merger >> between two large ISPs and then calls that "the largest combined >> ISP in the world" do you really think that ISP covers the entire >> Internet? I am sure not. >> > > actually by ISP aggregate i meant the ISP prefix (which aggregates > all the prefixes of the ISP's customers), so i guess we only have a > terminology misunderstanding here.. We should make sure that when we rely on an aggregated prefix advertised in core to solve mobility problems we don't also secretly assume specific prefixes being advertised too. >> "Large" ISP in terms of number of subscribers is different than >> "large" in terms of square kilometers covered, because population >> density varies so much. A single cellular ISP in China may be >> bigger than all biggest western ISPs if we talk subscriber count, >> but of course much smaller than Iridium if we talk square kilometer >> count. >> >>> and that the mobile node prefers to use the address of the >>> closest HA to initiate new communications. >> > > I haven't thought this approach completelly but i was thinking > something in these lines. > > Suppose that a mobile network wants to use 4 HA that are > geographically distributed. Each of these 4 home networks obtain a > topologically meaningful prefix from the ISP prefix (which does not > need to be announced seprately since it is part of the ISP prefix) > Suppose that the monet gets a home network prefix from each of the > Home networks So the monet gets 4 prefixes? 4 potential Home Addresses for LFNs? > Now, the goal at this point would be to use an address from the home > network prefix that is closer to the monet in a certain moment in > time to initiate new communications. If the node in the monet is able > to choose the right address, then it will be using the HA that is > closer to the monet achieving similar benefits to the ones obtained > by the global HAHA in terms of round trip time, agree? I think I get the picture. This looks feasible. MR may try to identify which MNP (out of the 4) to use for its Home Address. But this comes down to when MR moves in a different geographical area stop all its communications (LFNs' too) and then use another Home Address and start new communications. Ongoing apps stopped. Is this an accepted assumption? (that no session continuity is required) We've discussed this at some point in the mip4 WG. They have the Dynamic HA Assignment draft, now maybe an RFC. It was stressed that is about an initial assignment of the Home Address and that it wouldn't be about session continuity. And, if it's agreed that a globalHAHA-like solution is ok to not offer session continuity then the paradox is that one wouldn't need globalHAHA at all - it's sufficient for MR to _discover_ a local HA, obtain a "local" Home Address and all is fine. Which is: instead of pre-configuring 4 HAs do HA discovery. GlobalHAHA wouldn't be needed (i.e. no need of BGP route injection nor multi-homing). > So in order to do that, the gaol would be for the MR to measure the > distance with the different HA and inform the mobile network nodes to > preffer the prefix that is associated to the HA that is closer. > > In order to measure the distance with the HA, there are several > possibilities as you mention below. One option would be to listen to > BGP and use the AS-path length. I see, sounds as a good tool. It assumes though that MR will be exposed to the core BGP. An ISP for /24 end-users avoids that. But depends on the picture of course. > Other option is to send probes and measure the RTT, which is likely > to obtain better results than the BGP based technique. Yes, to write that we'd need routing experts involved. > In order to inform the hosts about which address to preffer, it could > easily do that by deprecating the prefixes of the HAs that are far > away and only leave the one of the closets HA as preferred. This > would make that all the new communications to be initiated with the > address associated to the closest HA Ok. > There is an underlying assumption here that the movement is slow, > comparing distance and HAs. I mean, that in most of the cases, the > address preferred when the communication is initiated will be good > enough during all the lifetime of the communication. So the solution > works better in this case. I think this is the case for the airplane > case presented, since even while the airplane moves very fast, it > doesn't changes that often of attachment point to the internet, soi > it doesn't need to change the HA that often (compared with the > lifetime of the communication) I think this is right. I think in the case of commercial airline jet liners, sea liners and similar the problem is even more simple because trajectories are well known. They are planned. Points of connection are well known in advance, there'd be little need to "discover" HA or to probe paths. >> How would a mobile _know_ which HA is "closest"? IP "distance" is >> an unreliable variable, nowhere defined. >> >> Maybe you meant that the BGP network will direct the BU to the HA >> that the network thinks it's closer, without MN involvement in the >> choice? >> >>> This with a route optimization solution should provide similar >>> benefits. >> >> An RO solution gets its first packets through HA and then straight >> to CN, thus eliminating HA from the path. Doesn't need global >> HAHA. >> > > Well, i hope it is clearer now what i had in mind. So there are two > cases that additional work is needed: - one case is for communication > that lasts for long and it was initiated with the address associated > with a previous HA. In this case, this communication would not > benefit from the closest HA. So in this case we need an additional RO > to optimize this situation. But this is only for a reduced set of > communications, so we can use something like host based RO, like > MIPv6 based RO for this particular nodes. - another case that needs > to be taken into account is incoming communications. I guess that > most of the application cases the mobile network nodes are clients, > rather than servers, but in any case, it would be worthy to consider > this case. The problem here is that the CN is the one selecting the > address, so it may well choose not the address corresponding to the > closest HA. For this cases, we also need some form of host based RO. > > > So, i guess this would work fine for outgoing communications that are > not very long lived. But my feeling is that these are the most of > the traffic to be expected in many scenarios, like planes and buses > and trains. I think once we accept that most communications are short lived then we can assume that Mobile IP is not needed at all. There exist widely deployed end-user applications that offer an illusion of "session continuity" without using Mobile IP, with reachability at a permanent address and all that. If we don't use Mobile IP and still offer users an idea of "session continuity" which s/he's ready to accept then the whole HA concept is not needed, so even less GlobalHAHA. > For the other communications a RO optimization mechanism is needed > > So i see this as an approach that is complementary with a RO > solution, makes any sense to you? I'm trying to, and understood most of the points you made. I think that what is needed by deployments is the most important thing. I think that if we have some set of scenarios from some people who wish to deploy the global HAHA tech (with an eventual description of the network topology) and some reasons to use Mobile IP then we'll get closer to delimiting the globalHAHA problem statement. Sorry for the too long text, Alex From nemo-bounces@ietf.org Tue Apr 04 18:53:45 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQuP9-00058V-KL; Tue, 04 Apr 2006 18:53:43 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQuP7-00058B-JP for nemo@ietf.org; Tue, 04 Apr 2006 18:53:41 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQuP6-0005hq-6a for nemo@ietf.org; Tue, 04 Apr 2006 18:53:41 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Tue, 4 Apr 2006 15:53:13 -0700 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZX7fDeTBsxSAStTeSHG1k/OPdE6QATBKig From: "Vijay Devarapalli" To: "marcelo bagnulo braun" , "Alexandru Petrescu" X-Spam-Score: 0.0 (/) X-Scan-Signature: 6907f330301e69261fa73bed91449a20 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Marcelo, you have to consider two things. first, MIPv6 and NEMO are based on the concept=20 of the mobile node and mobile router having the same home address and maintaining session continuity without changing the home address. this means for HAHA, the mobile router should have the same home address irrespective of which home agent (from the bunch of home agents that serve its home network) it attaches to. second, it is not realistic to assume the mobile router can figure out which home agent it is closest to based on IP hops, round trips, etc.. these are not always reliable ways of measuring distance from each home agent. Vijay > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > Sent: Tuesday, April 04, 2006 6:43 AM > To: Alexandru Petrescu > Cc: Pascal Thubert (pthubert); Davis, Terry L; T.J. Kniveton;=20 > Vijay Devarapalli; ml-nemo WG > Subject: Re: About route aggregation in global HAHA (was Re:=20 > draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) >=20 > Hi Alex, >=20 >=20 > El 04/04/2006, a las 16:01, Alexandru Petrescu escribi=F3: >=20 > > marcelo bagnulo braun wrote: > >> Hi all, > >> global HAHA draft states in section 5.1.1 External routing > >> In return, each site advertises the aggregation that=20 > encompasses the=20 > >> Distributed Home Network aggregation back into the service=20 > provider=20 > >> network, to the outside world (the internet). > >> So, the (distributed) home network prefix is announced through BGP=20 > >> and this implies that each Home Network prefix will impact with an=20 > >> additional route in the global BGP table, all agree? > >> The question at this point is whether this is a reasonable model > >> Let's see what the current practise in IPv6 is. As today, the=20 > >> reccommendation is that only ISPs can announce their=20 > blocks in BGP.=20 > >> In particular, end sites are discouraged to do that, in =20 > particular, > >> multihomed sites have to find an alternative solution rather than=20 > >> rely on BGP. > > > > Additionally, I think that even if two routes are announced=20 > in the BGP > > for same prefix, it's insured that the final destination of=20 > these two > > routes are in the _same_ physical place (same end user). > > >=20 > well, there are some configurations that use this, basically anycast=20 > services, like DNS infrastucutre >=20 > > Or, global HAHA claims its strengths in having a HA close=20 > to the point > > wherever the mobile may be, with HAs placed in various parts of the=20 > > Internet. I.e. global HAHA requires the use of a=20 > multiply-announced=20 > > BGP > > prefix that eventually ends in _different_ places. >=20 > exactly, but i am not sure if this is a problem w.r.t. the=20 > scalability=20 > of the global routing system. > I mean, do you think that there is a difference w.r.t. routing system=20 > scalability in the case that it is a single multihomed site that=20 > announces the same prefix through multiple links to the=20 > internet, or a=20 > multiple seperate sites announcing the same prefix? >=20 > > > >> Moreover, as today, there is no RIR policy that allows an=20 > end site to > >> get their own address space. Only Provider aggregatable=20 > address is=20 > >> being handed out by the RIRs. > >> Now, as is read this proposal, it implies that each=20 > distributed home > >> netwrok will need a prefix that is separately announced in the > >> global BGP table. Now, the question is if this is in line with > >> current tendency to consider global routing scalability as a top > >> priority. As i see it, there must be a very important=20 > reason to push > >> a solution that relies on this approach, and i really would like to > >> understand why the support of globally distributed HAs=20 > requires such > >> approach. > >> I mean one could think of an approach to support gloablly=20 > distributed > >> HAs that do not require having the home network prefix injected=20 > >> separately in the global BGP table. In particular, one=20 > could envision > >> a solution where each HA has its own prefix block,=20 > belonging to the > >> ISPs aggregate > > > > Would you elaborate on this perspective? How would an "ISP=20 > aggregate" > > be able to be present anywhere in the world where the=20 > mobile may roam? > > >=20 > see below... >=20 > > As a side note, do you think an "ISP aggregate" could=20 > outnumber all the > > ISPs present out there? If the media talks about some=20 > merger between > > two large ISPs and then calls that "the largest combined ISP in the > > world" do you really think that ISP covers the entire=20 > Internet? I am > > sure not. > > >=20 > actually by ISP aggregate i meant the ISP prefix (which=20 > aggregates all=20 > the prefixes of the ISP's customers), so i guess we only have a=20 > terminology misunderstanding here.. >=20 > > "Large" ISP in terms of number of subscribers is different=20 > than "large" > > in terms of square kilometers covered, because population density=20 > > varies > > so much. A single cellular ISP in China may be bigger than all=20 > > biggest western ISPs if we talk subscriber count, but of=20 > course much=20 > > smaller than Iridium if we talk square kilometer count. > > > >> and that the mobile node prefers to use the address of the=20 > closest HA > >> to initiate new communications. > > >=20 > I haven't thought this approach completelly but i was thinking=20 > something in these lines. >=20 > Suppose that a mobile network wants to use 4 HA that are=20 > geographically=20 > distributed. > Each of these 4 home networks obtain a topologically=20 > meaningful prefix=20 > from the ISP prefix (which does not need to be announced seprately=20 > since it is part of the ISP prefix) > Suppose that the monet gets a home network prefix from each=20 > of the Home=20 > networks >=20 > Now, the goal at this point would be to use an address from the home=20 > network prefix that is closer to the monet in a certain=20 > moment in time=20 > to initiate new communications. If the node in the monet is able to=20 > choose the right address, then it will be using the HA that is closer=20 > to the monet achieving similar benefits to the ones obtained by the=20 > global HAHA in terms of round trip time, agree? >=20 > So in order to do that, the gaol would be for the MR to measure the=20 > distance with the different HA and inform the mobile network nodes to=20 > preffer the prefix that is associated to the HA that is closer. >=20 > In order to measure the distance with the HA, there are several=20 > possibilities as you mention below. One option would be to listen to=20 > BGP and use the AS-path length. this is probably not the best=20 > measure.=20 > Other option is to send probes and measure the RTT, which is=20 > likely to=20 > obtain better results than the BGP based technique. > In order to inform the hosts about which address to preffer, it could=20 > easily do that by deprecating the prefixes of the HAs that=20 > are far away=20 > and only leave the one of the closets HA as preferred. This=20 > would make=20 > that all the new communications to be initiated with the address=20 > associated to the closest HA >=20 > There is an underlying assumption here that the movement is slow,=20 > comparing distance and HAs. I mean, that in most of the cases, the=20 > address preferred when the communication is initiated will be good=20 > enough during all the lifetime of the communication. So the solution=20 > works better in this case. I think this is the case for the airplane=20 > case presented, since even while the airplane moves very fast, it=20 > doesn't changes that often of attachment point to the=20 > internet, soi it=20 > doesn't need to change the HA that often (compared with the=20 > lifetime of=20 > the communication) >=20 > > How would a mobile _know_ which HA is "closest"? IP "distance" is > > an unreliable variable, nowhere defined. > > > > Maybe you meant that the BGP network will direct the BU to=20 > the HA that=20 > > the network thinks it's closer, without MN involvement in=20 > the choice? > > > >> This with a route optimization solution should provide similar > >> benefits. > > > > An RO solution gets its first packets through HA and then=20 > straight to=20 > > CN, thus eliminating HA from the path. Doesn't need global HAHA. > > >=20 > Well, i hope it is clearer now what i had in mind. > So there are two cases that additional work is needed: > - one case is for communication that lasts for long and it was=20 > initiated with the address associated with a previous HA. In=20 > this case,=20 > this communication would not benefit from the closest HA. So in this=20 > case we need an additional RO to optimize this situation. But this is=20 > only for a reduced set of communications, so we can use=20 > something like=20 > host based RO, like MIPv6 based RO for this particular nodes. > - another case that needs to be taken into account is incoming=20 > communications. I guess that most of the application cases the mobile=20 > network nodes are clients, rather than servers, but in any case, it=20 > would be worthy to consider this case. The problem here is=20 > that the CN=20 > is the one selecting the address, so it may well choose not the =20 > address corresponding to the closest HA. For this cases, we also need=20 > some form of host based RO. >=20 > So, i guess this would work fine for outgoing communications that are=20 > not very long lived. But my feeling is that these are the most of the=20 > traffic to be expected in many scenarios, like planes and buses and=20 > trains. >=20 > For the other communications a RO optimization mechanism is needed >=20 > So i see this as an approach that is complementary with a RO=20 > solution,=20 > makes any sense to you? >=20 > Regards, marcelo >=20 > > Alex > > >=20 >=20 From nemo-bounces@ietf.org Tue Apr 04 19:02:09 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQuXE-0000Jl-9m; Tue, 04 Apr 2006 19:02:04 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQuXC-0000Jg-TS for nemo@ietf.org; Tue, 04 Apr 2006 19:02:02 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQuXC-0006HH-2T for nemo@ietf.org; Tue, 04 Apr 2006 19:02:02 -0400 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: ingress filtering for (n,n,n) (was Re: [nemo] Charter proposal Date: Tue, 4 Apr 2006 16:01:54 -0700 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: ingress filtering for (n,n,n) (was Re: [nemo] Charter proposal Thread-Index: AcZX12di6HU+SunPQh2P9QmcCqkM9QAY77Zg From: "Vijay Devarapalli" To: "marcelo bagnulo braun" , "T.J. Kniveton" X-Spam-Score: 0.0 (/) X-Scan-Signature: f60d0f7806b0c40781eee6b9cd0b2135 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Chan Wah presented a bunch of slides at the last IETF, explaining what needs to be done in the NEMO WG and Monami6 WG when it comes to multihoming. see the following URL. http://www3.ietf.org/proceedings/05nov/slides/nemo-2/sld6.htm I think thats what the following means. > Aug 2006 Submit -00 draft on (n,n,n) ingress filtering Vijay > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > Sent: Tuesday, April 04, 2006 3:53 AM > To: T.J. Kniveton > Cc: ml-nemo WG > Subject: ingress filtering for (n,n,n) (was Re: [nemo]=20 > Charter proposal >=20 > Hi all, >=20 > In the proposed nemo charter, there is a bullet stating: >=20 > Aug 2006 Submit -00 draft on (n,n,n) ingress filtering >=20 > I am not sure i understand this point. > I don't think that there is much nemo specific in the ingress=20 > filtering problem itself > OTOH, i think that the interesting problem to deal with in nemo=20 > multihoming is how to provide an optimization for the provision of=20 > fault tolerance in the link between the MR and the AR (or the HA) >=20 > I mean, an important different between the nemo multihoming and the=20 > general site multihoming scenario, is that failures are much more=20 > likely to occur in the link between the MR and the AR (or the HA). So=20 > haivng special tools to deal with this case would be a useful=20 > optimization imho > The nemo multihoming analysis draft presents one of such solutions in=20 > the Appendix B and as i have mentioned already, imho the presented=20 > soilution has enough merits to be desribed in a separate draft.=20 > However, i don't see this as an ingress filtering solution,=20 > but rather=20 > an optimization for the fault tolerance support for the most common=20 > case of failure. It should be noted that the solution=20 > presented in the=20 > appendix does not requires new protocol extensions, but it would be=20 > more a informational work afaict >=20 > so, not sure what you have in mind with this bullet but i=20 > would suggest=20 > to reword it into something like > Nemo multihoming support Optimizations for MR-AR link fault tolerance=20 > or something more readable but with this ideas. >=20 > Regards, marcelo >=20 >=20 > El 24/03/2006, a las 22:27, T.J. Kniveton escribi=F3: >=20 > > Here is a starting point for charter update. Please send=20 > your comments. > > > > TJ > > >=20 >=20 From nemo-bounces@ietf.org Tue Apr 04 22:34:23 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQxqf-0002zC-7r; Tue, 04 Apr 2006 22:34:21 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQxqe-0002z7-1k for nemo@ietf.org; Tue, 04 Apr 2006 22:34:20 -0400 Received: from yui.nc.u-tokyo.ac.jp ([130.69.251.116]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQxqc-0006S9-ET for nemo@ietf.org; Tue, 04 Apr 2006 22:34:20 -0400 Received: from [203.178.139.231] (wifi-139-231.sfc.wide.ad.jp [203.178.139.231]) (authenticated bits=0) by yui.nc.u-tokyo.ac.jp (8.12.10/8.12.3/Debian-6.4) with ESMTP id k352YEUl025335; Wed, 5 Apr 2006 11:34:15 +0900 In-Reply-To: References: <20060330105501.9B9769518C@smtp02.uc3m.es> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=UTF-8; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: Ryuji Wakikawa Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo]Charter proposal) Date: Wed, 5 Apr 2006 11:34:08 +0900 To: marcelo bagnulo braun X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.0 (/) X-Scan-Signature: 4b800b1eab964a31702fa68f1ff0e955 Cc: 'ml-nemo WG' X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Marcelo Sorry, I'm a bit late to join the discussion.. It is not necessary to inject home prefix to BGP. The global HAHA can =20= run inside IGP, too. It depends how scale do you operate NEMO on the Internet. ryuji On 2006/03/30, at 20:01, marcelo bagnulo braun wrote: > So, we do agree that global HAHA implies announcing the home =20 > network into the global BGP routing table, and that each =20 > distributed home network impacts with an additional entry in the =20 > bgp table, right? > > wouldn't this imply global routing system scalability concerns? > > Regards, marcelo > > > El 30/03/2006, a las 13:54, Ben McCarthy escribi=C3=B3: > >> Sorry, I read your initial question wrong, I thought you were =20 >> referring to >> Boeing's BGP manipulation technique they use... >> http://www.apnic.net/meetings/19/docs/sigs/routing/routing-pres-=20 >> skeen-global >> -ip-netmob.pdf >> >> not Global HA-HA, my mistake. >> >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >> Sent: 30 March 2006 11:26 >> To: Ben McCarthy >> Cc: 'ml-nemo WG'; 'Simon Oosthoek' >> Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo]=20 >> Charter >> proposal) >> >>> >>> [Ben] Apart from optimising the route between the MR and CNs or >>> altering >>> BGP, then the other option is exactly what Pascal is proposing >>> (switching to >>> a HA closer to the point-of-attachment to optimise the triangular >>> route). >>> >> >> perhaps i am failing to understand how this works, but wouldn't this >> implies that each home network must be announced separately in BGP? >> >> regards, marcelo >> >>> Apart from that...??? >>> >>> Cheers, >>> >>> Ben >>> >>>> Ben >>>> >>>> >>> >> > From nemo-bounces@ietf.org Tue Apr 04 22:43:32 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQxzX-000625-S1; Tue, 04 Apr 2006 22:43:31 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQxzX-000620-A5 for nemo@ietf.org; Tue, 04 Apr 2006 22:43:31 -0400 Received: from yui.nc.u-tokyo.ac.jp ([130.69.251.116]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQxzW-0006pj-O7 for nemo@ietf.org; Tue, 04 Apr 2006 22:43:31 -0400 Received: from [203.178.139.231] (wifi-139-231.sfc.wide.ad.jp [203.178.139.231]) (authenticated bits=0) by yui.nc.u-tokyo.ac.jp (8.12.10/8.12.3/Debian-6.4) with ESMTP id k352hRUl026359; Wed, 5 Apr 2006 11:43:27 +0900 In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AA50@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AA50@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Ryuji Wakikawa Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 11:43:21 +0900 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.0 (/) X-Scan-Signature: cd26b070c2577ac175cd3a6d878c6248 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Andrew On 2006/03/31, at 2:06, Dul, Andrew L wrote: >> -----Original Message----- >> From: Vijay Devarapalli [mailto:Vijay.Devarapalli@AzaireNet.com] >> Sent: Wednesday, March 29, 2006 3:34 PM >>> >>> Aviation & maritime platforms are globally mobile, because they are >>> globally mobile this makes route optimization also an InterDomain >>> Routing issue. >>> >>> Even if you had global-haha implemented it would require >> multiple ASNs >>> to announce the mobile platform's address space. That would likely >>> require cooperation among a number of network operators to truly get >>> route-optimization. These network operators would also >> have to agree >>> to provide transit to other operators where the mobile platform was >>> currently homed. >>> >>> Alternatively the mobile service provider would have to setup >>> infrastructure in multiple well connected and well peered global >>> locations just to prevent triangle routing from occurring. >> The mobile >>> service provider would also have to provide a transit network to >>> transport traffic between the HA and the current attachment point. >> >> is that a problem? this problem is common to all roaming nodes, no? > > > The problem is common but the situation is exacerbated because the > roaming nodes are often far away from their home network. > > If you have to tunnel gigabytes of traffic across town because your > traffic didn't enter at the closest point, that isn't necessarily a > problem because the cost of carrying that traffic across town is > negligible. > > However, if you have tunnel gigabytes of traffic across oceans because > your traffic is being triangle routed that is a problem because the > cost > is substantial. > > Route optimization, to me, is not necessarily about always using the > shortest path from a physical perspective, but it is about using the > most economic path. Having the best path looks cool, but if the > cost to > tunnel the traffic is negligible compared with the increased cost of > complexity, I would vote for the negligible increase in transit cost. > However if the cost of transit is high then additional complexity > would > likely be needed to reduce the real cost of transporting bits. > > Andrew I agree with your statement. RO is not always effective (ex. HA is close enough to MN). It totally depends on topology. MIP6 RO requires several signaling between MN and CN, but a small optimization may be achieved in some case. This is problematic, since a mobile node involve this RO operation. And how shorter path the MN can obtain cannot be predicted before RO completed. HAHA is transparent to end nodes and can be controlled by operators (not end nodes). This type of solution is doable since the backbone gets quite first these days. ryuji From nemo-bounces@ietf.org Tue Apr 04 22:45:15 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQy15-00089r-Ki; Tue, 04 Apr 2006 22:45:07 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQy14-0007ye-2S for nemo@ietf.org; Tue, 04 Apr 2006 22:45:06 -0400 Received: from yui.nc.u-tokyo.ac.jp ([130.69.251.116]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQy13-0006rP-HH for nemo@ietf.org; Tue, 04 Apr 2006 22:45:06 -0400 Received: from [203.178.139.231] (wifi-139-231.sfc.wide.ad.jp [203.178.139.231]) (authenticated bits=0) by yui.nc.u-tokyo.ac.jp (8.12.10/8.12.3/Debian-6.4) with ESMTP id k352itUl026535; Wed, 5 Apr 2006 11:44:56 +0900 In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> Content-Transfer-Encoding: 7bit From: Ryuji Wakikawa Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 11:44:50 +0900 To: Vijay Devarapalli X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.0 (/) X-Scan-Signature: 41c17b4b16d1eedaa8395c26e9a251c4 Cc: marcelo bagnulo braun , ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org On 2006/03/31, at 3:17, Vijay Devarapalli wrote: > >>> The good news is that TJ's revised charter has a line that >> could cover >>> these discussions: >>> Aug 2006 Submit -00 draft on Route Optimization for >>> geographically distributed HAs >>> >> >> right, but the proposed charted also states that: >> >> The WG will: >> >> - Ensure that solutions will scale and function for the different >> mobile network configurations, without requiring changes to >> Correspondent Nodes in the Internet. All solutions will aim at >> preserving route aggregation within the Internet and will satisfy an >> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >> acceptable level of security (a thorough survey of new threats and an >> analysis of their severity will be conducted for any changes >> introduced >> to the network). >> >> I guess one of the points Andrew is pointing out is exactly with >> respect to this issues, namely: >> >>>> Even if you had global-haha implemented it would require >> multiple ASNs >>>> to announce the mobile platform's address space. That would likely >> >> how would this affect aggregation? > > it is very common today to advertise reachability > to a particular network from different points in > the Internet. Yes. This is common operation these days. > ofcourse, I do agree that route aggregation should > be preserved. We have NEMO home network models document:-) ryuji > Vijay > >> >> imho we should first discuss and figure out if the operational >> practices resulting from such a solution are acceptable before >> chartering the work. I mean we should first figure out at least if >> there seems to be a viable approach that fulfills the needs and >> preserves scalability of the routing system >> >> Regards, marcelo >> >> >>> We can propose global HAHA as a candidate, compete with >> other proposals >>> and try to keep the best from all. >>> >>> Pascal >>> >> >> > From nemo-bounces@ietf.org Tue Apr 04 22:52:18 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQy82-0001gW-NL; Tue, 04 Apr 2006 22:52:18 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQy82-0001gR-8O for nemo@ietf.org; Tue, 04 Apr 2006 22:52:18 -0400 Received: from yui.nc.u-tokyo.ac.jp ([130.69.251.116]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQy81-0006xH-Kw for nemo@ietf.org; Tue, 04 Apr 2006 22:52:18 -0400 Received: from [203.178.139.231] (wifi-139-231.sfc.wide.ad.jp [203.178.139.231]) (authenticated bits=0) by yui.nc.u-tokyo.ac.jp (8.12.10/8.12.3/Debian-6.4) with ESMTP id k352q7Ul027336; Wed, 5 Apr 2006 11:52:07 +0900 In-Reply-To: <44326E0F.4050100@motorola.com> References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <41C95BE6-9648-4CD0-827A-4B7E4AD7CB96@sfc.wide.ad.jp> Content-Transfer-Encoding: 7bit From: Ryuji Wakikawa Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 11:52:01 +0900 To: Alexandru Petrescu X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.0 (/) X-Scan-Signature: 34d35111647d654d033d58d318c0d21a Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , Vijay Devarapalli , "Davis, Terry L" , marcelo bagnulo braun X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Alex On 2006/04/04, at 22:01, Alexandru Petrescu wrote: > marcelo bagnulo braun wrote: >> Hi all, >> global HAHA draft states in section 5.1.1 External routing >> In return, each site advertises the aggregation that encompasses >> the Distributed Home Network aggregation back into the service >> provider network, to the outside world (the internet). >> So, the (distributed) home network prefix is announced through BGP >> and this implies that each Home Network prefix will impact with an >> additional route in the global BGP table, all agree? >> The question at this point is whether this is a reasonable model >> Let's see what the current practise in IPv6 is. As today, the >> reccommendation is that only ISPs can announce their blocks in >> BGP. In particular, end sites are discouraged to do that, in >> particular, >> multihomed sites have to find an alternative solution rather than >> rely on BGP. > > Additionally, I think that even if two routes are announced in the BGP > for same prefix, it's insured that the final destination of these two > routes are in the _same_ physical place (same end user). VPN can be used. Did you read draft-wakikawa-mip6-nemo-haha-spec-01.txt? We describe the various Home Agent configurations. > Or, global HAHA claims its strengths in having a HA close to the point > wherever the mobile may be, with HAs placed in various parts of the > Internet. I.e. global HAHA requires the use of a multiply- > announced BGP > prefix that eventually ends in _different_ places. For global HAHA, it is not necessarily to use BGP. Any routing protocol can be used. Of course, one example is BGP to obtain globally scale route optimization. ryuji >> Moreover, as today, there is no RIR policy that allows an end site to >> get their own address space. Only Provider aggregatable address >> is being handed out by the RIRs. >> Now, as is read this proposal, it implies that each distributed home >> netwrok will need a prefix that is separately announced in the >> global BGP table. Now, the question is if this is in line with >> current tendency to consider global routing scalability as a top >> priority. As i see it, there must be a very important reason to push >> a solution that relies on this approach, and i really would like to >> understand why the support of globally distributed HAs requires such >> approach. >> I mean one could think of an approach to support gloablly distributed >> HAs that do not require having the home network prefix injected >> separately in the global BGP table. In particular, one could envision >> a solution where each HA has its own prefix block, belonging to the >> ISPs aggregate > > Would you elaborate on this perspective? How would an "ISP aggregate" > be able to be present anywhere in the world where the mobile may roam? > > As a side note, do you think an "ISP aggregate" could outnumber all > the > ISPs present out there? If the media talks about some merger between > two large ISPs and then calls that "the largest combined ISP in the > world" do you really think that ISP covers the entire Internet? I am > sure not. > > "Large" ISP in terms of number of subscribers is different than > "large" > in terms of square kilometers covered, because population density > varies > so much. A single cellular ISP in China may be bigger than all > biggest western ISPs if we talk subscriber count, but of course > much smaller than Iridium if we talk square kilometer count. > >> and that the mobile node prefers to use the address of the closest HA >> to initiate new communications. > > How would a mobile _know_ which HA is "closest"? IP "distance" is > an unreliable variable, nowhere defined. > > Maybe you meant that the BGP network will direct the BU to the HA > that the network thinks it's closer, without MN involvement in the > choice? > >> This with a route optimization solution should provide similar >> benefits. > > An RO solution gets its first packets through HA and then straight > to CN, thus eliminating HA from the path. Doesn't need global HAHA. > > Alex > From nemo-bounces@ietf.org Tue Apr 04 23:42:43 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQyuk-0007Xv-3f; Tue, 04 Apr 2006 23:42:38 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FQyui-0007Xq-SA for nemo@ietf.org; Tue, 04 Apr 2006 23:42:36 -0400 Received: from smtp.mei.co.jp ([133.183.129.25] helo=smtp1.mei.co.jp) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FQyui-0008VZ-7b for nemo@ietf.org; Tue, 04 Apr 2006 23:42:36 -0400 Received: from mail-gw.jp.panasonic.com (dodgers.mei.co.jp [157.8.1.150]) by smtp1.mei.co.jp (8.12.11.20060308/3.7W/jazz) with ESMTP id k353gSSr010328; Wed, 5 Apr 2006 12:42:28 +0900 (JST) Received: by mail-gw.jp.panasonic.com (8.11.6p2/3.7W/somlx3) with ESMTP id k353gUA01071; Wed, 5 Apr 2006 12:42:30 +0900 (JST) Received: from pslexc01.psl.local (localhost [127.0.0.1]) by mail.jp.panasonic.com (8.11.6p2/3.7W/bluejays) with ESMTP id k353gTG19834; Wed, 5 Apr 2006 12:42:29 +0900 (JST) Received: from bach.sg.panasonic.com ([10.81.113.99]) by pslexc01.psl.local with Microsoft SMTPSVC(6.0.3790.1830); Wed, 5 Apr 2006 11:41:04 +0800 Received: by bach.sg.panasonic.com (Postfix, from userid 1000) id 9FBE1D41DA; Wed, 5 Apr 2006 11:49:31 +0800 (SGT) Subject: RE: ingress filtering for (n,n,n) (was Re: [nemo] Charter proposal From: Chan-Wah Ng To: Vijay Devarapalli In-Reply-To: References: Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Date: Wed, 05 Apr 2006 11:49:31 +0800 Message-Id: <1144208971.21148.60.camel@localhost> Mime-Version: 1.0 X-Mailer: Evolution 2.4.2.1 X-OriginalArrivalTime: 05 Apr 2006 03:41:04.0355 (UTC) FILETIME=[C3E9B330:01C65862] X-Spam-Score: 0.1 (/) X-Scan-Signature: 10d3e4e3c32e363f129e380e644649be Cc: marcelo bagnulo braun , ml-nemo WG , "T.J. Kniveton" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hello all, Thanks Vijay. I am certain that's what TJ have in mind when he placed in that bullet. In fact, if you read through the appendix B in draft-ietf-nemo-multihoming, you will find that the mechanism described can be used to overcome ingress filtering problem as described in 4.2, although it is worded as a fault tolerance solution. Ingress filtering, is in fact main obstacle for fault tolerance in NEMO.=20 Having said that, I do agree with marcelo on the rewording of the charter though. It would make it specific to NEMO, and raise less concern for work scope overlaps in WGs. /rgds /cwng On Tue, 2006-04-04 at 16:01 -0700, Vijay Devarapalli wrote: > Chan Wah presented a bunch of slides at the last > IETF, explaining what needs to be done in the > NEMO WG and Monami6 WG when it comes to multihoming. >=20 > see the following URL. > http://www3.ietf.org/proceedings/05nov/slides/nemo-2/sld6.htm >=20 > I think thats what the following means. >=20 > > Aug 2006 Submit -00 draft on (n,n,n) ingress filtering >=20 > Vijay >=20 > > -----Original Message----- > > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > > Sent: Tuesday, April 04, 2006 3:53 AM > > To: T.J. Kniveton > > Cc: ml-nemo WG > > Subject: ingress filtering for (n,n,n) (was Re: [nemo]=20 > > Charter proposal > >=20 > > Hi all, > >=20 > > In the proposed nemo charter, there is a bullet stating: > >=20 > > Aug 2006 Submit -00 draft on (n,n,n) ingress filtering > >=20 > > I am not sure i understand this point. > > I don't think that there is much nemo specific in the ingress=20 > > filtering problem itself > > OTOH, i think that the interesting problem to deal with in nemo=20 > > multihoming is how to provide an optimization for the provision of=20 > > fault tolerance in the link between the MR and the AR (or the HA) > >=20 > > I mean, an important different between the nemo multihoming and the=20 > > general site multihoming scenario, is that failures are much more=20 > > likely to occur in the link between the MR and the AR (or the HA). So=20 > > haivng special tools to deal with this case would be a useful=20 > > optimization imho > > The nemo multihoming analysis draft presents one of such solutions in=20 > > the Appendix B and as i have mentioned already, imho the presented=20 > > soilution has enough merits to be desribed in a separate draft.=20 > > However, i don't see this as an ingress filtering solution,=20 > > but rather=20 > > an optimization for the fault tolerance support for the most common=20 > > case of failure. It should be noted that the solution=20 > > presented in the=20 > > appendix does not requires new protocol extensions, but it would be=20 > > more a informational work afaict > >=20 > > so, not sure what you have in mind with this bullet but i=20 > > would suggest=20 > > to reword it into something like > > Nemo multihoming support Optimizations for MR-AR link fault tolerance=20 > > or something more readable but with this ideas. > >=20 > > Regards, marcelo > >=20 > >=20 > > El 24/03/2006, a las 22:27, T.J. Kniveton escribi=C3=B3: > >=20 > > > Here is a starting point for charter update. Please send=20 > > your comments. > > > > > > TJ > > > > >=20 > >=20 >=20 From nemo-bounces@ietf.org Wed Apr 05 06:00:11 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR4o4-0001hu-P4; Wed, 05 Apr 2006 06:00:08 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR4o3-0001hm-MM for nemo@ietf.org; Wed, 05 Apr 2006 06:00:07 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR4o2-0006Ql-Qx for nemo@ietf.org; Wed, 05 Apr 2006 06:00:07 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 17EA6212C4E; Wed, 5 Apr 2006 13:00:05 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id D22D0212C46; Wed, 5 Apr 2006 13:00:04 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 61470BDC40; Wed, 5 Apr 2006 13:00:04 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 27B92BDC3D; Wed, 5 Apr 2006 13:00:04 +0300 (EEST) In-Reply-To: <4432AE55.7010805@motorola.com> References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> <4432AE55.7010805@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 13:00:03 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 90e8b0e368115979782f8b3d811b226b Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 04/04/2006, a las 20:35, Alexandru Petrescu escribi=F3: > marcelo bagnulo braun wrote: >> well, there are some configurations that use this, basically anycast >> services, like DNS infrastucutre > > yes, right, but infrastructure not edge fringes. > absolutely >>> Or, global HAHA claims its strengths in having a HA close to the=20 >>> point wherever the mobile may be, with HAs placed in various parts >>> of the Internet. I.e. global HAHA requires the use of a=20 >>> multiply-announced BGP prefix that eventually ends in _different_=20 >>> places. >> exactly, but i am not sure if this is a problem w.r.t. the=20 >> scalability of the global routing system. I mean, do you think that=20= >> there is a difference w.r.t. routing system scalability in the case=20= >> that it is a single multihomed site that announces the same prefix=20 >> through multiple links to the internet, or a multiple seperate sites >> announcing the same prefix? > > Yes, while it's feasible to multi home a few yearly-changing > short prefixes it's infeasible to multi home several minutely-changing=20= > long prefixes. > completely agree, but my point is that this is because stress on the=20 global routing table (due to additional entries and the dynamic of such=20= entries) rather than because they are seperate sites announcing the=20 same prefix. I mean, having many long prefixes of multihomed sites that=20= appear and disapear would cause a similar effect even if they are=20 announced by a single site connected through multiple points to the=20 internet >>>> Moreover, as today, there is no RIR policy that allows an end site=20= >>>> to get their own address space. Only Provider aggregatable address=20= >>>> is being handed out by the RIRs. Now, as is read this proposal, it=20= >>>> implies that each distributed home netwrok will need >>>> a prefix that is separately announced in the global BGP table.=20 >>>> Now, the question is if this is in line with current tendency to >>>> consider global routing scalability as a top priority. As i see >>>> it, there must be a very important reason to push a solution >>>> that relies on this approach, and i really would like to >>>> understand why the support of globally distributed HAs requires >>>> such approach. I mean one could think of an approach to support=20 >>>> gloablly distributed HAs that do not require having the home=20 >>>> network prefix injected separately in the global BGP table. In=20 >>>> particular, one could envision a solution where each HA has its own=20= >>>> prefix block, belonging to the ISPs aggregate >>> Would you elaborate on this perspective? How would an "ISP=20 >>> aggregate" be able to be present anywhere in the world where the=20 >>> mobile may roam? >> see below... >>> As a side note, do you think an "ISP aggregate" could outnumber all >>> the ISPs present out there? If the media talks about some merger >>> between two large ISPs and then calls that "the largest combined=20 >>> ISP in the world" do you really think that ISP covers the entire=20 >>> Internet? I am sure not. >> actually by ISP aggregate i meant the ISP prefix (which aggregates=20 >> all the prefixes of the ISP's customers), so i guess we only have a=20= >> terminology misunderstanding here.. > > We should make sure that when we rely on an aggregated prefix=20 > advertised > in core to solve mobility problems we don't also secretly assume > specific prefixes being advertised too. > completely agree >>> "Large" ISP in terms of number of subscribers is different than=20 >>> "large" in terms of square kilometers covered, because population=20 >>> density varies so much. A single cellular ISP in China may be=20 >>> bigger than all biggest western ISPs if we talk subscriber count,=20 >>> but of course much smaller than Iridium if we talk square kilometer >>> count. >>>> and that the mobile node prefers to use the address of the closest=20= >>>> HA to initiate new communications. >> I haven't thought this approach completelly but i was thinking=20 >> something in these lines. >> Suppose that a mobile network wants to use 4 HA that are=20 >> geographically distributed. Each of these 4 home networks obtain a=20 >> topologically meaningful prefix from the ISP prefix (which does not=20= >> need to be announced seprately since it is part of the ISP prefix)=20 >> Suppose that the monet gets a home network prefix from each of the=20 >> Home networks > > So the monet gets 4 prefixes? 4 potential Home Addresses for LFNs? > yes, like any multihomed site that gets PA addresses from their=20 upstream providers >> Now, the goal at this point would be to use an address from the home >> network prefix that is closer to the monet in a certain moment in=20 >> time to initiate new communications. If the node in the monet is able >> to choose the right address, then it will be using the HA that is=20 >> closer to the monet achieving similar benefits to the ones obtained=20= >> by the global HAHA in terms of round trip time, agree? > > I think I get the picture. This looks feasible. MR may try to=20 > identify > which MNP (out of the 4) to use for its Home Address. > yes > But this comes down to when MR moves in a different geographical area > stop all its communications (LFNs' too) and then use another Home > Address and start new communications. Ongoing apps stopped. well, no, because you still can get the packets arriving to the=20 previous HA, because there is still the tunnel between the MR and the=20 old HA (i mean, it is not like the MR can use only one HA at the time) The problem is that as the MR moves away from a given HA, the delay in=20= the communication that are being routed through the HA increases. The MR will have 4 tunnels with 4 different HAs. It can communicate=20 through all of them (if they are working of course), and it can receive=20= packets addressed to the HoAs of each HA. The point is that one of the HAs will be closer to the MR, so this=20 would be the preferred one for starting new communications. As the MR moves, the preferred MR can change. this doesn't means that=20 the previous preferred HA becomes unreachable, or that the=20 communications thorugh the previous HA need to be interrupted, just=20 that there is a HA that would provide better performance for new=20 communications. It may well be the case that communication through previous HA get=20 really log delays, that is why i think that a host based RO mechanism=20 could be a good complement to this approach > Is this an accepted assumption? (that no session continuity is=20 > required) > > We've discussed this at some point in the mip4 WG. They have the > Dynamic HA Assignment draft, now maybe an RFC. It was stressed that = is > about an initial assignment of the Home Address and that it wouldn't = be > about session continuity. > i think this is different. Here you have session continuity, it is just=20= that the perfromance can be poor and you may want to use additional RO=20= tools to overcome this limitation > And, if it's agreed that a globalHAHA-like solution is ok to not offer > session continuity then the paradox is that one wouldn't need=20 > globalHAHA > at all - it's sufficient for MR to _discover_ a local HA, obtain a > "local" Home Address and all is fine. > see above > Which is: instead of pre-configuring 4 HAs do HA discovery. = GlobalHAHA > wouldn't be needed (i.e. no need of BGP route injection nor=20 > multi-homing). > I think that supporting multiple HAs in paralel allows to softly moving=20= from one HA to another one, since the MR can evaluate that it is=20 approaching to a different HA and start using the new for new=20 communications, but still preserving the sessions established using the=20= previously preferred HA >> So in order to do that, the gaol would be for the MR to measure the=20= >> distance with the different HA and inform the mobile network nodes to >> preffer the prefix that is associated to the HA that is closer. >> In order to measure the distance with the HA, there are several=20 >> possibilities as you mention below. One option would be to listen to >> BGP and use the AS-path length. > > I see, sounds as a good tool. It assumes though that MR will be=20 > exposed > to the core BGP. An ISP for /24 end-users avoids that. But depends = on > the picture of course. agree > >> Other option is to send probes and measure the RTT, which is likely=20= >> to obtain better results than the BGP based technique. > > Yes, to write that we'd need routing experts involved. > not sure why do you think so... probing to different Has seems quite=20 straight forward to me... what complexities do you see? >> In order to inform the hosts about which address to preffer, it could >> easily do that by deprecating the prefixes of the HAs that are far >> away and only leave the one of the closets HA as preferred. This >> would make that all the new communications to be initiated with the >> address associated to the closest HA > > Ok. > >> There is an underlying assumption here that the movement is slow,=20 >> comparing distance and HAs. I mean, that in most of the cases, the=20 >> address preferred when the communication is initiated will be good=20 >> enough during all the lifetime of the communication. So the solution >> works better in this case. I think this is the case for the airplane >> case presented, since even while the airplane moves very fast, it=20 >> doesn't changes that often of attachment point to the internet, soi=20= >> it doesn't need to change the HA that often (compared with the=20 >> lifetime of the communication) > > I think this is right. > > I think in the case of commercial airline jet liners, sea liners and=20= > similar the problem is even more simple because trajectories are well=20= > known. They are planned. Points of connection are well known in=20 > advance, there'd be little need to "discover" HA or to probe paths. > makes sense to me >>> How would a mobile _know_ which HA is "closest"? IP "distance" is=20= >>> an unreliable variable, nowhere defined. >>> Maybe you meant that the BGP network will direct the BU to the HA=20 >>> that the network thinks it's closer, without MN involvement in the >>> choice? >>>> This with a route optimization solution should provide similar=20 >>>> benefits. >>> An RO solution gets its first packets through HA and then straight >>> to CN, thus eliminating HA from the path. Doesn't need global = HAHA. >> Well, i hope it is clearer now what i had in mind. So there are two=20= >> cases that additional work is needed: - one case is for communication >> that lasts for long and it was initiated with the address associated >> with a previous HA. In this case, this communication would not=20 >> benefit from the closest HA. So in this case we need an additional RO >> to optimize this situation. But this is only for a reduced set of=20 >> communications, so we can use something like host based RO, like=20 >> MIPv6 based RO for this particular nodes. - another case that needs=20= >> to be taken into account is incoming communications. I guess that=20 >> most of the application cases the mobile network nodes are clients,=20= >> rather than servers, but in any case, it would be worthy to consider >> this case. The problem here is that the CN is the one selecting the >> address, so it may well choose not the address corresponding to the >> closest HA. For this cases, we also need some form of host based RO. >> So, i guess this would work fine for outgoing communications that are >> not very long lived. But my feeling is that these are the most of=20 >> the traffic to be expected in many scenarios, like planes and buses=20= >> and trains. > > I think once we accept that most communications are short lived then=20= > we can assume that Mobile IP is not needed at all. There exist widely=20= > deployed end-user applications that offer an illusion of "session=20 > continuity" without using Mobile IP, with reachability at a permanent=20= > address and all that. If we don't use Mobile IP and still offer users=20= > an idea of "session continuity" which s/he's ready to accept then the=20= > whole HA concept is not needed, so even less GlobalHAHA. > >> For the other communications a RO optimization mechanism is needed >> So i see this as an approach that is complementary with a RO=20 >> solution, makes any sense to you? > > I'm trying to, and understood most of the points you made. > > I think that what is needed by deployments is the most important=20 > thing. I think that if we have some set of scenarios from some people=20= > who wish to deploy the global HAHA tech (with an eventual description=20= > of the network topology) and some reasons to use Mobile IP then we'll=20= > get closer to delimiting the globalHAHA problem statement. > well i was taking as a starting point the problem presented by the=20 boeing people and what they are currently doing in the planes to=20 provide connectivity this seems a real problem with a suboptimal expensive solution in IPv4=20= today, and i think that it would be important to provide an better=20 solution than what they are doing today. my goal was to try to figure out a solution that has no impact in the=20 global routing table and that it fullfills their needs regards, marcelo > Sorry for the too long text, > > Alex > From nemo-bounces@ietf.org Wed Apr 05 06:15:09 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR52Y-00073b-OV; Wed, 05 Apr 2006 06:15:06 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR52X-00073W-Se for nemo@ietf.org; Wed, 05 Apr 2006 06:15:05 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR52X-0006vd-2o for nemo@ietf.org; Wed, 05 Apr 2006 06:15:05 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 038EB212C4E; Wed, 5 Apr 2006 13:15:04 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id BF80C212C46; Wed, 5 Apr 2006 13:15:03 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 4D104BDC40; Wed, 5 Apr 2006 13:15:03 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 1418ABDC3D; Wed, 5 Apr 2006 13:15:03 +0300 (EEST) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 13:15:02 +0300 To: "Vijay Devarapalli" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: c0aa019322dfce838bd8604f5a841b57 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "Davis, Terry L" , Alexandru Petrescu , "T.J. Kniveton" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Vijay, El 05/04/2006, a las 1:53, Vijay Devarapalli escribi=F3: > Marcelo, > > you have to consider two things. > > first, MIPv6 and NEMO are based on the concept > of the mobile node and mobile router having the > same home address and maintaining session > continuity without changing the home address. agree > this means for HAHA, the mobile router should > have the same home address irrespective of which > home agent (from the bunch of home agents that > serve its home network) it attaches to. > depends what you are looking for from the presentation from boeing, it seemed to me that their goal was=20= to use the HA that is closer to the current location of the MR. It is=20 not a problem of fault tolerance, but rather a problem of performance.=20= I don't think that the general case there are many long lived=20 communications, but rather some how short lived communications=20 (compared to the time required to change the geographical area covered=20= by one of the HAs). So it makes sense to use MIP because the CoA is=20 changing, so you want to preserve communications through CoA changes,=20 but i guess that is not so common to have communications that last=20 longer than the time required to change HA. for scuh communications=20 additional host based RO mechanisms can be used however, i think i more and more agree with Alex here, perhaps we need=20= to figure out what are the exact requirements that we are trying to=20 address personally, i see that general fault tolerance support for multiple=20 distributed HA and using the HA that is closer to the current location=20= of the MR can be tackled with different tools.... i mean, the first=20 case is just a multihomed site that is attached to multiple points in=20 the internet... > second, it is not realistic to assume the mobile > router can figure out which home agent it is > closest to based on IP hops, round trips, etc.. > these are not always reliable ways of measuring > distance from each home agent. not sure what difficulties you here, i mean defining a probe protocol=20 between the MR and the HA shouldn't be hard... regards, marcelo > > Vijay > >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >> Sent: Tuesday, April 04, 2006 6:43 AM >> To: Alexandru Petrescu >> Cc: Pascal Thubert (pthubert); Davis, Terry L; T.J. Kniveton; >> Vijay Devarapalli; ml-nemo WG >> Subject: Re: About route aggregation in global HAHA (was Re: >> draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) >> >> Hi Alex, >> >> >> El 04/04/2006, a las 16:01, Alexandru Petrescu escribi=F3: >> >>> marcelo bagnulo braun wrote: >>>> Hi all, >>>> global HAHA draft states in section 5.1.1 External routing >>>> In return, each site advertises the aggregation that >> encompasses the >>>> Distributed Home Network aggregation back into the service >> provider >>>> network, to the outside world (the internet). >>>> So, the (distributed) home network prefix is announced through BGP >>>> and this implies that each Home Network prefix will impact with an >>>> additional route in the global BGP table, all agree? >>>> The question at this point is whether this is a reasonable model >>>> Let's see what the current practise in IPv6 is. As today, the >>>> reccommendation is that only ISPs can announce their >> blocks in BGP. >>>> In particular, end sites are discouraged to do that, in >> particular, >>>> multihomed sites have to find an alternative solution rather than >>>> rely on BGP. >>> >>> Additionally, I think that even if two routes are announced >> in the BGP >>> for same prefix, it's insured that the final destination of >> these two >>> routes are in the _same_ physical place (same end user). >>> >> >> well, there are some configurations that use this, basically anycast >> services, like DNS infrastucutre >> >>> Or, global HAHA claims its strengths in having a HA close >> to the point >>> wherever the mobile may be, with HAs placed in various parts of the >>> Internet. I.e. global HAHA requires the use of a >> multiply-announced >>> BGP >>> prefix that eventually ends in _different_ places. >> >> exactly, but i am not sure if this is a problem w.r.t. the >> scalability >> of the global routing system. >> I mean, do you think that there is a difference w.r.t. routing system >> scalability in the case that it is a single multihomed site that >> announces the same prefix through multiple links to the >> internet, or a >> multiple seperate sites announcing the same prefix? >> >>> >>>> Moreover, as today, there is no RIR policy that allows an >> end site to >>>> get their own address space. Only Provider aggregatable >> address is >>>> being handed out by the RIRs. >>>> Now, as is read this proposal, it implies that each >> distributed home >>>> netwrok will need a prefix that is separately announced in the >>>> global BGP table. Now, the question is if this is in line with >>>> current tendency to consider global routing scalability as a top >>>> priority. As i see it, there must be a very important >> reason to push >>>> a solution that relies on this approach, and i really would like to >>>> understand why the support of globally distributed HAs >> requires such >>>> approach. >>>> I mean one could think of an approach to support gloablly >> distributed >>>> HAs that do not require having the home network prefix injected >>>> separately in the global BGP table. In particular, one >> could envision >>>> a solution where each HA has its own prefix block, >> belonging to the >>>> ISPs aggregate >>> >>> Would you elaborate on this perspective? How would an "ISP >> aggregate" >>> be able to be present anywhere in the world where the >> mobile may roam? >>> >> >> see below... >> >>> As a side note, do you think an "ISP aggregate" could >> outnumber all the >>> ISPs present out there? If the media talks about some >> merger between >>> two large ISPs and then calls that "the largest combined ISP in the >>> world" do you really think that ISP covers the entire >> Internet? I am >>> sure not. >>> >> >> actually by ISP aggregate i meant the ISP prefix (which >> aggregates all >> the prefixes of the ISP's customers), so i guess we only have a >> terminology misunderstanding here.. >> >>> "Large" ISP in terms of number of subscribers is different >> than "large" >>> in terms of square kilometers covered, because population density >>> varies >>> so much. A single cellular ISP in China may be bigger than all >>> biggest western ISPs if we talk subscriber count, but of >> course much >>> smaller than Iridium if we talk square kilometer count. >>> >>>> and that the mobile node prefers to use the address of the >> closest HA >>>> to initiate new communications. >>> >> >> I haven't thought this approach completelly but i was thinking >> something in these lines. >> >> Suppose that a mobile network wants to use 4 HA that are >> geographically >> distributed. >> Each of these 4 home networks obtain a topologically >> meaningful prefix >> from the ISP prefix (which does not need to be announced seprately >> since it is part of the ISP prefix) >> Suppose that the monet gets a home network prefix from each >> of the Home >> networks >> >> Now, the goal at this point would be to use an address from the home >> network prefix that is closer to the monet in a certain >> moment in time >> to initiate new communications. If the node in the monet is able to >> choose the right address, then it will be using the HA that is closer >> to the monet achieving similar benefits to the ones obtained by the >> global HAHA in terms of round trip time, agree? >> >> So in order to do that, the gaol would be for the MR to measure the >> distance with the different HA and inform the mobile network nodes to >> preffer the prefix that is associated to the HA that is closer. >> >> In order to measure the distance with the HA, there are several >> possibilities as you mention below. One option would be to listen to >> BGP and use the AS-path length. this is probably not the best >> measure. >> Other option is to send probes and measure the RTT, which is >> likely to >> obtain better results than the BGP based technique. >> In order to inform the hosts about which address to preffer, it could >> easily do that by deprecating the prefixes of the HAs that >> are far away >> and only leave the one of the closets HA as preferred. This >> would make >> that all the new communications to be initiated with the address >> associated to the closest HA >> >> There is an underlying assumption here that the movement is slow, >> comparing distance and HAs. I mean, that in most of the cases, the >> address preferred when the communication is initiated will be good >> enough during all the lifetime of the communication. So the solution >> works better in this case. I think this is the case for the airplane >> case presented, since even while the airplane moves very fast, it >> doesn't changes that often of attachment point to the >> internet, soi it >> doesn't need to change the HA that often (compared with the >> lifetime of >> the communication) >> >>> How would a mobile _know_ which HA is "closest"? IP "distance" is >>> an unreliable variable, nowhere defined. >>> >>> Maybe you meant that the BGP network will direct the BU to >> the HA that >>> the network thinks it's closer, without MN involvement in >> the choice? >>> >>>> This with a route optimization solution should provide similar >>>> benefits. >>> >>> An RO solution gets its first packets through HA and then >> straight to >>> CN, thus eliminating HA from the path. Doesn't need global HAHA. >>> >> >> Well, i hope it is clearer now what i had in mind. >> So there are two cases that additional work is needed: >> - one case is for communication that lasts for long and it was >> initiated with the address associated with a previous HA. In >> this case, >> this communication would not benefit from the closest HA. So in this >> case we need an additional RO to optimize this situation. But this is >> only for a reduced set of communications, so we can use >> something like >> host based RO, like MIPv6 based RO for this particular nodes. >> - another case that needs to be taken into account is incoming >> communications. I guess that most of the application cases the mobile >> network nodes are clients, rather than servers, but in any case, it >> would be worthy to consider this case. The problem here is >> that the CN >> is the one selecting the address, so it may well choose not the >> address corresponding to the closest HA. For this cases, we also need >> some form of host based RO. >> >> So, i guess this would work fine for outgoing communications that are >> not very long lived. But my feeling is that these are the most of the >> traffic to be expected in many scenarios, like planes and buses and >> trains. >> >> For the other communications a RO optimization mechanism is needed >> >> So i see this as an approach that is complementary with a RO >> solution, >> makes any sense to you? >> >> Regards, marcelo >> >>> Alex >>> >> >> > From nemo-bounces@ietf.org Wed Apr 05 06:18:08 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR55U-0001xS-GH; Wed, 05 Apr 2006 06:18:08 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR55S-0001vO-Kd for nemo@ietf.org; Wed, 05 Apr 2006 06:18:06 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR55R-00075a-4Z for nemo@ietf.org; Wed, 05 Apr 2006 06:18:06 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 36741212C4E; Wed, 5 Apr 2006 13:18:04 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id F3B70212C46; Wed, 5 Apr 2006 13:18:03 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id B9595BDC40; Wed, 5 Apr 2006 13:18:03 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 82823BDC3D; Wed, 5 Apr 2006 13:18:03 +0300 (EEST) In-Reply-To: References: <20060330105501.9B9769518C@smtp02.uc3m.es> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo]Charter proposal) Date: Wed, 5 Apr 2006 13:18:02 +0300 To: Ryuji Wakikawa X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: a2c12dacc0736f14d6b540e805505a86 Cc: 'ml-nemo WG' X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Ryuji, El 05/04/2006, a las 5:34, Ryuji Wakikawa escribi=F3: > Hi Marcelo > > Sorry, I'm a bit late to join the discussion.. > > It is not necessary to inject home prefix to BGP. The global HAHA can =20= > run inside IGP, too. > It depends how scale do you operate NEMO on the Internet. > section 5.1.1 states In return, each site advertises the aggregation that encompasses the Distributed Home Network aggregation back into the service provider network, to the outside world (the internet). this seems to me to imply that the home network prefix is announced =20 through BGP to the rest of the internet... i mean if you want to =20 announce something to the rest of the internet it needs to be done with =20= BGP right? so i am not following your point here... regards, marcelo > ryuji > > On 2006/03/30, at 20:01, marcelo bagnulo braun wrote: > >> So, we do agree that global HAHA implies announcing the home network =20= >> into the global BGP routing table, and that each distributed home =20 >> network impacts with an additional entry in the bgp table, right? >> >> wouldn't this imply global routing system scalability concerns? >> >> Regards, marcelo >> >> >> El 30/03/2006, a las 13:54, Ben McCarthy escribi=F3: >> >>> Sorry, I read your initial question wrong, I thought you were =20 >>> referring to >>> Boeing's BGP manipulation technique they use... >>> http://www.apnic.net/meetings/19/docs/sigs/routing/routing-pres-=20 >>> skeen-global >>> -ip-netmob.pdf >>> >>> not Global HA-HA, my mistake. >>> >>> -----Original Message----- >>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>> Sent: 30 March 2006 11:26 >>> To: Ben McCarthy >>> Cc: 'ml-nemo WG'; 'Simon Oosthoek' >>> Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: =20 >>> [nemo]Charter >>> proposal) >>> >>>> >>>> [Ben] Apart from optimising the route between the MR and CNs or >>>> altering >>>> BGP, then the other option is exactly what Pascal is proposing >>>> (switching to >>>> a HA closer to the point-of-attachment to optimise the triangular >>>> route). >>>> >>> >>> perhaps i am failing to understand how this works, but wouldn't this >>> implies that each home network must be announced separately in BGP? >>> >>> regards, marcelo >>> >>>> Apart from that...??? >>>> >>>> Cheers, >>>> >>>> Ben >>>> >>>>> Ben >>>>> >>>>> >>>> >>> >> > From nemo-bounces@ietf.org Wed Apr 05 06:21:04 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR58K-00047C-7T; Wed, 05 Apr 2006 06:21:04 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR58I-00046G-Lt for nemo@ietf.org; Wed, 05 Apr 2006 06:21:02 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR58I-0007F9-7l for nemo@ietf.org; Wed, 05 Apr 2006 06:21:02 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 6200A212C4E; Wed, 5 Apr 2006 13:21:01 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 0081B212C46; Wed, 5 Apr 2006 13:21:01 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id BF096BDC40; Wed, 5 Apr 2006 13:21:00 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 876C6BDC3D; Wed, 5 Apr 2006 13:21:00 +0300 (EEST) In-Reply-To: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> References: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 13:20:59 +0300 To: Ryuji Wakikawa X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 5a9a1bd6c2d06a21d748b7d0070ddcb8 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 5:44, Ryuji Wakikawa escribi=F3: >> it is very common today to advertise reachability >> to a particular network from different points in >> the Internet. > > Yes. This is common operation these days. > in the v4 internet, you mean? In v6 there is no RIR policy that allows an end site to obtain a prefix=20= to announce it to the internet only Provider Aggregatable prefixes are allocated to LIRs (ISPs), so=20 that they can announce a single prefix covering all its providers. trying to announce an end site prefix to the global routing table seems=20= to be against current efforts to preserve the scalability of the global=20= routing table, and should be very well justified imho Regards, marcelo >> ofcourse, I do agree that route aggregation should >> be preserved. > > We have NEMO home network models document:-) > > ryuji > >> Vijay >> >>> >>> imho we should first discuss and figure out if the operational >>> practices resulting from such a solution are acceptable before >>> chartering the work. I mean we should first figure out at least if >>> there seems to be a viable approach that fulfills the needs and >>> preserves scalability of the routing system >>> >>> Regards, marcelo >>> >>> >>>> We can propose global HAHA as a candidate, compete with >>> other proposals >>>> and try to keep the best from all. >>>> >>>> Pascal >>>> >>> >>> >> > From nemo-bounces@ietf.org Wed Apr 05 06:37:07 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5Nq-0008M7-Eq; Wed, 05 Apr 2006 06:37:06 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5Nn-0008M1-VK for nemo@ietf.org; Wed, 05 Apr 2006 06:37:03 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR5Nn-0007Z4-Ah for nemo@ietf.org; Wed, 05 Apr 2006 06:37:03 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 137BF212C4E; Wed, 5 Apr 2006 13:37:02 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id CFC36212C46; Wed, 5 Apr 2006 13:37:01 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 64537BDC40; Wed, 5 Apr 2006 13:37:01 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 2DBE3BDC3D; Wed, 5 Apr 2006 13:37:01 +0300 (EEST) In-Reply-To: <1144208971.21148.60.camel@localhost> References: <1144208971.21148.60.camel@localhost> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <8060790bcb760dced0a56f05ced0002c@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: ingress filtering for (n,n,n) (was Re: [nemo] Charter proposal Date: Wed, 5 Apr 2006 13:37:00 +0300 To: Chan-Wah Ng X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 22bbb45ef41b733eb2d03ee71ece8243 Cc: ml-nemo WG , "T.J. Kniveton" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 6:49, Chan-Wah Ng escribi=F3: > Hello all, > > Thanks Vijay. I am certain that's what TJ have in mind when he placed > in that bullet. > > In fact, if you read through the appendix B in > draft-ietf-nemo-multihoming, you will find that the mechanism = described > can be used to overcome ingress filtering problem as described in 4.2, > although it is worded as a fault tolerance solution. Ingress=20 > filtering, > is in fact main obstacle for fault tolerance in NEMO. > i guess this is the point we disagree imho fault tolerance support presents several difficulties, being=20 ingress filtering one of them, but not sure the most importnat one. And=20= i think that the solution presented in appendix B addresses soem of the=20= difficulties, including ingress filtering but more than that I mean, consider the case of incoming traffic. In this case, there is=20 absolutelly no issues with ingress filtering, right? However, there are still issues with the multiprefix configuration,=20 since packets addressed to a given prefix, will be routed towards a=20 given HA. So, in this case we need that the HA reroute the packets=20 through the alternative HA, so that packets can be delivered to the=20 nemo. This is what is done in the solution of appendix B if i remember=20= correctly, but this has no relation with ingress filtering. So, my point is that the problem that needs to be addressed is the=20 general problem of providing fault tolerance support for the case that=20= the failure affects the MR - HA links. this is an optimization of the=20 general fault tolerance for multihomed nemo support, but it makes sense=20= to optimize it because seems to be a quite common type of failure imho Regards, marcelo > Having said that, I do agree with marcelo on the rewording of the > charter though. It would make it specific to NEMO, and raise less > concern for work scope overlaps in WGs. > > /rgds > /cwng > > On Tue, 2006-04-04 at 16:01 -0700, Vijay Devarapalli wrote: >> Chan Wah presented a bunch of slides at the last >> IETF, explaining what needs to be done in the >> NEMO WG and Monami6 WG when it comes to multihoming. >> >> see the following URL. >> http://www3.ietf.org/proceedings/05nov/slides/nemo-2/sld6.htm >> >> I think thats what the following means. >> >>> Aug 2006 Submit -00 draft on (n,n,n) ingress filtering >> >> Vijay >> >>> -----Original Message----- >>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>> Sent: Tuesday, April 04, 2006 3:53 AM >>> To: T.J. Kniveton >>> Cc: ml-nemo WG >>> Subject: ingress filtering for (n,n,n) (was Re: [nemo] >>> Charter proposal >>> >>> Hi all, >>> >>> In the proposed nemo charter, there is a bullet stating: >>> >>> Aug 2006 Submit -00 draft on (n,n,n) ingress filtering >>> >>> I am not sure i understand this point. >>> I don't think that there is much nemo specific in the ingress >>> filtering problem itself >>> OTOH, i think that the interesting problem to deal with in nemo >>> multihoming is how to provide an optimization for the provision of >>> fault tolerance in the link between the MR and the AR (or the HA) >>> >>> I mean, an important different between the nemo multihoming and the >>> general site multihoming scenario, is that failures are much more >>> likely to occur in the link between the MR and the AR (or the HA). = So >>> haivng special tools to deal with this case would be a useful >>> optimization imho >>> The nemo multihoming analysis draft presents one of such solutions = in >>> the Appendix B and as i have mentioned already, imho the presented >>> soilution has enough merits to be desribed in a separate draft. >>> However, i don't see this as an ingress filtering solution, >>> but rather >>> an optimization for the fault tolerance support for the most common >>> case of failure. It should be noted that the solution >>> presented in the >>> appendix does not requires new protocol extensions, but it would be >>> more a informational work afaict >>> >>> so, not sure what you have in mind with this bullet but i >>> would suggest >>> to reword it into something like >>> Nemo multihoming support Optimizations for MR-AR link fault = tolerance >>> or something more readable but with this ideas. >>> >>> Regards, marcelo >>> >>> >>> El 24/03/2006, a las 22:27, T.J. Kniveton escribi=F3: >>> >>>> Here is a starting point for charter update. Please send >>> your comments. >>>> >>>> TJ >>>> >>> >>> >> > From nemo-bounces@ietf.org Wed Apr 05 06:53:00 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5dD-0005fF-SG; Wed, 05 Apr 2006 06:52:59 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5dC-0005fA-R7 for nemo@ietf.org; Wed, 05 Apr 2006 06:52:58 -0400 Received: from smtp.mei.co.jp ([133.183.129.25] helo=smtp1.mei.co.jp) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR5dC-000881-4k for nemo@ietf.org; Wed, 05 Apr 2006 06:52:58 -0400 Received: from mail-gw.jp.panasonic.com (dodgers.mei.co.jp [157.8.1.150]) by smtp1.mei.co.jp (8.12.11.20060308/3.7W/kings) with ESMTP id k35AqoZr004406; Wed, 5 Apr 2006 19:52:50 +0900 (JST) Received: by mail-gw.jp.panasonic.com (8.11.6p2/3.7W/somlx2) with ESMTP id k35AqpL17187; Wed, 5 Apr 2006 19:52:51 +0900 (JST) Received: from pslexc01.psl.local (localhost [127.0.0.1]) by mail.jp.panasonic.com (8.11.6p2/3.7W/redsox) with ESMTP id k35Aqpc25150; Wed, 5 Apr 2006 19:52:51 +0900 (JST) Received: from bach.sg.panasonic.com ([10.81.113.99]) by pslexc01.psl.local with Microsoft SMTPSVC(6.0.3790.1830); Wed, 5 Apr 2006 18:51:25 +0800 Received: by bach.sg.panasonic.com (Postfix, from userid 1000) id 46E86D41DA; Wed, 5 Apr 2006 18:59:54 +0800 (SGT) Subject: Re: ingress filtering for (n,n,n) (was Re: [nemo] Charter proposal From: Chan-Wah Ng To: marcelo bagnulo braun In-Reply-To: <8060790bcb760dced0a56f05ced0002c@it.uc3m.es> References: <1144208971.21148.60.camel@localhost> <8060790bcb760dced0a56f05ced0002c@it.uc3m.es> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Date: Wed, 05 Apr 2006 18:59:54 +0800 Message-Id: <1144234794.21148.87.camel@localhost> Mime-Version: 1.0 X-Mailer: Evolution 2.4.2.1 X-OriginalArrivalTime: 05 Apr 2006 10:51:25.0344 (UTC) FILETIME=[E26BF200:01C6589E] X-Spam-Score: 0.1 (/) X-Scan-Signature: 29dc808194f5fb921c09d0040806d6eb Cc: ml-nemo WG , "T.J. Kniveton" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hello, Marcelo, On Wed, 2006-04-05 at 13:37 +0300, marcelo bagnulo braun wrote: > El 05/04/2006, a las 6:49, Chan-Wah Ng escribi=C3=B3: >=20 > > Hello all, > > > > Thanks Vijay. I am certain that's what TJ have in mind when he placed > > in that bullet. > > > > In fact, if you read through the appendix B in > > draft-ietf-nemo-multihoming, you will find that the mechanism described > > can be used to overcome ingress filtering problem as described in 4.2, > > although it is worded as a fault tolerance solution. Ingress=20 > > filtering, > > is in fact main obstacle for fault tolerance in NEMO. > > >=20 > i guess this is the point we disagree >=20 > imho fault tolerance support presents several difficulties, being=20 > ingress filtering one of them, but not sure the most importnat one. And=20 > i think that the solution presented in appendix B addresses soem of the=20 > difficulties, including ingress filtering but more than that >=20 > I mean, consider the case of incoming traffic. In this case, there is=20 > absolutelly no issues with ingress filtering, right? True. >=20 > However, there are still issues with the multiprefix configuration,=20 > since packets addressed to a given prefix, will be routed towards a=20 > given HA. So, in this case we need that the HA reroute the packets=20 > through the alternative HA, so that packets can be delivered to the=20 > nemo. This is what is done in the solution of appendix B if i remember=20 > correctly, but this has no relation with ingress filtering. This is going deeper than the scope of the original thread, but this rerouting of packet through the alternative HA is done transparently to the original HA. >=20 > So, my point is that the problem that needs to be addressed is the=20 > general problem of providing fault tolerance support for the case that=20 > the failure affects the MR - HA links. this is an optimization of the=20 > general fault tolerance for multihomed nemo support, but it makes sense=20 > to optimize it because seems to be a quite common type of failure imho >=20 Again, true. I have no disagreement with what you pointed out, but for the sake of discussion, I like to point out a few things: (1) By definition of ingress filtering, the problem will only occur from the direction of MNN towards CN, not the other way round. (2) In this direction, ingress filtering pose a significant, if not the most, problem for fault tolerance to work. (3) A solution which provides fault tolerance capability to a (n, n, n) network can be easily used to overcome ingress filtering problem even if there is no fault. /rgds /cwng > Regards, marcelo >=20 > > Having said that, I do agree with marcelo on the rewording of the > > charter though. It would make it specific to NEMO, and raise less > > concern for work scope overlaps in WGs. > > > > /rgds > > /cwng > > > > On Tue, 2006-04-04 at 16:01 -0700, Vijay Devarapalli wrote: > >> Chan Wah presented a bunch of slides at the last > >> IETF, explaining what needs to be done in the > >> NEMO WG and Monami6 WG when it comes to multihoming. > >> > >> see the following URL. > >> http://www3.ietf.org/proceedings/05nov/slides/nemo-2/sld6.htm > >> > >> I think thats what the following means. > >> > >>> Aug 2006 Submit -00 draft on (n,n,n) ingress filtering > >> > >> Vijay > >> > >>> -----Original Message----- > >>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] > >>> Sent: Tuesday, April 04, 2006 3:53 AM > >>> To: T.J. Kniveton > >>> Cc: ml-nemo WG > >>> Subject: ingress filtering for (n,n,n) (was Re: [nemo] > >>> Charter proposal > >>> > >>> Hi all, > >>> > >>> In the proposed nemo charter, there is a bullet stating: > >>> > >>> Aug 2006 Submit -00 draft on (n,n,n) ingress filtering > >>> > >>> I am not sure i understand this point. > >>> I don't think that there is much nemo specific in the ingress > >>> filtering problem itself > >>> OTOH, i think that the interesting problem to deal with in nemo > >>> multihoming is how to provide an optimization for the provision of > >>> fault tolerance in the link between the MR and the AR (or the HA) > >>> > >>> I mean, an important different between the nemo multihoming and the > >>> general site multihoming scenario, is that failures are much more > >>> likely to occur in the link between the MR and the AR (or the HA). So > >>> haivng special tools to deal with this case would be a useful > >>> optimization imho > >>> The nemo multihoming analysis draft presents one of such solutions in > >>> the Appendix B and as i have mentioned already, imho the presented > >>> soilution has enough merits to be desribed in a separate draft. > >>> However, i don't see this as an ingress filtering solution, > >>> but rather > >>> an optimization for the fault tolerance support for the most common > >>> case of failure. It should be noted that the solution > >>> presented in the > >>> appendix does not requires new protocol extensions, but it would be > >>> more a informational work afaict > >>> > >>> so, not sure what you have in mind with this bullet but i > >>> would suggest > >>> to reword it into something like > >>> Nemo multihoming support Optimizations for MR-AR link fault tolerance > >>> or something more readable but with this ideas. > >>> > >>> Regards, marcelo > >>> > >>> > >>> El 24/03/2006, a las 22:27, T.J. Kniveton escribi=C3=B3: > >>> > >>>> Here is a starting point for charter update. Please send > >>> your comments. > >>>> > >>>> TJ > >>>> > >>> > >>> > >> > > >=20 >=20 From nemo-bounces@ietf.org Wed Apr 05 06:54:52 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5f2-00066M-M5; Wed, 05 Apr 2006 06:54:52 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5f1-000667-1w for nemo@ietf.org; Wed, 05 Apr 2006 06:54:51 -0400 Received: from yui.nc.u-tokyo.ac.jp ([130.69.251.116]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR5f0-0008Cb-FK for nemo@ietf.org; Wed, 05 Apr 2006 06:54:51 -0400 Received: from [203.178.128.44] (n128-44.sfc.wide.ad.jp [203.178.128.44]) (authenticated bits=0) by yui.nc.u-tokyo.ac.jp (8.12.10/8.12.3/Debian-6.4) with ESMTP id k35AslUl017668; Wed, 5 Apr 2006 19:54:47 +0900 In-Reply-To: References: <20060330105501.9B9769518C@smtp02.uc3m.es> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=UTF-8; delsp=yes; format=flowed Message-Id: <32ECD40B-D303-4233-BC87-495A25D5B086@sfc.wide.ad.jp> Content-Transfer-Encoding: quoted-printable From: Ryuji Wakikawa Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo]Charter proposal) Date: Wed, 5 Apr 2006 19:54:46 +0900 To: marcelo bagnulo braun X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.0 (/) X-Scan-Signature: 7fa173a723009a6ca8ce575a65a5d813 Cc: 'ml-nemo WG' X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Marcelo On 2006/04/05, at 19:18, marcelo bagnulo braun wrote: > Hi Ryuji, > > El 05/04/2006, a las 5:34, Ryuji Wakikawa escribi=C3=B3: > >> Hi Marcelo >> >> Sorry, I'm a bit late to join the discussion.. >> >> It is not necessary to inject home prefix to BGP. The global HAHA =20 >> can run inside IGP, too. >> It depends how scale do you operate NEMO on the Internet. >> > > section 5.1.1 states > > In return, each site advertises the aggregation that encompasses =20= > the > Distributed Home Network aggregation back into the service provider > network, to the outside world (the internet). > > this seems to me to imply that the home network prefix is announced =20= > through BGP to the rest of the internet... i mean if you want to =20 > announce something to the rest of the internet it needs to be done =20 > with BGP right? > > so i am not following your point here... Well, if a mobile node roams inside a network (AS) , IGP based =20 anycast routing is enough. Ex. a mobile terminal always connects to operator's network. HAHA running in an AS is also useful if the operator's network covers =20= big area. 1 HA in Madrid, another HA in Barcelona. ryuji > regards, marcelo > > > >> ryuji >> >> On 2006/03/30, at 20:01, marcelo bagnulo braun wrote: >> >>> So, we do agree that global HAHA implies announcing the home =20 >>> network into the global BGP routing table, and that each =20 >>> distributed home network impacts with an additional entry in the =20 >>> bgp table, right? >>> >>> wouldn't this imply global routing system scalability concerns? >>> >>> Regards, marcelo >>> >>> >>> El 30/03/2006, a las 13:54, Ben McCarthy escribi=C3=B3: >>> >>>> Sorry, I read your initial question wrong, I thought you were =20 >>>> referring to >>>> Boeing's BGP manipulation technique they use... >>>> http://www.apnic.net/meetings/19/docs/sigs/routing/routing-pres-=20 >>>> skeen-global >>>> -ip-netmob.pdf >>>> >>>> not Global HA-HA, my mistake. >>>> >>>> -----Original Message----- >>>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>>> Sent: 30 March 2006 11:26 >>>> To: Ben McCarthy >>>> Cc: 'ml-nemo WG'; 'Simon Oosthoek' >>>> Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo]=20 >>>> Charter >>>> proposal) >>>> >>>>> >>>>> [Ben] Apart from optimising the route between the MR and CNs or >>>>> altering >>>>> BGP, then the other option is exactly what Pascal is proposing >>>>> (switching to >>>>> a HA closer to the point-of-attachment to optimise the triangular >>>>> route). >>>>> >>>> >>>> perhaps i am failing to understand how this works, but wouldn't =20 >>>> this >>>> implies that each home network must be announced separately in BGP? >>>> >>>> regards, marcelo >>>> >>>>> Apart from that...??? >>>>> >>>>> Cheers, >>>>> >>>>> Ben >>>>> >>>>>> Ben >>>>>> >>>>>> >>>>> >>>> >>> >> > From nemo-bounces@ietf.org Wed Apr 05 06:59:22 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5jO-0007YS-3u; Wed, 05 Apr 2006 06:59:22 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5jN-0007Ur-5n for nemo@ietf.org; Wed, 05 Apr 2006 06:59:21 -0400 Received: from lennon.multicasttech.com ([63.105.122.7] helo=multicasttech.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR5jM-0008TW-UM for nemo@ietf.org; Wed, 05 Apr 2006 06:59:21 -0400 Received: from [63.105.122.7] (account marshall_eubanks HELO [IPv6:::1]) by multicasttech.com (CommuniGate Pro SMTP 3.4.8) with ESMTP id 4313568; Wed, 05 Apr 2006 06:59:01 -0400 In-Reply-To: References: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: Marshall Eubanks Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 06:59:37 -0400 To: marcelo bagnulo braun X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.0 (/) X-Scan-Signature: 4b800b1eab964a31702fa68f1ff0e955 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , Ryuji Wakikawa , "T.J. Kniveton" , Vijay Devarapalli , "Davis, Terry L" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org On Apr 5, 2006, at 6:20 AM, marcelo bagnulo braun wrote: > > El 05/04/2006, a las 5:44, Ryuji Wakikawa escribi=F3: >>> it is very common today to advertise reachability >>> to a particular network from different points in >>> the Internet. >> >> Yes. This is common operation these days. >> > > in the v4 internet, you mean? > > In v6 there is no RIR policy that allows an end site to obtain a =20 > prefix to announce it to the internet > Yet. There are two ARIN proposals, 2005-1 and 2006-4, to allow this on the =20= table for Montreal. If you are interested, you might want to join the discussion on the ppml list. Regards Marshal Eubanks > only Provider Aggregatable prefixes are allocated to LIRs (ISPs), =20 > so that they can announce a single prefix covering all its providers. > > trying to announce an end site prefix to the global routing table =20 > seems to be against current efforts to preserve the scalability of =20 > the global routing table, and should be very well justified imho > > Regards, marcelo > > >>> ofcourse, I do agree that route aggregation should >>> be preserved. >> >> We have NEMO home network models document:-) >> >> ryuji >> >>> Vijay >>> >>>> >>>> imho we should first discuss and figure out if the operational >>>> practices resulting from such a solution are acceptable before >>>> chartering the work. I mean we should first figure out at least if >>>> there seems to be a viable approach that fulfills the needs and >>>> preserves scalability of the routing system >>>> >>>> Regards, marcelo >>>> >>>> >>>>> We can propose global HAHA as a candidate, compete with >>>> other proposals >>>>> and try to keep the best from all. >>>>> >>>>> Pascal >>>>> >>>> >>>> >>> >> > > From nemo-bounces@ietf.org Wed Apr 05 07:08:41 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5sJ-0004Pe-EV; Wed, 05 Apr 2006 07:08:35 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5sI-0004PZ-88 for nemo@ietf.org; Wed, 05 Apr 2006 07:08:34 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR5sF-0000pC-Hy for nemo@ietf.org; Wed, 05 Apr 2006 07:08:34 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id BC9FE212C4E; Wed, 5 Apr 2006 14:08:29 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 8A1E9212C46; Wed, 5 Apr 2006 14:08:29 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 217B7BDC40; Wed, 5 Apr 2006 14:08:29 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id DFC24BDC3D; Wed, 5 Apr 2006 14:08:28 +0300 (EEST) In-Reply-To: <1144234794.21148.87.camel@localhost> References: <1144208971.21148.60.camel@localhost> <8060790bcb760dced0a56f05ced0002c@it.uc3m.es> <1144234794.21148.87.camel@localhost> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: ingress filtering for (n,n,n) (was Re: [nemo] Charter proposal Date: Wed, 5 Apr 2006 14:08:27 +0300 To: Chan-Wah Ng X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 848ed35f2a4fc0638fa89629cb640f48 Cc: ml-nemo WG , "T.J. Kniveton" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 13:59, Chan-Wah Ng escribi=F3: > > I have no disagreement with what you pointed out, but for the sake of > discussion, I like to point out a few things: > > (1) By definition of ingress filtering, the problem will only occur =20= > from > the direction of MNN towards CN, not the other way round. > agree that is why i described the difficulties in the provision of =20 fault tolernce in this direction, to show that there are relevant =20 issues that have no relation with ingress filtering :-) > (2) In this direction, ingress filtering pose a significant, if not = the > most, problem for fault tolerance to work. well, i guess that the most important issues are posed by failures, =20 rather that filters :-) I mean, in the case of filters, you already know what is wrong and what =20= needs to be done in order to make it work. I mean, you can think of =20 much more general solutions to the ingress filtering problem, like the =20= ones described in =20 http://www.watersprings.org/pub/id/draft-huitema-shim6-ingress-=20 filtering-00.txt This approach allows to send the packets through the correct exit =20 router, so no problems will occur with ingress filters. however, such approach is not good enough to deal with failures. This is because we don't know when failures occur and we need to =20 reroute packets through alternative ISPs which do not match with the =20 prefix included in the source address. I mean, if we want to provide ingress filtering capability, we can do =20= two things: i) allow that packets with an alternative prefix in the =20 source address get routed through an HA/ISP that has not delegated the =20= prefix in the source prefix or ii) make sure that packets are always =20= routed through the correct HA/ISP. However, if you want to deal with failures, the ii) option is no longer =20= avialble, and only i) is possible since the original HA/ISP is no =20 longer there. That is why i think that ingress filtering is just one =20 piece of the faul tolerance provision, and the real problem is fault =20 tolerance as a whole > > (3) A solution which provides fault tolerance capability to a (n, n, = n) > network can be easily used to overcome ingress filtering problem even =20= > if > there is no fault. > agree, but the other way around is not true. I mean a solution that =20 provides ingress fitlering capability may not provide fault tolerance, =20= see =20 http://www.watersprings.org/pub/id/draft-huitema-shim6-ingress-=20 filtering-00.txt regards, marcelo > /rgds > /cwng > >> Regards, marcelo >> >>> Having said that, I do agree with marcelo on the rewording of the >>> charter though. It would make it specific to NEMO, and raise less >>> concern for work scope overlaps in WGs. >>> >>> /rgds >>> /cwng >>> >>> On Tue, 2006-04-04 at 16:01 -0700, Vijay Devarapalli wrote: >>>> Chan Wah presented a bunch of slides at the last >>>> IETF, explaining what needs to be done in the >>>> NEMO WG and Monami6 WG when it comes to multihoming. >>>> >>>> see the following URL. >>>> http://www3.ietf.org/proceedings/05nov/slides/nemo-2/sld6.htm >>>> >>>> I think thats what the following means. >>>> >>>>> Aug 2006 Submit -00 draft on (n,n,n) ingress filtering >>>> >>>> Vijay >>>> >>>>> -----Original Message----- >>>>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>>>> Sent: Tuesday, April 04, 2006 3:53 AM >>>>> To: T.J. Kniveton >>>>> Cc: ml-nemo WG >>>>> Subject: ingress filtering for (n,n,n) (was Re: [nemo] >>>>> Charter proposal >>>>> >>>>> Hi all, >>>>> >>>>> In the proposed nemo charter, there is a bullet stating: >>>>> >>>>> Aug 2006 Submit -00 draft on (n,n,n) ingress filtering >>>>> >>>>> I am not sure i understand this point. >>>>> I don't think that there is much nemo specific in the ingress >>>>> filtering problem itself >>>>> OTOH, i think that the interesting problem to deal with in nemo >>>>> multihoming is how to provide an optimization for the provision of >>>>> fault tolerance in the link between the MR and the AR (or the HA) >>>>> >>>>> I mean, an important different between the nemo multihoming and = the >>>>> general site multihoming scenario, is that failures are much more >>>>> likely to occur in the link between the MR and the AR (or the HA). = =20 >>>>> So >>>>> haivng special tools to deal with this case would be a useful >>>>> optimization imho >>>>> The nemo multihoming analysis draft presents one of such solutions = =20 >>>>> in >>>>> the Appendix B and as i have mentioned already, imho the presented >>>>> soilution has enough merits to be desribed in a separate draft. >>>>> However, i don't see this as an ingress filtering solution, >>>>> but rather >>>>> an optimization for the fault tolerance support for the most = common >>>>> case of failure. It should be noted that the solution >>>>> presented in the >>>>> appendix does not requires new protocol extensions, but it would = be >>>>> more a informational work afaict >>>>> >>>>> so, not sure what you have in mind with this bullet but i >>>>> would suggest >>>>> to reword it into something like >>>>> Nemo multihoming support Optimizations for MR-AR link fault =20 >>>>> tolerance >>>>> or something more readable but with this ideas. >>>>> >>>>> Regards, marcelo >>>>> >>>>> >>>>> El 24/03/2006, a las 22:27, T.J. Kniveton escribi=F3: >>>>> >>>>>> Here is a starting point for charter update. Please send >>>>> your comments. >>>>>> >>>>>> TJ >>>>>> >>>>> >>>>> >>>> >>> >> >> > From nemo-bounces@ietf.org Wed Apr 05 07:12:53 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5wT-0005O8-CE; Wed, 05 Apr 2006 07:12:53 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR5wR-0005Nz-Q8 for nemo@ietf.org; Wed, 05 Apr 2006 07:12:51 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR5wR-0001ES-Ad for nemo@ietf.org; Wed, 05 Apr 2006 07:12:51 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 45EF4212C4E; Wed, 5 Apr 2006 14:12:50 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id D3229212C46; Wed, 5 Apr 2006 14:12:49 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 97E4EBDC40; Wed, 5 Apr 2006 14:12:49 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 5EC24BDC3D; Wed, 5 Apr 2006 14:12:49 +0300 (EEST) In-Reply-To: References: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <8ed1ed6a1bb39d2a5f786f2f500a15bd@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 14:12:48 +0300 To: Marshall Eubanks X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: cf3becbbd6d1a45acbe2ffd4ab88bdc2 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , Ryuji Wakikawa , "T.J. Kniveton" , Vijay Devarapalli , "Davis, Terry L" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 13:59, Marshall Eubanks escribi=F3: > > On Apr 5, 2006, at 6:20 AM, marcelo bagnulo braun wrote: > >> >> El 05/04/2006, a las 5:44, Ryuji Wakikawa escribi=F3: >>>> it is very common today to advertise reachability >>>> to a particular network from different points in >>>> the Internet. >>> >>> Yes. This is common operation these days. >>> >> >> in the v4 internet, you mean? >> >> In v6 there is no RIR policy that allows an end site to obtain a=20 >> prefix to announce it to the internet >> > > Yet. > > There are two ARIN proposals, 2005-1 and 2006-4, to allow this on the=20= > table for Montreal. If you > are interested, you might want to join the discussion on the ppml = list. > indeed, but this exactly refers to what i am arguing for. routing system scalability is a major concern and the community is=20 taking this very seriously. Solutions that inject additional routes to=20= the global routing table are analyzed in depth and very important=20 reasons need to be presented in order to accept them I guess that the discussion in the ppml list proves that, there have=20 been years of discussion, several times the proposal was presented and=20= rejected and studied in more depth this is why i think that we need to be very carefull in proposing=20 solutions that contribute with additional entries in the global routing=20= table and we need to make sure that we have compelling reasons to do=20 so. i don't thionk we are quite there yet in HAHA regards, marcelo > Regards > Marshal Eubanks > >> only Provider Aggregatable prefixes are allocated to LIRs (ISPs), so=20= >> that they can announce a single prefix covering all its providers. >> >> trying to announce an end site prefix to the global routing table=20 >> seems to be against current efforts to preserve the scalability of=20 >> the global routing table, and should be very well justified imho >> >> Regards, marcelo >> >> >>>> ofcourse, I do agree that route aggregation should >>>> be preserved. >>> >>> We have NEMO home network models document:-) >>> >>> ryuji >>> >>>> Vijay >>>> >>>>> >>>>> imho we should first discuss and figure out if the operational >>>>> practices resulting from such a solution are acceptable before >>>>> chartering the work. I mean we should first figure out at least if >>>>> there seems to be a viable approach that fulfills the needs and >>>>> preserves scalability of the routing system >>>>> >>>>> Regards, marcelo >>>>> >>>>> >>>>>> We can propose global HAHA as a candidate, compete with >>>>> other proposals >>>>>> and try to keep the best from all. >>>>>> >>>>>> Pascal >>>>>> >>>>> >>>>> >>>> >>> >> >> > From nemo-bounces@ietf.org Wed Apr 05 07:18:48 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR62A-0000Zw-OP; Wed, 05 Apr 2006 07:18:46 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR629-0000Zr-N1 for nemo@ietf.org; Wed, 05 Apr 2006 07:18:45 -0400 Received: from yui.nc.u-tokyo.ac.jp ([130.69.251.116]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR628-0001i8-57 for nemo@ietf.org; Wed, 05 Apr 2006 07:18:45 -0400 Received: from [203.178.128.44] (n128-44.sfc.wide.ad.jp [203.178.128.44]) (authenticated bits=0) by yui.nc.u-tokyo.ac.jp (8.12.10/8.12.3/Debian-6.4) with ESMTP id k35BIUUl020268; Wed, 5 Apr 2006 20:18:30 +0900 In-Reply-To: References: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=UTF-8; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: Ryuji Wakikawa Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 20:18:29 +0900 To: marcelo bagnulo braun X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.0 (/) X-Scan-Signature: 92df29fa99cf13e554b84c8374345c17 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org On 2006/04/05, at 19:20, marcelo bagnulo braun wrote: > El 05/04/2006, a las 5:44, Ryuji Wakikawa escribi=C3=B3: >>> it is very common today to advertise reachability >>> to a particular network from different points in >>> the Internet. >> >> Yes. This is common operation these days. >> > > in the v4 internet, you mean? > In v6 there is no RIR policy that allows an end site to obtain a =20 > prefix to announce it to the internet > only Provider Aggregatable prefixes are allocated to LIRs (ISPs), =20 > so that they can announce a single prefix covering all its providers. Policy is policy. > trying to announce an end site prefix to the global routing table =20 > seems to be against current efforts to preserve the scalability of =20 > the global routing table, and should be very well justified imho I don't think it effects the scalability of the global routing table. We shouldn't prohibit this network configuration. DNS anycast is already operated in IPv4. It can be operated for IPv6, =20= too. We don't have in mind that /64 prefixes will be used for HAHA. This =20 should be prohibited, but aggregated prefix is possible. The problem you raised here is all about operations. It is important to deliver protocol spec for operators who want to =20 deploy globally scalable system (ex. haha). ryuji > Regards, marcelo > > >>> ofcourse, I do agree that route aggregation should >>> be preserved. >> >> We have NEMO home network models document:-) >> >> ryuji >> >>> Vijay >>> >>>> >>>> imho we should first discuss and figure out if the operational >>>> practices resulting from such a solution are acceptable before >>>> chartering the work. I mean we should first figure out at least if >>>> there seems to be a viable approach that fulfills the needs and >>>> preserves scalability of the routing system >>>> >>>> Regards, marcelo >>>> >>>> >>>>> We can propose global HAHA as a candidate, compete with >>>> other proposals >>>>> and try to keep the best from all. >>>>> >>>>> Pascal >>>>> >>>> >>>> >>> >> > From nemo-bounces@ietf.org Wed Apr 05 07:39:18 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR6M2-0006uj-5z; Wed, 05 Apr 2006 07:39:18 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR6M1-0006ue-3P for nemo@ietf.org; Wed, 05 Apr 2006 07:39:17 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR6M0-0002FF-Jy for nemo@ietf.org; Wed, 05 Apr 2006 07:39:17 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 657F3212C4E; Wed, 5 Apr 2006 14:39:15 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id F1061212C46; Wed, 5 Apr 2006 14:39:14 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 82016BDC40; Wed, 5 Apr 2006 14:39:14 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 49D21BDC3D; Wed, 5 Apr 2006 14:39:14 +0300 (EEST) In-Reply-To: References: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 14:39:13 +0300 To: Ryuji Wakikawa X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 2086112c730e13d5955355df27e3074b Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 14:18, Ryuji Wakikawa escribi=F3: > > On 2006/04/05, at 19:20, marcelo bagnulo braun wrote: >> El 05/04/2006, a las 5:44, Ryuji Wakikawa escribi=F3: >>>> it is very common today to advertise reachability >>>> to a particular network from different points in >>>> the Internet. >>> >>> Yes. This is common operation these days. >>> >> >> in the v4 internet, you mean? >> In v6 there is no RIR policy that allows an end site to obtain a=20 >> prefix to announce it to the internet >> only Provider Aggregatable prefixes are allocated to LIRs (ISPs), so=20= >> that they can announce a single prefix covering all its providers. > > Policy is policy. > >> trying to announce an end site prefix to the global routing table=20 >> seems to be against current efforts to preserve the scalability of=20 >> the global routing table, and should be very well justified imho > > I don't think it effects the scalability of the global routing table. > what do you mean? The global HAHA solution requires (for geographically=20= distibuted HAs in different sites) to inject a seperate prefix. How=20 does this preserves the scalability of the routing system? the number of entries on the global routing system grows linearly with=20= the number of distributed home networks It is exactly the same argument that for multihomed sites. It is not=20 scalable because their contribution to the bgp global table is linear=20 with the number of multihomed sites. > We shouldn't prohibit this network configuration. IETF cannot prohibit configurations i guess but i still don't think we should endorse it by standarizing it > DNS anycast is already operated in IPv4. It can be operated for IPv6,=20= > too. > yes, right, but imho root dns servers are much more critical than=20 distributed home networks > We don't have in mind that /64 prefixes will be used for HAHA. This=20= > should be prohibited, but aggregated prefix is possible. how would you achieve aggregation? > > The problem you raised here is all about operations. > It is important to deliver protocol spec for operators who want to=20 > deploy globally scalable system (ex. haha). > again, how is global haha scalable? I have pointed out that the contribution of global haha to the DFZ is=20 linear with the number of distributed home networks, this is not=20 scalable by my definition from the wikipedia: A routing protocol is considered scalable with respect to network size,=20= if the size of the necessary routing table on each node grows as O(log=20= N), where N is the number of nodes in the network. so linear is not scalable regards, marcelo > ryuji > >> Regards, marcelo >> >> >>>> ofcourse, I do agree that route aggregation should >>>> be preserved. >>> >>> We have NEMO home network models document:-) >>> >>> ryuji >>> >>>> Vijay >>>> >>>>> >>>>> imho we should first discuss and figure out if the operational >>>>> practices resulting from such a solution are acceptable before >>>>> chartering the work. I mean we should first figure out at least if >>>>> there seems to be a viable approach that fulfills the needs and >>>>> preserves scalability of the routing system >>>>> >>>>> Regards, marcelo >>>>> >>>>> >>>>>> We can propose global HAHA as a candidate, compete with >>>>> other proposals >>>>>> and try to keep the best from all. >>>>>> >>>>>> Pascal >>>>>> >>>>> >>>>> >>>> >>> >> > From nemo-bounces@ietf.org Wed Apr 05 11:13:13 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9h3-0004ev-3a; Wed, 05 Apr 2006 11:13:13 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9h2-0004el-5U for nemo@ietf.org; Wed, 05 Apr 2006 11:13:12 -0400 Received: from motgate7.mot.com ([129.188.136.7]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR9h0-0002dG-VI for nemo@ietf.org; Wed, 05 Apr 2006 11:13:12 -0400 Received: from az33exr02.mot.com ([10.64.251.232]) by motgate7.mot.com (8.12.11/Motgate7) with ESMTP id k35Fhre2022318; Wed, 5 Apr 2006 08:43:53 -0700 (MST) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by az33exr02.mot.com (8.13.1/8.13.0) with ESMTP id k35FPp6O001548; Wed, 5 Apr 2006 10:25:52 -0500 (CDT) Message-ID: <4433DE80.9070307@motorola.com> Date: Wed, 05 Apr 2006 17:13:04 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Ryuji Wakikawa References: <20060330105501.9B9769518C@smtp02.uc3m.es> In-Reply-To: Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 08170828343bcf1325e4a0fb4584481c Cc: marcelo bagnulo braun , 'ml-nemo WG' Subject: [nemo] Re: draft-thubert-nemo-global-haha-01.txt X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Ryuji Wakikawa wrote: > Hi Marcelo > > Sorry, I'm a bit late to join the discussion.. > > It is not necessary to inject home prefix to BGP. The global HAHA can > run inside IGP, too. It depends how scale do you operate NEMO on > the Internet. "Global" HAHA inside a "Interior"GP? Inside an IGP domain netlmm can be used, not HAHA. Alex From nemo-bounces@ietf.org Wed Apr 05 11:13:18 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9h8-0004hi-DE; Wed, 05 Apr 2006 11:13:18 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9h7-0004ha-4N for nemo@ietf.org; Wed, 05 Apr 2006 11:13:17 -0400 Received: from blv-smtpout-01.boeing.com ([130.76.32.69]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR9h6-0002dX-JH for nemo@ietf.org; Wed, 05 Apr 2006 11:13:17 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by blv-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id IAA08929; Wed, 5 Apr 2006 08:13:08 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k35FD8T07589; Wed, 5 Apr 2006 08:13:08 -0700 (PDT) Received: from XCH-NW-8V1.nw.nos.boeing.com ([130.247.55.69]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Wed, 5 Apr 2006 08:13:02 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 08:13:02 -0700 Message-ID: <0D090F1E0F5536449C7E6527AFFA280A21BE4F@XCH-NW-8V1.nw.nos.boeing.com> In-Reply-To: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZYorKI4KYDyLF2RDyHvxwJpUL3wQAFWsPA From: "Davis, Terry L" To: "Ryuji Wakikawa" , "marcelo bagnulo braun" X-OriginalArrivalTime: 05 Apr 2006 15:13:02.0749 (UTC) FILETIME=[6ECDDCD0:01C658C3] X-Spam-Score: 0.0 (/) X-Scan-Signature: 22bbb45ef41b733eb2d03ee71ece8243 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Ryuji/Marcelo I would hope that we are NOT designing protocol based on RIR policy! = These do change. And by definition, a mobile network platform (aircraft/ship) is going to = create "route churn" unless you are willing accept declining QOS and = increasing congestion/cost as you move further from your home port/base. I might also point out that if you are planning on designing a protocol = based on the current PA policy, although the protocol might work, the = implementation would most likely fail, at least for those of us that = actually move the links to our mobile platforms (ships & planes) between = continents. I'm not aware of an ISP that provides service on all = continents, and in all countries, and at all airports and ports. I think a global mobility architecture requires ISP independence in its = design. Also to make a fully functional global mobility service, I think we will = need to finally get around to defining "Internet Docking" as an aircraft = could easily need to "dock" with a dozen or more ISP's in a single day. = Today I get to "enjoy the benefits" of NAT at every airport; I would = hope that this is not our future vision either? Finally I'll comment that I can NEVER ever envision anyone building = "critical infrastructure" based on the existing PA policy. Down the = road, we are "considering" whether to use IP based links for the next = global air traffic management system. In that context, consider this: - Would you like your airline to switch providers while you are on their = aircraft at 35,000 feet over the North Pacific? - Or while you just coming in to land at 5 PM in rain storm to JFK? How far do you willing to trust "re-addressing"? Take care Terry L Davis, P.E. IP & Security Aircraft Architect Boeing Commercial Airplanes Formerly: Chief Network Engineer Connexion by Boeing > -----Original Message----- > From: Ryuji Wakikawa [mailto:ryuji@sfc.wide.ad.jp] > Sent: Wednesday, April 05, 2006 4:18 AM > To: marcelo bagnulo braun > Cc: Pascal Thubert (pthubert); T.J. Kniveton; Davis, Terry L; ml-nemo = WG; > Vijay Devarapalli > Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] = Charter > proposal) >=20 >=20 > On 2006/04/05, at 19:20, marcelo bagnulo braun wrote: > > El 05/04/2006, a las 5:44, Ryuji Wakikawa escribi=F3: > >>> it is very common today to advertise reachability > >>> to a particular network from different points in > >>> the Internet. > >> > >> Yes. This is common operation these days. > >> > > > > in the v4 internet, you mean? > > In v6 there is no RIR policy that allows an end site to obtain a > > prefix to announce it to the internet > > only Provider Aggregatable prefixes are allocated to LIRs (ISPs), > > so that they can announce a single prefix covering all its = providers. >=20 > Policy is policy. >=20 > > trying to announce an end site prefix to the global routing table > > seems to be against current efforts to preserve the scalability of > > the global routing table, and should be very well justified imho >=20 > I don't think it effects the scalability of the global routing table. >=20 > We shouldn't prohibit this network configuration. > DNS anycast is already operated in IPv4. It can be operated for IPv6, > too. > We don't have in mind that /64 prefixes will be used for HAHA. This > should be prohibited, but aggregated prefix is possible. >=20 > The problem you raised here is all about operations. > It is important to deliver protocol spec for operators who want to > deploy globally scalable system (ex. haha). >=20 > ryuji >=20 > > Regards, marcelo > > > > > >>> ofcourse, I do agree that route aggregation should > >>> be preserved. > >> > >> We have NEMO home network models document:-) > >> > >> ryuji > >> > >>> Vijay > >>> > >>>> > >>>> imho we should first discuss and figure out if the operational > >>>> practices resulting from such a solution are acceptable before > >>>> chartering the work. I mean we should first figure out at least = if > >>>> there seems to be a viable approach that fulfills the needs and > >>>> preserves scalability of the routing system > >>>> > >>>> Regards, marcelo > >>>> > >>>> > >>>>> We can propose global HAHA as a candidate, compete with > >>>> other proposals > >>>>> and try to keep the best from all. > >>>>> > >>>>> Pascal > >>>>> > >>>> > >>>> > >>> > >> > > From nemo-bounces@ietf.org Wed Apr 05 11:14:35 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9iN-00055Y-50; Wed, 05 Apr 2006 11:14:35 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9iL-00055T-Qx for nemo@ietf.org; Wed, 05 Apr 2006 11:14:33 -0400 Received: from motgate4.mot.com ([144.189.100.102]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR9iJ-0002hj-Ef for nemo@ietf.org; Wed, 05 Apr 2006 11:14:33 -0400 Received: from az33exr02.mot.com (az33exr02.mot.com [10.64.251.232]) by motgate4.mot.com (8.12.11/Motgate4) with ESMTP id k35FRAuC019712; Wed, 5 Apr 2006 08:27:10 -0700 (MST) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by az33exr02.mot.com (8.13.1/8.13.0) with ESMTP id k35FR4FJ002491; Wed, 5 Apr 2006 10:27:05 -0500 (CDT) Message-ID: <4433DECA.2050203@motorola.com> Date: Wed, 05 Apr 2006 17:14:18 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Ryuji Wakikawa Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <108F5BE9DC629140A6E119C8BCDC28FC0109AA50@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: d8ae4fd88fcaf47c1a71c804d04f413d Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Ryuji Wakikawa wrote: > Hi Andrew > > On 2006/03/31, at 2:06, Dul, Andrew L wrote: > >>> -----Original Message----- From: Vijay Devarapalli >>> [mailto:Vijay.Devarapalli@AzaireNet.com] Sent: Wednesday, March >>> 29, 2006 3:34 PM >>>> >>>> Aviation & maritime platforms are globally mobile, because they >>>> are globally mobile this makes route optimization also an >>>> InterDomain Routing issue. >>>> >>>> Even if you had global-haha implemented it would require >>> multiple ASNs >>>> to announce the mobile platform's address space. That would >>>> likely require cooperation among a number of network operators >>>> to truly get route-optimization. These network operators >>>> would also >>> have to agree >>>> to provide transit to other operators where the mobile platform >>>> was currently homed. >>>> >>>> Alternatively the mobile service provider would have to setup >>>> infrastructure in multiple well connected and well peered >>>> global locations just to prevent triangle routing from >>>> occurring. >>> The mobile >>>> service provider would also have to provide a transit network >>>> to transport traffic between the HA and the current attachment >>>> point. >>> >>> is that a problem? this problem is common to all roaming nodes, >>> no? >> >> >> The problem is common but the situation is exacerbated because the >> roaming nodes are often far away from their home network. >> >> If you have to tunnel gigabytes of traffic across town because your >> traffic didn't enter at the closest point, that isn't necessarily >> a problem because the cost of carrying that traffic across town is >> negligible. >> >> However, if you have tunnel gigabytes of traffic across oceans >> because your traffic is being triangle routed that is a problem >> because the cost is substantial. >> >> Route optimization, to me, is not necessarily about always using >> the shortest path from a physical perspective, but it is about >> using the most economic path. Having the best path looks cool, but >> if the cost to tunnel the traffic is negligible compared with the >> increased cost of complexity, I would vote for the negligible >> increase in transit cost. However if the cost of transit is high >> then additional complexity would likely be needed to reduce the >> real cost of transporting bits. >> >> Andrew > > I agree with your statement. > > RO is not always effective (ex. HA is close enough to MN). It totally > depends on topology. MIP6 RO requires several signaling between MN > and CN, but a small optimization may be achieved in some case. This > is problematic, since a mobile node involve this RO operation. And > how shorter path the MN can obtain cannot be predicted before RO > completed. > > HAHA is transparent to end nodes and can be controlled by operators > (not end nodes). This type of solution is doable since the backbone > gets quite first these days. Sorry for disgressing but are we talking netlmm here? Alex From nemo-bounces@ietf.org Wed Apr 05 11:16:38 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9kM-0005Nb-Gt; Wed, 05 Apr 2006 11:16:38 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9kL-0005NW-28 for nemo@ietf.org; Wed, 05 Apr 2006 11:16:37 -0400 Received: from motgate3.mot.com ([144.189.100.103]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR9kJ-0002th-NA for nemo@ietf.org; Wed, 05 Apr 2006 11:16:37 -0400 Received: from az33exr02.mot.com (az33exr02.mot.com [10.64.251.232]) by motgate3.mot.com (8.12.11/Motgate3) with ESMTP id k35Fbcp8029593; Wed, 5 Apr 2006 08:37:38 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by az33exr02.mot.com (8.13.1/8.13.0) with ESMTP id k35FSqja003946; Wed, 5 Apr 2006 10:28:53 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 84F02865980; Wed, 5 Apr 2006 17:16:06 +0200 (CEST) Message-ID: <4433DF36.8060602@motorola.com> Date: Wed, 05 Apr 2006 17:16:06 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Ryuji Wakikawa Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> In-Reply-To: <8D0842AE-EC69-4AA1-A6D6-599FC77C4E4D@sfc.wide.ad.jp> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 4d87d2aa806f79fed918a62e834505ca Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , Vijay Devarapalli , "Davis, Terry L" , marcelo bagnulo braun X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Ryuji Wakikawa wrote: > > On 2006/03/31, at 3:17, Vijay Devarapalli wrote: > >> >>>> The good news is that TJ's revised charter has a line that >>> could cover >>>> these discussions: >>>> Aug 2006 Submit -00 draft on Route Optimization for >>>> geographically distributed HAs >>>> >>> >>> right, but the proposed charted also states that: >>> >>> The WG will: >>> >>> - Ensure that solutions will scale and function for the different >>> mobile network configurations, without requiring changes to >>> Correspondent Nodes in the Internet. All solutions will aim at >>> preserving route aggregation within the Internet and will satisfy an >>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >>> acceptable level of security (a thorough survey of new threats and an >>> analysis of their severity will be conducted for any changes >>> introduced >>> to the network). >>> >>> I guess one of the points Andrew is pointing out is exactly with >>> respect to this issues, namely: >>> >>>>> Even if you had global-haha implemented it would require >>> multiple ASNs >>>>> to announce the mobile platform's address space. That would likely >>> >>> how would this affect aggregation? >> >> it is very common today to advertise reachability >> to a particular network from different points in >> the Internet. > > Yes. This is common operation these days. > >> ofcourse, I do agree that route aggregation should >> be preserved. > > We have NEMO home network models document:-) Itself having issues. We discussed some of them recently here. IIRC it's still not clear how a MR returning home may "match" a prefix (the home prefix) by checking it being shorter than its MNP. Alex From nemo-bounces@ietf.org Wed Apr 05 11:30:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9xg-0003Vu-KX; Wed, 05 Apr 2006 11:30:24 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FR9xf-0003Vp-R2 for nemo@ietf.org; Wed, 05 Apr 2006 11:30:23 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FR9xe-0003V9-I5 for nemo@ietf.org; Wed, 05 Apr 2006 11:30:23 -0400 Received: from il06exr02.mot.com (il06exr02.mot.com [129.188.137.132]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k35FkEOK028599; Wed, 5 Apr 2006 08:46:19 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr02.mot.com (8.13.1/8.13.0) with ESMTP id k35Fhv81009766; Wed, 5 Apr 2006 10:43:57 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 2DC2D865980; Wed, 5 Apr 2006 17:29:58 +0200 (CEST) Message-ID: <4433E275.2080701@motorola.com> Date: Wed, 05 Apr 2006 17:29:57 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Ryuji Wakikawa Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> <41C95BE6-9648-4CD0-827A-4B7E4AD7CB96@sfc.wide.ad.jp> In-Reply-To: <41C95BE6-9648-4CD0-827A-4B7E4AD7CB96@sfc.wide.ad.jp> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 769a46790fb42fbb0b0cc700c82f7081 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , Vijay Devarapalli , "Davis, Terry L" , marcelo bagnulo braun X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Ryuji Wakikawa wrote: > Hi Alex > On 2006/04/04, at 22:01, Alexandru Petrescu wrote: > >> marcelo bagnulo braun wrote: >>> Hi all, >>> global HAHA draft states in section 5.1.1 External routing >>> In return, each site advertises the aggregation that encompasses the >>> Distributed Home Network aggregation back into the service provider >>> network, to the outside world (the internet). >>> So, the (distributed) home network prefix is announced through BGP >>> and this implies that each Home Network prefix will impact with an >>> additional route in the global BGP table, all agree? >>> The question at this point is whether this is a reasonable model >>> Let's see what the current practise in IPv6 is. As today, the >>> reccommendation is that only ISPs can announce their blocks in BGP. >>> In particular, end sites are discouraged to do that, in particular, >>> multihomed sites have to find an alternative solution rather than >>> rely on BGP. >> >> Additionally, I think that even if two routes are announced in the BGP >> for same prefix, it's insured that the final destination of these two >> routes are in the _same_ physical place (same end user). > > VPN can be used. > Did you read draft-wakikawa-mip6-nemo-haha-spec-01.txt? > We describe the various Home Agent configurations. The draft has no text "VPN". How can the VPN concepts be used? Since we've already discussed the draft when it was initially announced, I am open to discuss any scenario that is described inside. Just explain one. >> Or, global HAHA claims its strengths in having a HA close to the point >> wherever the mobile may be, with HAs placed in various parts of the >> Internet. I.e. global HAHA requires the use of a multiply-announced BGP >> prefix that eventually ends in _different_ places. > > For global HAHA, it is not necessarily to use BGP. Any routing protocol > can be used. > Of course, one example is BGP to obtain globally scale route optimization. I am talking about route in the core Internet towards the current CoA, I think you're talking about routes to the HoA. Alex From nemo-bounces@ietf.org Wed Apr 05 11:40:50 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRA7l-0004ei-NZ; Wed, 05 Apr 2006 11:40:49 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRA7k-0004ea-5B for nemo@ietf.org; Wed, 05 Apr 2006 11:40:48 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRA7k-00043m-1j for nemo@ietf.org; Wed, 05 Apr 2006 11:40:48 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FRA7i-00036j-Mg for nemo@ietf.org; Wed, 05 Apr 2006 11:40:47 -0400 Received: from il06exr02.mot.com (il06exr02.mot.com [129.188.137.132]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k35FuqAU000883; Wed, 5 Apr 2006 08:56:52 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr02.mot.com (8.13.1/8.13.0) with ESMTP id k35Fsaeo014919; Wed, 5 Apr 2006 10:54:36 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 8437F865980; Wed, 5 Apr 2006 17:40:36 +0200 (CEST) Message-ID: <4433E4F4.4000306@motorola.com> Date: Wed, 05 Apr 2006 17:40:36 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> <4432AE55.7010805@motorola.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by motgate8.mot.com id k35FuqAU000883 X-Spam-Score: -1.3 (-) X-Scan-Signature: 3b8eea209b62bd15620865bc4fbef8cd Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: >=20 > El 04/04/2006, a las 20:35, Alexandru Petrescu escribi=F3: >=20 >> marcelo bagnulo braun wrote: >>> well, there are some configurations that use this, basically anycast >>> services, like DNS infrastucutre >> >> yes, right, but infrastructure not edge fringes. >> >=20 > absolutely >=20 >>>> Or, global HAHA claims its strengths in having a HA close to the=20 >>>> point wherever the mobile may be, with HAs placed in various parts >>>> of the Internet. I.e. global HAHA requires the use of a=20 >>>> multiply-announced BGP prefix that eventually ends in _different_=20 >>>> places. >>> exactly, but i am not sure if this is a problem w.r.t. the=20 >>> scalability of the global routing system. I mean, do you think that=20 >>> there is a difference w.r.t. routing system scalability in the case=20 >>> that it is a single multihomed site that announces the same prefix=20 >>> through multiple links to the internet, or a multiple seperate sites >>> announcing the same prefix? >> >> Yes, while it's feasible to multi home a few yearly-changing >> short prefixes it's infeasible to multi home several minutely-changing= =20 >> long prefixes. >> >=20 > completely agree, but my point is that this is because stress on the=20 > global routing table (due to additional entries and the dynamic of such= =20 > entries) rather than because they are seperate sites announcing the sam= e=20 > prefix. I mean, having many long prefixes of multihomed sites that=20 > appear and disapear would cause a similar effect even if they are=20 > announced by a single site connected through multiple points to the=20 > internet >=20 >=20 >>>>> Moreover, as today, there is no RIR policy that allows an end site=20 >>>>> to get their own address space. Only Provider aggregatable address=20 >>>>> is being handed out by the RIRs. Now, as is read this proposal, it=20 >>>>> implies that each distributed home netwrok will need >>>>> a prefix that is separately announced in the global BGP table.=20 >>>>> Now, the question is if this is in line with current tendency to >>>>> consider global routing scalability as a top priority. As i see >>>>> it, there must be a very important reason to push a solution >>>>> that relies on this approach, and i really would like to >>>>> understand why the support of globally distributed HAs requires >>>>> such approach. I mean one could think of an approach to support=20 >>>>> gloablly distributed HAs that do not require having the home=20 >>>>> network prefix injected separately in the global BGP table. In=20 >>>>> particular, one could envision a solution where each HA has its own= =20 >>>>> prefix block, belonging to the ISPs aggregate >>>> Would you elaborate on this perspective? How would an "ISP=20 >>>> aggregate" be able to be present anywhere in the world where the=20 >>>> mobile may roam? >>> see below... >>>> As a side note, do you think an "ISP aggregate" could outnumber all >>>> the ISPs present out there? If the media talks about some merger >>>> between two large ISPs and then calls that "the largest combined=20 >>>> ISP in the world" do you really think that ISP covers the entire=20 >>>> Internet? I am sure not. >>> actually by ISP aggregate i meant the ISP prefix (which aggregates=20 >>> all the prefixes of the ISP's customers), so i guess we only have a=20 >>> terminology misunderstanding here.. >> >> We should make sure that when we rely on an aggregated prefix advertis= ed >> in core to solve mobility problems we don't also secretly assume >> specific prefixes being advertised too. >> >=20 > completely agree >=20 >>>> "Large" ISP in terms of number of subscribers is different than=20 >>>> "large" in terms of square kilometers covered, because population=20 >>>> density varies so much. A single cellular ISP in China may be=20 >>>> bigger than all biggest western ISPs if we talk subscriber count,=20 >>>> but of course much smaller than Iridium if we talk square kilometer >>>> count. >>>>> and that the mobile node prefers to use the address of the closest=20 >>>>> HA to initiate new communications. >>> I haven't thought this approach completelly but i was thinking=20 >>> something in these lines. >>> Suppose that a mobile network wants to use 4 HA that are=20 >>> geographically distributed. Each of these 4 home networks obtain a=20 >>> topologically meaningful prefix from the ISP prefix (which does not=20 >>> need to be announced seprately since it is part of the ISP prefix)=20 >>> Suppose that the monet gets a home network prefix from each of the=20 >>> Home networks >> >> So the monet gets 4 prefixes? 4 potential Home Addresses for LFNs? >> >=20 > yes, like any multihomed site that gets PA addresses from their upstrea= m=20 > providers >=20 >>> Now, the goal at this point would be to use an address from the home >>> network prefix that is closer to the monet in a certain moment in=20 >>> time to initiate new communications. If the node in the monet is able >>> to choose the right address, then it will be using the HA that is=20 >>> closer to the monet achieving similar benefits to the ones obtained=20 >>> by the global HAHA in terms of round trip time, agree? >> >> I think I get the picture. This looks feasible. MR may try to identi= fy >> which MNP (out of the 4) to use for its Home Address. >> >=20 > yes >=20 >> But this comes down to when MR moves in a different geographical area >> stop all its communications (LFNs' too) and then use another Home >> Address and start new communications. Ongoing apps stopped. >=20 > well, no, because you still can get the packets arriving to the previou= s=20 > HA, because there is still the tunnel between the MR and the old HA (i=20 > mean, it is not like the MR can use only one HA at the time) >=20 > The problem is that as the MR moves away from a given HA, the delay in=20 > the communication that are being routed through the HA increases. >=20 > The MR will have 4 tunnels with 4 different HAs. It can communicate=20 > through all of them (if they are working of course), and it can receive= =20 > packets addressed to the HoAs of each HA. > The point is that one of the HAs will be closer to the MR, so this woul= d=20 > be the preferred one for starting new communications. >=20 > As the MR moves, the preferred MR can change. this doesn't means that=20 > the previous preferred HA becomes unreachable, or that the=20 > communications thorugh the previous HA need to be interrupted, just tha= t=20 > there is a HA that would provide better performance for new communicati= ons. >=20 > It may well be the case that communication through previous HA get=20 > really log delays, that is why i think that a host based RO mechanism=20 > could be a good complement to this approach >=20 >> Is this an accepted assumption? (that no session continuity is requir= ed) >> >> We've discussed this at some point in the mip4 WG. They have the >> Dynamic HA Assignment draft, now maybe an RFC. It was stressed that i= s >> about an initial assignment of the Home Address and that it wouldn't b= e >> about session continuity. >> >=20 > i think this is different. Here you have session continuity, it is just= =20 > that the perfromance can be poor and you may want to use additional RO=20 > tools to overcome this limitation >=20 >> And, if it's agreed that a globalHAHA-like solution is ok to not offer >> session continuity then the paradox is that one wouldn't need globalHA= HA >> at all - it's sufficient for MR to _discover_ a local HA, obtain a >> "local" Home Address and all is fine. >> >=20 > see above >=20 >> Which is: instead of pre-configuring 4 HAs do HA discovery. GlobalHAH= A >> wouldn't be needed (i.e. no need of BGP route injection nor=20 >> multi-homing). >> >=20 > I think that supporting multiple HAs in paralel allows to softly moving= =20 > from one HA to another one, since the MR can evaluate that it is=20 > approaching to a different HA and start using the new for new=20 > communications, but still preserving the sessions established using the= =20 > previously preferred HA >=20 >>> So in order to do that, the gaol would be for the MR to measure the=20 >>> distance with the different HA and inform the mobile network nodes to >>> preffer the prefix that is associated to the HA that is closer. >>> In order to measure the distance with the HA, there are several=20 >>> possibilities as you mention below. One option would be to listen to >>> BGP and use the AS-path length. >> >> I see, sounds as a good tool. It assumes though that MR will be expos= ed >> to the core BGP. An ISP for /24 end-users avoids that. But depends o= n >> the picture of course. >=20 > agree >=20 >> >>> Other option is to send probes and measure the RTT, which is likely=20 >>> to obtain better results than the BGP based technique. >> >> Yes, to write that we'd need routing experts involved. >> >=20 > not sure why do you think so... probing to different Has seems quite=20 > straight forward to me... what complexities do you see? >=20 >>> In order to inform the hosts about which address to preffer, it could >>> easily do that by deprecating the prefixes of the HAs that are far >>> away and only leave the one of the closets HA as preferred. This >>> would make that all the new communications to be initiated with the >>> address associated to the closest HA >> >> Ok. >> >>> There is an underlying assumption here that the movement is slow,=20 >>> comparing distance and HAs. I mean, that in most of the cases, the=20 >>> address preferred when the communication is initiated will be good=20 >>> enough during all the lifetime of the communication. So the solution >>> works better in this case. I think this is the case for the airplane >>> case presented, since even while the airplane moves very fast, it=20 >>> doesn't changes that often of attachment point to the internet, soi=20 >>> it doesn't need to change the HA that often (compared with the=20 >>> lifetime of the communication) >> >> I think this is right. >> >> I think in the case of commercial airline jet liners, sea liners and=20 >> similar the problem is even more simple because trajectories are well=20 >> known. They are planned. Points of connection are well known in=20 >> advance, there'd be little need to "discover" HA or to probe paths. >> >=20 > makes sense to me >=20 >>>> How would a mobile _know_ which HA is "closest"? IP "distance" is=20 >>>> an unreliable variable, nowhere defined. >>>> Maybe you meant that the BGP network will direct the BU to the HA=20 >>>> that the network thinks it's closer, without MN involvement in the >>>> choice? >>>>> This with a route optimization solution should provide similar=20 >>>>> benefits. >>>> An RO solution gets its first packets through HA and then straight >>>> to CN, thus eliminating HA from the path. Doesn't need global HAHA. >>> Well, i hope it is clearer now what i had in mind. So there are two=20 >>> cases that additional work is needed: - one case is for communication >>> that lasts for long and it was initiated with the address associated >>> with a previous HA. In this case, this communication would not=20 >>> benefit from the closest HA. So in this case we need an additional RO >>> to optimize this situation. But this is only for a reduced set of=20 >>> communications, so we can use something like host based RO, like=20 >>> MIPv6 based RO for this particular nodes. - another case that needs=20 >>> to be taken into account is incoming communications. I guess that=20 >>> most of the application cases the mobile network nodes are clients,=20 >>> rather than servers, but in any case, it would be worthy to consider >>> this case. The problem here is that the CN is the one selecting the >>> address, so it may well choose not the address corresponding to the >>> closest HA. For this cases, we also need some form of host based RO. >>> So, i guess this would work fine for outgoing communications that are >>> not very long lived. But my feeling is that these are the most of=20 >>> the traffic to be expected in many scenarios, like planes and buses=20 >>> and trains. >> >> I think once we accept that most communications are short lived then=20 >> we can assume that Mobile IP is not needed at all. There exist widely= =20 >> deployed end-user applications that offer an illusion of "session=20 >> continuity" without using Mobile IP, with reachability at a permanent=20 >> address and all that. If we don't use Mobile IP and still offer users= =20 >> an idea of "session continuity" which s/he's ready to accept then the=20 >> whole HA concept is not needed, so even less GlobalHAHA. >> >>> For the other communications a RO optimization mechanism is needed >>> So i see this as an approach that is complementary with a RO=20 >>> solution, makes any sense to you? >> >> I'm trying to, and understood most of the points you made. >> >> I think that what is needed by deployments is the most important=20 >> thing. I think that if we have some set of scenarios from some people= =20 >> who wish to deploy the global HAHA tech (with an eventual description=20 >> of the network topology) and some reasons to use Mobile IP then we'll=20 >> get closer to delimiting the globalHAHA problem statement. >> >=20 > well i was taking as a starting point the problem presented by the=20 > boeing people and what they are currently doing in the planes to provid= e=20 > connectivity I'd have reserves. The reports from the aviation industry seem to talk=20 about systems that cover large geographical areas but are still very=20 small closed systems. When the aviation industry says they wouldn't like the plane over US to=20 connect to the HA in EU it's because the satelite link is slow, not=20 because EU is far from US (even if they connected to the HA placed in=20 the US instead - that would still be slow, because of the slow first-hop=20 satelite link). > this seems a real problem with a suboptimal expensive solution in IPv4=20 > today, and i think that it would be important to provide an better=20 > solution than what they are doing today. Having a customer for a potential document is very important, helps=20 delimiting the scope. We should discuss too whether netlmm type of solutions are adapted to=20 that customer's type of problem. > my goal was to try to figure out a solution that has no impact in the=20 > global routing table and that it fullfills their needs Easiest thing would be to get from them whether their system needs to=20 connect to the Internet or not, and how, in the first place. If the=20 system connects to the Internet with a single exit point then there's no=20 question of impact on the global routing table. Alex From nemo-bounces@ietf.org Wed Apr 05 11:48:24 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRAF5-0001AK-SC; Wed, 05 Apr 2006 11:48:23 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRAF4-00018M-TY for nemo@ietf.org; Wed, 05 Apr 2006 11:48:22 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRAF4-0004NY-Kt for nemo@ietf.org; Wed, 05 Apr 2006 11:48:22 -0400 Received: from il06exr02.mot.com (il06exr02.mot.com [129.188.137.132]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k35G4YF8002681; Wed, 5 Apr 2006 09:04:34 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr02.mot.com (8.13.1/8.13.0) with ESMTP id k35G2HC2018975; Wed, 5 Apr 2006 11:02:17 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 0431B865980; Wed, 5 Apr 2006 17:48:18 +0200 (CEST) Message-ID: <4433E6C1.8050209@motorola.com> Date: Wed, 05 Apr 2006 17:48:17 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> In-Reply-To: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> Content-Type: text/plain; charset=ISO-8859-1; format=flowed X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by motgate8.mot.com id k35G4YF8002681 X-Spam-Score: 0.0 (/) X-Scan-Signature: bdc523f9a54890b8a30dd6fd53d5d024 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , Vijay Devarapalli , "Davis, Terry L" , "T.J. Kniveton" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: > Hi Vijay, >=20 > El 05/04/2006, a las 1:53, Vijay Devarapalli escribi=F3: >=20 >> Marcelo, >>=20 >> you have to consider two things. >>=20 >> first, MIPv6 and NEMO are based on the concept of the mobile node >> and mobile router having the same home address and maintaining >> session continuity without changing the home address. >=20 > agree >=20 >> this means for HAHA, the mobile router should have the same home >> address irrespective of which home agent (from the bunch of home >> agents that serve its home network) it attaches to. >>=20 >=20 > depends what you are looking for >=20 > from the presentation from boeing, it seemed to me that their goal > was to use the HA that is closer to the current location of the MR. > It is not a problem of fault tolerance, but rather a problem of > performance. I think there is some misunderstanding here... It is true that in general any mobile needs to connect to the system that is closer to it. But the HA concept was never designed to be close to the mobile, instead it was designed to be able to serve its mobile wherever it may be, especially far away cases. It may be the case that the boeing presentation may actually not need Mobile IP at all, and its HA. Connecting to the access point closest to the mobile, and having all access points linked together can easily offer session continuity without Mobile IP. In general, I am not at all sure where Mobile IP is needed in a closed (or small) system. > I don't think that the general case there are many long lived=20 > communications, but rather some how short lived communications > (compared to the time required to change the geographical area > covered by one of the HAs). So it makes sense to use MIP because the > CoA is changing, so you want to preserve communications through CoA > changes, but i guess that is not so common to have communications > that last longer than the time required to change HA. for scuh > communications additional host based RO mechanisms can be used >=20 > however, i think i more and more agree with Alex here, perhaps we > need to figure out what are the exact requirements that we are trying > to address >=20 > personally, i see that general fault tolerance support for multiple=20 > distributed HA and using the HA that is closer to the current > location of the MR can be tackled with different tools.... i mean, > the first case is just a multihomed site that is attached to multiple > points in the internet... >=20 >> second, it is not realistic to assume the mobile router can figure >> out which home agent it is closest to based on IP hops, round >> trips, etc.. these are not always reliable ways of measuring=20 >> distance from each home agent. >=20 > not sure what difficulties you here, i mean defining a probe protocol > between the MR and the HA shouldn't be hard... It's a probe that draws a conclusion about the intermediary network=20 topology, which may vary a lot after the probe... I think in the routing=20 wisdom these kinds of probing across large domains are considered=20 useless... eliminating the variance of the topology (get a more precise=20 idea about who's closer) involves even more probes, which adds to the=20 load itself... Alex From nemo-bounces@ietf.org Wed Apr 05 12:01:39 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRARu-0005Sk-5T; Wed, 05 Apr 2006 12:01:38 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRARs-0005PA-FH for nemo@ietf.org; Wed, 05 Apr 2006 12:01:36 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRARs-0005Bu-Dp for nemo@ietf.org; Wed, 05 Apr 2006 12:01:36 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FRAJn-0003Ex-Dv for nemo@ietf.org; Wed, 05 Apr 2006 11:53:15 -0400 Received: from il06exr02.mot.com (il06exr02.mot.com [129.188.137.132]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k35G9TUT003750; Wed, 5 Apr 2006 09:09:29 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr02.mot.com (8.13.1/8.13.0) with ESMTP id k35G7CLc021323; Wed, 5 Apr 2006 11:07:13 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 6237B865980; Wed, 5 Apr 2006 17:53:13 +0200 (CEST) Message-ID: <4433E7E9.9020005@motorola.com> Date: Wed, 05 Apr 2006 17:53:13 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Ryuji Wakikawa References: <20060330105501.9B9769518C@smtp02.uc3m.es> <32ECD40B-D303-4233-BC87-495A25D5B086@sfc.wide.ad.jp> In-Reply-To: <32ECD40B-D303-4233-BC87-495A25D5B086@sfc.wide.ad.jp> Content-Type: text/plain; charset=UTF-8; format=flowed X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by motgate8.mot.com id k35G9TUT003750 X-Spam-Score: -1.9 (-) X-Scan-Signature: 82c9bddb247d9ba4471160a9a865a5f3 Cc: marcelo bagnulo braun , 'ml-nemo WG' Subject: [nemo] Re: draft-thubert-nemo-global-haha-01.txt X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Ryuji Wakikawa wrote: > Hi Marcelo >=20 > On 2006/04/05, at 19:18, marcelo bagnulo braun wrote: >=20 >> Hi Ryuji, >> >> El 05/04/2006, a las 5:34, Ryuji Wakikawa escribi=C3=B3: >> >>> Hi Marcelo >>> >>> Sorry, I'm a bit late to join the discussion.. >>> >>> It is not necessary to inject home prefix to BGP. The global HAHA can= =20 >>> run inside IGP, too. >>> It depends how scale do you operate NEMO on the Internet. >>> >> >> section 5.1.1 states >> >> In return, each site advertises the aggregation that encompasses th= e >> Distributed Home Network aggregation back into the service provider >> network, to the outside world (the internet). >> >> this seems to me to imply that the home network prefix is announced=20 >> through BGP to the rest of the internet... i mean if you want to=20 >> announce something to the rest of the internet it needs to be done=20 >> with BGP right? >> >> so i am not following your point here... >=20 > Well, if a mobile node roams inside a network (AS) , IGP based anycast=20 > routing is enough. > Ex. a mobile terminal always connects to operator's network. Right. Ryuji, if anycast inside an IGP is able to solve the mobility problem=20 inside that IGP then it wouldn't need Mobile IP at all, neither the HA=20 concept. Would you agree? > HAHA running in an AS is also useful if the operator's network covers=20 > big area. 1 HA in Madrid, another HA in Barcelona. I think we really have a problem of assimilating geographical distances=20 to IP routing distances... I really think it's a problem. They're so=20 different: my phone and my pda are both in my pocket and they talk to=20 each other but the number of hops between them is about 10. A plane=20 over US talks to a ground site in UK and the IP distance is 1. Mixing these arguments can happen in any direction, to help any proposal. Alex From nemo-bounces@ietf.org Wed Apr 05 12:23:41 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRAnE-0005kE-Ea; Wed, 05 Apr 2006 12:23:40 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRAnD-0005k8-Gd for nemo@ietf.org; Wed, 05 Apr 2006 12:23:39 -0400 Received: from slb-smtpout-01.boeing.com ([130.76.64.48]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRAn9-0006V2-0g for nemo@ietf.org; Wed, 05 Apr 2006 12:23:39 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by slb-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id JAA03653; Wed, 5 Apr 2006 09:23:27 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k35GNQT21442; Wed, 5 Apr 2006 09:23:27 -0700 (PDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Wed, 5 Apr 2006 09:23:19 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 09:23:19 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AA8A@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: <4433E4F4.4000306@motorola.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZYx2GEXuzhLzBaQ/OIOyZ4A7oFaAAAvV+g From: "Dul, Andrew L" To: "ml-nemo WG" , X-OriginalArrivalTime: 05 Apr 2006 16:23:19.0486 (UTC) FILETIME=[402CDDE0:01C658CD] X-Spam-Score: 0.0 (/) X-Scan-Signature: 4d87d2aa806f79fed918a62e834505ca Cc: X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > -----Original Message----- > From: Alexandru Petrescu [mailto:alexandru.petrescu@motorola.com]=20 > Sent: Wednesday, April 05, 2006 8:41 AM > > my goal was to try to figure out a solution that has no=20 > impact in the=20 > > global routing table and that it fullfills their needs I think any solution will have some impact, but the trick is to make it a globally scalable impact. >=20 > Easiest thing would be to get from them whether their system=20 > needs to connect to the Internet or not, and how, in the=20 > first place. If the system connects to the Internet with a=20 > single exit point then there's no question of impact on the=20 > global routing table. >=20 I think this is a question for me... I think you can assume that the aircraft will have a connection to the Internet both for passenger services as well as other crew services. I also think you have to assume that for Internet services we will have to use multiple ISPs. =20 I believe Global-HAHA can be scalable if the individual BGP prefixes are not injected into the global BGP table. However, to still achieve reasonable route optimization one would likely have to push this routing complexity in to the operational world by requiring transit relationships between HA's which are operated by different ISPs. My concern is that the operational complexity will create an environment where global-haha cannot succeed for aircraft and other globally mobile platforms.=20 Another option would be to split the network identifier & locator. There probably should be a bit of thinking about if this would be the best way to go, not just for mobile networks but for BGP scaling in general. This was briefly discussed at the grow/shim6 group last month. http://www.ietf.org/proceedings/06mar/slides/grow-1.pdf http://ietfreport.isoc.org/idref/draft-ietf-ipngwg-gseaddr/=20 An aircraft of the future will also likely have connections to other closed networks. Those networks may have different mobility/scaling problems compared to the Internet and thus the solutions there may be completely different. Andrew From nemo-bounces@ietf.org Wed Apr 05 13:03:16 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRBPW-0003lL-IE; Wed, 05 Apr 2006 13:03:14 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRBPV-0003lF-95 for nemo@ietf.org; Wed, 05 Apr 2006 13:03:13 -0400 Received: from stl-smtpout-01.boeing.com ([130.76.96.56]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRBPT-0007oq-W9 for nemo@ietf.org; Wed, 05 Apr 2006 13:03:13 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by stl-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id MAA02685; Wed, 5 Apr 2006 12:03:05 -0500 (CDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k35H35T06767; Wed, 5 Apr 2006 10:03:05 -0700 (PDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Wed, 5 Apr 2006 10:03:04 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Wed, 5 Apr 2006 10:03:04 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AA8C@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZYmdb+VoUldTR4QNir3ll3EXESUgANjkFw From: "Dul, Andrew L" To: "marcelo bagnulo braun" , "ml-nemo WG" X-OriginalArrivalTime: 05 Apr 2006 17:03:04.0797 (UTC) FILETIME=[CDEE70D0:01C658D2] X-Spam-Score: 0.0 (/) X-Scan-Signature: e5ba305d0e64821bf3d8bc5d3bb07228 Cc: X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org =20 > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > from the presentation from boeing, it seemed to me that their=20 > goal was to use the HA that is closer to the current location=20 > of the MR. It is not a problem of fault tolerance, but rather=20 > a problem of performance.=20 True, one of our main objectives was performance not fault tolerance. > I don't think that the general case there are many long lived=20 > communications, but rather some how short lived=20 > communications (compared to the time required to change the=20 > geographical area covered by one of the HAs). So it makes=20 > sense to use MIP because the CoA is changing, so you want to=20 > preserve communications through CoA changes, but i guess that=20 > is not so common to have communications that last longer than=20 > the time required to change HA. for scuh communications=20 > additional host based RO mechanisms can be used Just one example...we needed to ensure that if you had a VPN session established your connection wouldn't break when we changed satellites & ground stations from Europe to Asia. If you were using host based MIP on the airplane your next upstream network does not change, rather the transport network between the plane and the ground changes. =20 From nemo-bounces@ietf.org Thu Apr 06 00:53:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRMUp-0002tv-0M; Thu, 06 Apr 2006 00:53:27 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRMUm-0002sP-F6 for nemo@ietf.org; Thu, 06 Apr 2006 00:53:24 -0400 Received: from f.csce.kyushu-u.ac.jp ([133.5.33.1]) by ietf-mx.ietf.org with smtp (Exim 4.43) id 1FRMUg-0006rq-JC for nemo@ietf.org; Thu, 06 Apr 2006 00:53:24 -0400 Received: (qmail 9138 invoked from network); 6 Apr 2006 04:53:16 -0000 Received: from unknown (HELO ?172.17.3.211?) (192.168.77.254) by 192.168.77.101 with SMTP; 6 Apr 2006 04:53:16 -0000 Date: Thu, 06 Apr 2006 13:53:16 +0900 From: Teruaki Kitasuka To: manet@ietf.org, mip4@ietf.org, mip6@ietf.org, mipshop@ietf.org, monami6@ietf.org, nemo@ietf.org Message-Id: <20060406135026.5228.KITASUKA@f.csce.kyushu-u.ac.jp> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver. 2.24.02 [ja] X-Spam-Score: 0.0 (/) X-Scan-Signature: 848ed35f2a4fc0638fa89629cb640f48 Cc: Subject: [nemo] CfP - 3rd Int. Conf. on Mobile Computing and Ubiquitous Networking (ICMU2006) at London, UK X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org (Apologies if you receive multiple copies of this message) 2 weeks left for submission. ======================================================================== C A L L F O R P A P E R S ======================================================================== The Third International Conference on Mobile Computing and Ubiquitous Networking (ICMU2006) October 11-13, 2006 BCS London Office, London, UK http://www.icmu.org/icmu2006/ Sponsored by: IPSJ SIG-MBL (Information Processing Society of Japan, Special Interest Group of Mobile Computing and Ubiquitous Networking) Technically co-sponsored by: BCS (British Computer Society), IEE (Institution of Electrical Engineers), and IEEE ITS Society, Technical Committee on Mobile Commuications & Applications Supported by: ICF (International Communications Foundation) SCOPE: Mobile computing aims at ubiquitous user access to computer application. To make it possible, networking technology which enables Internet access anytime and everywhere, i.e. ubiquitous networking is necessary. It is obvious that with the accelerating trends towards mobile networks, ad-hoc networks, and wireless networks, mobile computing and ubiquitous networking will play an important role in future network. This conference is aimed at providing researchers and practitioners in this hot research area a forum for discussion and collaboration. Authors are invited to submit papers addressing, but not limited to, the following topics: - Network management for ubiquitous networking - Data management for mobile computing - Mobile device management - Performance evaluation for mobile computing and ubiquitous networking systems - Broadcast communications - Network architectures, protocols, or service models for ubiquitous networking - Security in mobile computing and ubiquitous networking - Mobile OS - Wireless and mobile communications - 4G wireless communications - Mobile Internetworking - Ad-hoc networks - Network mobility - Mesh networks - Sensor networks - Location-based services SUBMISSION INSTRUCTIONS: Papers are solicited as full papers of no more than 8 pages, each of which will be subject to a full review process. Submission should already follow the author guidelines as specified in the web site below. An electronic, PDF-based submission of papers is mandatory. Please check the web site of the conference http://www.icmu.org/icmu2006/ for further submission instructions. After the conference, a selected number of papers will be published as special issues in IPSJ journal and IPSJ Digital Courier. Please check the web site of IPSJ http://www.ipsj.or.jp/08editt/dc/index.html for the details of IPSJ Digital Coulier. IMPORTANT DATES: ########################################################## # Deadline for submissions: April 22, 2006 # # Notification of acceptance: July 8, 2006 # # Camera ready due: July 31, 2006 # ########################################################## ORGANIZING COMMITTEE: General Chair: Chai-Keong Toh (Queen Mary University of London, UK) Technical Program Committee Co-Chairs: Susumu Ishihara (Shizuoka University, Japan) Eiji Kamioka (National Institute of Informatics, Japan) Publication Co-Chairs: Gen Kitagata (Tohoku University, Japan) Yang Yang (University College London, UK) Publicity Co-Chairs: Teruaki Kitasuka (Kyushu University, Japan) Kun Yang (University of Essex, UK) Local Arrangement Chair: David Martland (Kingston University London, UK) Treasurers: David Martland (Kingston University London, UK) Tomohiko Yagyu (NEC, Japan) Registration Co-Chairs: Jindong Hou (France Telecom, UK) Homare Murakami (NICT, Japan) ICMU STEERING COMMITTEE: Tadanori Mizuno (Shizuoka University, Japan) Osamu Takahashi (Future University-Hakodate, Japan) Takashi Watanabe (Shizuoka University, Japan) Miki Yamamoto (Kansai University, Japan) TECHNICAL PROGRAM COMMITTEE: Mohammed Atiquzzaman (University of Oklahoma, USA) Xiuzhen Cheng (George Washington University, USA) We Duke Cho (Ajou University, Korea) John Gardiner (Bradford University, UK) Teruo Higashino (Osaka University, Japan) Choong Seon Hong (Kyung Hee University, Korea) Russell Hsing (Telcordia Technologies, USA) Chung Ming Huang (National Cheng Kung University, Taiwan) Joseph Hui (Arizona State University, USA) Masugi Inoue (NICT, Japan) Masahiro Ishiyama (Toshiba Corporation, Japan) Holger Karl (University of Paderborn, Germany) Hughes Kester (Qinetiq Ltd, UK) Dongkyun Kim (Kyungpook National University, Korea) Peter Langendoerfer (IHP Microelectronics, Germany) Madjid Merabti (Liverpool John Moores University, UK) Isabelle Moreau (Mitsubishi Electric ITE, France) Hiroyuki Morikawa (University of Tokyo, Japan) Yasuto Nakanishi (Keio University, Japan) Ken Ohta (NTT DoCoMo, Japan) Ryouji Ono (Mitsubishi Electric, Japan) Mohamed Ould-Khaoua (University of Glasgow, UK) Antonio Pescape (University of Napoli, Italy) Petar Popovski (Aalborg University, Denmark) Jeremy Randles (British Telecom, UK) Ichiro Satoh (National Institute of Informatics, Japan) Winston Seah (Institute for Infocomm Research, Singapore) Hiroshi Shigeno (Keio University, Japan) Biplab Sikdar (Rensselaer Polytechnic Institute, USA) David Simplot-Ryl (University of Lille 1, France) Tatsuya Suda (University of California Irvine, USA) Shinta Sugimoto (Nippon Ericsson, Japan) Keisuke Suwa (Musashi Institute of Technology, Japan) Kevin W. Tsai (University of California, Irvine, USA) Xin Wang (Fudan University, China) Neil Williams (ERA Technology Ltd, UK) Nik Van den Wingaert (University of Antwerp, Belgium) Hirozumi Yamaguchi (Osaka University, Japan) Kun Yang (Essex University, UK) Yang Yang (University College London, UK) Masashi Yano (Hitachi Ltd, Japan) Hidetoshi Yokota (KDDI R&D Laboratories, Japan) Ekio Yoneki (Cambridge University, UK) Qing-An Zeng (University of Cincinnati, USA) ======================================================================== From nemo-bounces@ietf.org Thu Apr 06 02:03:19 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNZy-0004Iz-RD; Thu, 06 Apr 2006 02:02:50 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNZx-0004Iu-2i for nemo@ietf.org; Thu, 06 Apr 2006 02:02:49 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRNZu-0000sE-A6 for nemo@ietf.org; Thu, 06 Apr 2006 02:02:49 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 30BFF212CB0; Thu, 6 Apr 2006 09:02:44 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id C1FDF212C4E; Thu, 6 Apr 2006 09:02:43 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 59080BDC40; Thu, 6 Apr 2006 09:02:43 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id D402BBDC3D; Thu, 6 Apr 2006 09:02:42 +0300 (EEST) In-Reply-To: <0D090F1E0F5536449C7E6527AFFA280A21BE4F@XCH-NW-8V1.nw.nos.boeing.com> References: <0D090F1E0F5536449C7E6527AFFA280A21BE4F@XCH-NW-8V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Thu, 6 Apr 2006 09:02:42 +0300 To: "Davis, Terry L" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 325b777e1a3a618c889460b612a65510 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , Ryuji Wakikawa , "T.J. Kniveton" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Terry, El 05/04/2006, a las 18:13, Davis, Terry L escribi=F3: > Ryuji/Marcelo > > I would hope that we are NOT designing protocol based on RIR policy! =20= > These do change. > no, my point about policy was merely to refute arguments that seemed to=20= imply that is common to have end sites injecting their own prefix as=20 seperate entries in bgp. this is not the case in today's IPv6 internet. The goal is to design protocols based on fundamental principles like=20 routing system scalability, which are long lasting > And by definition, a mobile network platform (aircraft/ship) is going=20= > to create "route churn" unless you are willing accept declining QOS=20 > and increasing congestion/cost as you move further from your home=20 > port/base. > I don't follow your argument. You can perfectly envision route optimization mechanisms that allow you=20= to support mobility with enhanced performance wihtout any kind of=20 impact in the global routing. See the RO document for a compilation of=20= many of those approaches. Of course this implies additional work in the=20= edges, but that is what scalability is about, moving load from the core=20= to the edges > I might also point out that if you are planning on designing a=20 > protocol based on the current PA policy, although the protocol might=20= > work, the implementation would most likely fail, at least for those of=20= > us that actually move the links to our mobile platforms (ships &=20 > planes) between continents. I'm not aware of an ISP that provides=20 > service on all continents, and in all countries, and at all airports=20= > and ports. > But that is exactly the point, not needing a single ISP. I think that=20 it would be worth exploring the possibility of having different ISPs,=20 each one of them providing their own address block, and build a system=20= that supports changing from one address block to another. Such a system=20= wouldn't have any impact in the global routing table > I think a global mobility architecture requires ISP independence in=20 > its design. > Why do you think that? i think RO schemes can support global mobility without need for ISP=20 independent addresses being injected in the DFZ I mean with a RO scheme, the HA is simply a rendez vous point and after=20= the initial contact has been achieved, you can just move to RO mode,=20 where the direct communication will flow. If to that you add that you can have multiple HAs in different=20 locations, each one with their own address block, i think the result=20 may be worth exploring > Also to make a fully functional global mobility service, I think we=20 > will need to finally get around to defining "Internet Docking" as an=20= > aircraft could easily need to "dock" with a dozen or more ISP's in a=20= > single day. Today I get to "enjoy the benefits" of NAT at every=20 > airport; I would hope that this is not our future vision either? > > Finally I'll comment that I can NEVER ever envision anyone building=20 > "critical infrastructure" based on the existing PA policy. Down the=20= > road, we are "considering" whether to use IP based links for the next=20= > global air traffic management system. ok, i think we are mixing things here (or at least i am not following=20 this now) I mean, i think there are two different kinds of traffic here: - one is the traffic of the machines that hook up inside the nemo of=20 the plane. This are basically people that are communicating with the=20 rest of the internet doing web, email or any kind of other stuff. This=20= is not critical traffic, in the sense that in general, neither the life=20= of people is at stake nor the health of the Internet depends on that.=20 This traffic can suffer from different delays, but of course the=20 service needs to be acceptable. I think it is somehow acceptable to=20 have handovers delays, like in any other internet communication - the other type of traffic is the traffic control traffic, signalling=20= of the airplane control and so on, which is definitely critical.=20 However, this is very different type of traffic, since i guess the=20 communicating parties are very well detemrined and closed. For this=20 case, i don't think it makes sense to treat it as normal internet=20 traffic. I guess you should have direct links with the earth=20 infrastructure in order to communicate with all the guarantees=20 required. So, for this type of traffic, i fail to see why do you need=20 global internet reachability (probably because of security reasons,=20 this is likely (i hope) to be a closed system and reachability is=20 limited to the trusted parties). So no need to inject global routes for=20= this case. > In that context, consider this: > - Would you like your airline to switch providers while you are on=20 > their aircraft at 35,000 feet over the North Pacific? > - Or while you just coming in to land at 5 PM in rain storm to JFK? > No, but all this is the second type of traffic, that i defeinitly=20 expect to be routed inside a closed secured network, separated from the=20= internet, with private links. Regards, marcelo > How far do you willing to trust "re-addressing"? > > Take care > > Terry L Davis, P.E. > IP & Security Aircraft Architect > Boeing Commercial Airplanes > > Formerly: > Chief Network Engineer > Connexion by Boeing > > >> -----Original Message----- >> From: Ryuji Wakikawa [mailto:ryuji@sfc.wide.ad.jp] >> Sent: Wednesday, April 05, 2006 4:18 AM >> To: marcelo bagnulo braun >> Cc: Pascal Thubert (pthubert); T.J. Kniveton; Davis, Terry L; ml-nemo=20= >> WG; >> Vijay Devarapalli >> Subject: Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo]=20 >> Charter >> proposal) >> >> >> On 2006/04/05, at 19:20, marcelo bagnulo braun wrote: >>> El 05/04/2006, a las 5:44, Ryuji Wakikawa escribi=F3: >>>>> it is very common today to advertise reachability >>>>> to a particular network from different points in >>>>> the Internet. >>>> >>>> Yes. This is common operation these days. >>>> >>> >>> in the v4 internet, you mean? >>> In v6 there is no RIR policy that allows an end site to obtain a >>> prefix to announce it to the internet >>> only Provider Aggregatable prefixes are allocated to LIRs (ISPs), >>> so that they can announce a single prefix covering all its = providers. >> >> Policy is policy. >> >>> trying to announce an end site prefix to the global routing table >>> seems to be against current efforts to preserve the scalability of >>> the global routing table, and should be very well justified imho >> >> I don't think it effects the scalability of the global routing table. >> >> We shouldn't prohibit this network configuration. >> DNS anycast is already operated in IPv4. It can be operated for IPv6, >> too. >> We don't have in mind that /64 prefixes will be used for HAHA. This >> should be prohibited, but aggregated prefix is possible. >> >> The problem you raised here is all about operations. >> It is important to deliver protocol spec for operators who want to >> deploy globally scalable system (ex. haha). >> >> ryuji >> >>> Regards, marcelo >>> >>> >>>>> ofcourse, I do agree that route aggregation should >>>>> be preserved. >>>> >>>> We have NEMO home network models document:-) >>>> >>>> ryuji >>>> >>>>> Vijay >>>>> >>>>>> >>>>>> imho we should first discuss and figure out if the operational >>>>>> practices resulting from such a solution are acceptable before >>>>>> chartering the work. I mean we should first figure out at least = if >>>>>> there seems to be a viable approach that fulfills the needs and >>>>>> preserves scalability of the routing system >>>>>> >>>>>> Regards, marcelo >>>>>> >>>>>> >>>>>>> We can propose global HAHA as a candidate, compete with >>>>>> other proposals >>>>>>> and try to keep the best from all. >>>>>>> >>>>>>> Pascal >>>>>>> >>>>>> >>>>>> >>>>> >>>> >>> > > From nemo-bounces@ietf.org Thu Apr 06 02:06:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNdz-0006MK-Aa; Thu, 06 Apr 2006 02:06:59 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNdt-0006M1-3K for nemo@ietf.org; Thu, 06 Apr 2006 02:06:53 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRNds-00011K-J3 for nemo@ietf.org; Thu, 06 Apr 2006 02:06:53 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 3B32B212CB0; Thu, 6 Apr 2006 09:06:44 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 049FE212C4E; Thu, 6 Apr 2006 09:06:44 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id BE453BDC40; Thu, 6 Apr 2006 09:06:43 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 85790BDC3D; Thu, 6 Apr 2006 09:06:43 +0300 (EEST) In-Reply-To: <4433E4F4.4000306@motorola.com> References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> <4432AE55.7010805@motorola.com> <4433E4F4.4000306@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <0e3184d045b3e08746684b7ad0d030f6@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Aviation requirements (was Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Thu, 6 Apr 2006 09:06:43 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: f607d15ccc2bc4eaf3ade8ffa8af02a0 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 18:40, Alexandru Petrescu escribi=F3: > marcelo bagnulo braun wrote: >>> >> well i was taking as a starting point the problem presented by the=20 >> boeing people and what they are currently doing in the planes to=20 >> provide connectivity > > I'd have reserves. The reports from the aviation industry seem to=20 > talk about systems that cover large geographical areas but are still=20= > very small closed systems. > > When the aviation industry says they wouldn't like the plane over US=20= > to connect to the HA in EU it's because the satelite link is slow, not=20= > because EU is far from US (even if they connected to the HA placed in=20= > the US instead - that would still be slow, because of the slow=20 > first-hop satelite link). > >> this seems a real problem with a suboptimal expensive solution in=20 >> IPv4 today, and i think that it would be important to provide an=20 >> better solution than what they are doing today. > > Having a customer for a potential document is very important, helps=20 > delimiting the scope. > > We should discuss too whether netlmm type of solutions are adapted to=20= > that customer's type of problem. > >> my goal was to try to figure out a solution that has no impact in the=20= >> global routing table and that it fullfills their needs > > Easiest thing would be to get from them whether their system needs to=20= > connect to the Internet or not, and how, in the first place. If the=20= > system connects to the Internet with a single exit point then there's=20= > no question of impact on the global routing table. > Yes, i think that we should first really understand what they need and=20= then try to provide a solution. Perhaps a good input to the wg would a=20= description of their requirements (not limited to the aviation but to=20 devices that move globally, like ships, planes) regards, marcelo > Alex > From nemo-bounces@ietf.org Thu Apr 06 02:14:42 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNlA-0006nU-1a; Thu, 06 Apr 2006 02:14:24 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNl7-0006nM-OT for nemo@ietf.org; Thu, 06 Apr 2006 02:14:21 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRNl7-0001FO-7V for nemo@ietf.org; Thu, 06 Apr 2006 02:14:21 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 2D455212CB0; Thu, 6 Apr 2006 09:14:20 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id E9FB6212C4E; Thu, 6 Apr 2006 09:14:19 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 7E7E4BDC40; Thu, 6 Apr 2006 09:14:19 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 43FEABDC3D; Thu, 6 Apr 2006 09:14:19 +0300 (EEST) In-Reply-To: <4433E6C1.8050209@motorola.com> References: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> <4433E6C1.8050209@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <814b89a1f881f0a369230c3d81760162@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Thu, 6 Apr 2006 09:14:19 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 944ecb6e61f753561f559a497458fb4f Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 18:48, Alexandru Petrescu escribi=F3: > marcelo bagnulo braun wrote: >> Hi Vijay, >> El 05/04/2006, a las 1:53, Vijay Devarapalli escribi=F3: >>> Marcelo, >>> you have to consider two things. >>> first, MIPv6 and NEMO are based on the concept of the mobile node >>> and mobile router having the same home address and maintaining >>> session continuity without changing the home address. >> agree >>> this means for HAHA, the mobile router should have the same home >>> address irrespective of which home agent (from the bunch of home >>> agents that serve its home network) it attaches to. >> depends what you are looking for >> from the presentation from boeing, it seemed to me that their goal >> was to use the HA that is closer to the current location of the MR. >> It is not a problem of fault tolerance, but rather a problem of >> performance. > > I think there is some misunderstanding here... > > It is true that in general any mobile needs to connect to the system > that is closer to it. But the HA concept was never designed to be=20 > close > to the mobile, instead it was designed to be able to serve its mobile > wherever it may be, especially far away cases. > agree, but it is also true that if the HA is far away in the topology,=20= then the resulting performance can be really bad, if all the traffic=20 needs to flow through the HA (which is the current nemo case where=20 there is no RO support). So trying to switch to a more direct path=20 makes sense (how do we do this, is a different story of course) > It may be the case that the boeing presentation may actually not need > Mobile IP at all, and its HA. > > Connecting to the access point closest to the mobile, and having all > access points linked together can easily offer session continuity > without Mobile IP. > > In general, I am not at all sure where Mobile IP is needed in a closed > (or small) system. > yes, but from other comments, i understood that they have different=20 ISPs in different locations, so it is far from being a closed system.=20 They seem to be a mobile network that gets connectivity from different=20= providers as it moves, so proabably they need something like mobile IP=20= that supports moving through different providers >> I don't think that the general case there are many long lived=20 >> communications, but rather some how short lived communications >> (compared to the time required to change the geographical area >> covered by one of the HAs). So it makes sense to use MIP because the >> CoA is changing, so you want to preserve communications through CoA >> changes, but i guess that is not so common to have communications >> that last longer than the time required to change HA. for scuh >> communications additional host based RO mechanisms can be used >> however, i think i more and more agree with Alex here, perhaps we >> need to figure out what are the exact requirements that we are trying >> to address >> personally, i see that general fault tolerance support for multiple=20= >> distributed HA and using the HA that is closer to the current >> location of the MR can be tackled with different tools.... i mean, >> the first case is just a multihomed site that is attached to multiple >> points in the internet... >>> second, it is not realistic to assume the mobile router can figure >>> out which home agent it is closest to based on IP hops, round >>> trips, etc.. these are not always reliable ways of measuring=20 >>> distance from each home agent. >> not sure what difficulties you here, i mean defining a probe protocol >> between the MR and the HA shouldn't be hard... > > It's a probe that draws a conclusion about the intermediary network=20 > topology, which may vary a lot after the probe... I think in the=20 > routing wisdom these kinds of probing across large domains are=20 > considered useless... eliminating the variance of the topology (get a=20= > more precise idea about who's closer) involves even more probes, which=20= > adds to the load itself... > well, their current option is AS path length which doesn't seems to be=20= a very accurate metric in any case. While i agree that probes may not=20 be accurate, i am not sure how inaccurate they are if you compare it=20 with AS path length metric. In any case, i guess that as you say, having more probes can improve=20 the precision, with a cost. OTOH, they are running BGP today, and this=20= implies quite some traffic, with the BGP session and the BGP=20 keepalives, so, perhaps it may ok to analyze how acceptable probes are. regards, marcelo > Alex > From nemo-bounces@ietf.org Thu Apr 06 02:16:40 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNnB-0006z9-65; Thu, 06 Apr 2006 02:16:29 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNnA-0006z4-AE for nemo@ietf.org; Thu, 06 Apr 2006 02:16:28 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRNn5-0001Q4-R0 for nemo@ietf.org; Thu, 06 Apr 2006 02:16:28 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 0534E212CB0; Thu, 6 Apr 2006 09:16:23 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id C3409212C4E; Thu, 6 Apr 2006 09:16:22 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 8B471BDC40; Thu, 6 Apr 2006 09:16:22 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 55D5EBDC3D; Thu, 6 Apr 2006 09:16:22 +0300 (EEST) In-Reply-To: <4433E7E9.9020005@motorola.com> References: <20060330105501.9B9769518C@smtp02.uc3m.es> <32ECD40B-D303-4233-BC87-495A25D5B086@sfc.wide.ad.jp> <4433E7E9.9020005@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <9369ee79e1a57abc78865193bc4fa99a@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Date: Thu, 6 Apr 2006 09:16:21 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 93238566e09e6e262849b4f805833007 Cc: 'ml-nemo WG' , Ryuji Wakikawa Subject: [nemo] Re: draft-thubert-nemo-global-haha-01.txt X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 18:53, Alexandru Petrescu escribi=F3: >> Well, if a mobile node roams inside a network (AS) , IGP based=20 >> anycast routing is enough. >> Ex. a mobile terminal always connects to operator's network. > > Right. > > Ryuji, if anycast inside an IGP is able to solve the mobility problem=20= > inside that IGP then it wouldn't need Mobile IP at all, neither the HA=20= > concept. Would you agree? > this is especailly true considering than the current solution that they=20= are using relies in announcing the airplane prefix through BGP, so if=20 they can rely on BGP convergence times, i guess they can definetly rely=20= on IGP convergence times regards, marcelo From nemo-bounces@ietf.org Thu Apr 06 02:18:36 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNp9-0007Nb-A7; Thu, 06 Apr 2006 02:18:31 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRNp7-0007NP-VI for nemo@ietf.org; Thu, 06 Apr 2006 02:18:29 -0400 Received: from smtp.mei.co.jp ([133.183.129.25] helo=smtp1.mei.co.jp) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRNp6-0001Te-A5 for nemo@ietf.org; Thu, 06 Apr 2006 02:18:29 -0400 Received: from mail-gw.jp.panasonic.com (dodgers.mei.co.jp [157.8.1.150]) by smtp1.mei.co.jp (8.12.11.20060308/3.7W/bulls) with ESMTP id k366IKOB014548; Thu, 6 Apr 2006 15:18:20 +0900 (JST) Received: by mail-gw.jp.panasonic.com (8.11.6p2/3.7W/somlx3) with ESMTP id k366IJ126270; Thu, 6 Apr 2006 15:18:19 +0900 (JST) Received: from pslexc01.psl.local (localhost [127.0.0.1]) by mail.jp.panasonic.com (8.11.6p2/3.7W/redsox) with ESMTP id k366II227268; Thu, 6 Apr 2006 15:18:18 +0900 (JST) Received: from bach.sg.panasonic.com ([10.81.113.99]) by pslexc01.psl.local with Microsoft SMTPSVC(6.0.3790.1830); Thu, 6 Apr 2006 14:16:53 +0800 Received: by bach.sg.panasonic.com (Postfix, from userid 1000) id 35DA7D41DA; Thu, 6 Apr 2006 14:25:29 +0800 (SGT) Subject: Re: Aviation requirements (was Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) From: Chan-Wah Ng To: marcelo bagnulo braun In-Reply-To: <0e3184d045b3e08746684b7ad0d030f6@it.uc3m.es> References: <01851ccb4f5de4f060ab033153790536@it.uc3m.es> <44326E0F.4050100@motorola.com> <4432AE55.7010805@motorola.com> <4433E4F4.4000306@motorola.com> <0e3184d045b3e08746684b7ad0d030f6@it.uc3m.es> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Date: Thu, 06 Apr 2006 14:25:28 +0800 Message-Id: <1144304728.21148.93.camel@localhost> Mime-Version: 1.0 X-Mailer: Evolution 2.4.2.1 X-OriginalArrivalTime: 06 Apr 2006 06:16:53.0227 (UTC) FILETIME=[B2B01BB0:01C65941] X-Spam-Score: 0.1 (/) X-Scan-Signature: 538aad3a3c4f01d8b6a6477ca4248793 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , Alexandru Petrescu , "T.J. Kniveton" , Vijay Devarapalli , "Davis, Terry L" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org On Thu, 2006-04-06 at 09:06 +0300, marcelo bagnulo braun wrote: > El 05/04/2006, a las 18:40, Alexandru Petrescu escribi=C3=B3: >=20 > > marcelo bagnulo braun wrote: > >>> > >> well i was taking as a starting point the problem presented by the=20 > >> boeing people and what they are currently doing in the planes to=20 > >> provide connectivity > > > > I'd have reserves. The reports from the aviation industry seem to=20 > > talk about systems that cover large geographical areas but are still=20 > > very small closed systems. > > > > When the aviation industry says they wouldn't like the plane over US=20 > > to connect to the HA in EU it's because the satelite link is slow, not=20 > > because EU is far from US (even if they connected to the HA placed in=20 > > the US instead - that would still be slow, because of the slow=20 > > first-hop satelite link). > > > >> this seems a real problem with a suboptimal expensive solution in=20 > >> IPv4 today, and i think that it would be important to provide an=20 > >> better solution than what they are doing today. > > > > Having a customer for a potential document is very important, helps=20 > > delimiting the scope. > > > > We should discuss too whether netlmm type of solutions are adapted to=20 > > that customer's type of problem. > > > >> my goal was to try to figure out a solution that has no impact in the=20 > >> global routing table and that it fullfills their needs > > > > Easiest thing would be to get from them whether their system needs to=20 > > connect to the Internet or not, and how, in the first place. If the=20 > > system connects to the Internet with a single exit point then there's=20 > > no question of impact on the global routing table. > > >=20 > Yes, i think that we should first really understand what they need and=20 > then try to provide a solution. Perhaps a good input to the wg would a=20 > description of their requirements (not limited to the aviation but to=20 > devices that move globally, like ships, planes) >=20 That is an excellent idea. That would certainly indicates to us what kind of route optimization schemes (if any) is needed more. And on a related note, I wonder what are the providers/operators view on route optimization? /rgds /cwng From nemo-bounces@ietf.org Thu Apr 06 02:35:13 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRO5C-00008q-Il; Thu, 06 Apr 2006 02:35:06 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRO5B-00008l-Jl for nemo@ietf.org; Thu, 06 Apr 2006 02:35:05 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRO5B-00026R-4j for nemo@ietf.org; Thu, 06 Apr 2006 02:35:05 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 2654F212CB0; Thu, 6 Apr 2006 09:35:04 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id E7EE4212C4E; Thu, 6 Apr 2006 09:35:03 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id B2BE2BDC40; Thu, 6 Apr 2006 09:35:03 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 7F884BDC3D; Thu, 6 Apr 2006 09:35:03 +0300 (EEST) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AA8C@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AA8C@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <9a001afec8c802d2b089dc8b4d97a710@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Thu, 6 Apr 2006 09:35:02 +0300 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 0a7aa2e6e558383d84476dc338324fab Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 20:03, Dul, Andrew L escribi=F3: > > >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] > >> from the presentation from boeing, it seemed to me that their >> goal was to use the HA that is closer to the current location >> of the MR. It is not a problem of fault tolerance, but rather >> a problem of performance. > > True, one of our main objectives was performance not fault tolerance. > >> I don't think that the general case there are many long lived >> communications, but rather some how short lived >> communications (compared to the time required to change the >> geographical area covered by one of the HAs). So it makes >> sense to use MIP because the CoA is changing, so you want to >> preserve communications through CoA changes, but i guess that >> is not so common to have communications that last longer than >> the time required to change HA. for scuh communications >> additional host based RO mechanisms can be used > > Just one example...we needed to ensure that if you had a VPN session > established your connection wouldn't break when we changed satellites = & > ground stations from Europe to Asia. If you were using host based MIP > on the airplane your next upstream network does not change, rather the > transport network between the plane and the ground changes. > ok, but in this case, i guess the micro/localized mobility solutions=20 should provide support for this. Basically what is happening here is that the plane is moving its=20 attachment point inside a closed domain, as Alex was mentioning, so=20 localized mobility mechanisms should work here. So agree that there is no need for MIP to support this case, but i=20 guess the a solution like global HAHA doesn't buy you anything neither.=20= Moreover, i don't understand how injecting a PI prefix would provide=20 better performance than using a PA block from the global provider.... i=20= mean, the global provider prefix is being injected through all the=20 points where the global provide attaches to the internet, so, the CN=20 will reach the global provider through the point that is closer to the=20= CN and then the traffic will be routed through the internal global ISP=20= network Regards, marcelo > > > From nemo-bounces@ietf.org Thu Apr 06 03:19:41 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FROmH-00071U-Mg; Thu, 06 Apr 2006 03:19:37 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FROmG-00071P-SV for nemo@ietf.org; Thu, 06 Apr 2006 03:19:36 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FROmE-0003Gq-Ak for nemo@ietf.org; Thu, 06 Apr 2006 03:19:36 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id A625C212C4E; Thu, 6 Apr 2006 10:19:33 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 73145212C46; Thu, 6 Apr 2006 10:19:33 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 3DF01BDC40; Thu, 6 Apr 2006 10:19:33 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 07CE4BDC3D; Thu, 6 Apr 2006 10:19:33 +0300 (EEST) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AA8A@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AA8A@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Thu, 6 Apr 2006 10:19:31 +0300 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: cd26b070c2577ac175cd3a6d878c6248 Cc: ml-nemo WG , alexandru.petrescu@motorola.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 05/04/2006, a las 19:23, Dul, Andrew L escribi=F3: > > I think this is a question for me... > > I think you can assume that the aircraft will have a connection to the > Internet both for passenger services as well as other crew services. By crew services are you talking about air traffic and navigation=20 control traffic? i.e. critical traffic? (in the sense that the=20 integrity of the airplane depends on it) If this is the case, but wouldn't these be routed through different=20 mechanisms? I mean, i guess you won't share the same channel between=20 the web browsing of one passenger and the communication to the control=20= of the airport. Imho this distinction is quite important because they are likely to=20 have very different QoS requirements > I > also think you have to assume that for Internet services we will have=20= > to > use multiple ISPs. > ok > I believe Global-HAHA can be scalable if the individual BGP prefixes=20= > are > not injected into the global BGP table. what do you call individual prefixes? I mean, as i understand it, if you want to use global HAHA with=20 multiple ISPs, you need to inject the distributed home prefix through=20 the different ISPs, which imho precludes the routing system=20 scalability, so are you considering the distributed home prefix is not an individual=20= prefix, or do you envision some way of running global HAHA wihtout=20 injecting the dsitributed home prefix in bgp or other option that i=20 missing? regards, marcelo > However, to still achieve > reasonable route optimization one would likely have to push this=20 > routing > complexity in to the operational world by requiring transit > relationships between HA's which are operated by different ISPs. My > concern is that the operational complexity will create an environment > where global-haha cannot succeed for aircraft and other globally = mobile > platforms. > > Another option would be to split the network identifier & locator. > There probably should be a bit of thinking about if this would be the > best way to go, not just for mobile networks but for BGP scaling in > general. This was briefly discussed at the grow/shim6 group last=20 > month. > > http://www.ietf.org/proceedings/06mar/slides/grow-1.pdf > http://ietfreport.isoc.org/idref/draft-ietf-ipngwg-gseaddr/ > > An aircraft of the future will also likely have connections to other > closed networks. Those networks may have different mobility/scaling > problems compared to the Internet and thus the solutions there may be > completely different. > > Andrew > > From nemo-bounces@ietf.org Thu Apr 06 11:02:47 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRW0P-0007jq-Vg; Thu, 06 Apr 2006 11:02:41 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRW0N-0007jl-Su for nemo@ietf.org; Thu, 06 Apr 2006 11:02:40 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRW0N-0002sr-Ig for nemo@ietf.org; Thu, 06 Apr 2006 11:02:39 -0400 Received: from il06exr03.mot.com (il06exr03.mot.com [129.188.137.133]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k36FIkDP009951; Thu, 6 Apr 2006 08:18:50 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr03.mot.com (8.13.1/8.13.0) with ESMTP id k36FJ8Yk002125; Thu, 6 Apr 2006 10:19:08 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 13520865980; Thu, 6 Apr 2006 17:02:30 +0200 (CEST) Message-ID: <44352D85.5040006@motorola.com> Date: Thu, 06 Apr 2006 17:02:29 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> <4433E6C1.8050209@motorola.com> <814b89a1f881f0a369230c3d81760162@it.uc3m.es> In-Reply-To: <814b89a1f881f0a369230c3d81760162@it.uc3m.es> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 057ebe9b96adec30a7efb2aeda4c26a4 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: >> It is true that in general any mobile needs to connect to the >> system that is closer to it. But the HA concept was never designed >> to be close to the mobile, instead it was designed to be able to >> serve its mobile wherever it may be, especially far away cases. >> > > agree, but it is also true that if the HA is far away in the > topology, then the resulting performance can be really bad, if all > the traffic needs to flow through the HA (which is the current nemo > case where there is no RO support). ... _if_ Mobile IP is used. One can build moving networks without Mobile IP, and still offer continuous sessions or reachability at a more permanent address. I'm trying to say that when we have an clear idea about the aviation topology, requirements, points of control then we can decide whether or not we need Mobile IP for aviation. The word "Mobile" in Mobile IP makes people tend to think that all that moves and connects to Internet should use Mobile IP. That's not true. >> In general, I am not at all sure where Mobile IP is needed in a >> closed (or small) system. >> > > yes, but from other comments, i understood that they have different > ISPs in different locations, so it is far from being a closed system. If they control all these ISPs it all becomes a "closed" system, manageable by a single entity. If the aviation industry _can_ say to ISP1 "please place a HA here" to ISP2 "please place a HA there" then it means that they can already tell ISPs what to do. And if they can do that, then they can also tell ISPs "ISPs please exchange routes about the MR when you sense it in your location". And if the ISPs exchange these routes then no need of Mobile IP. If however the ISPs are reluctant to place mutually trusted HAs, or if the ISPs are not straighlty connected together (connected without BGP) then they won't be able to place HAs nor to exchange routes. To say that if ISPs can place HAs at different places then they can also exchange routes; and if they can exchange routes then no Mobile IP, no HA. > They seem to be a mobile network that gets connectivity from > different providers as it moves, so proabably they need something > like mobile IP that supports moving through different providers IMHO no no; if the providers can use Mobile IP for different HAs then they can also exchange routes. If they can't exchange routes then they can't place HAs either. _If_ there's a link (physical) between the ISPs which doesn't influence the global routing tables then it is possible for the ISPs to exchange routes directly, again no Mobile IP. >>> I don't think that the general case there are many long lived >>> communications, but rather some how short lived communications >>> (compared to the time required to change the geographical area >>> covered by one of the HAs). So it makes sense to use MIP because >>> the CoA is changing, so you want to preserve communications >>> through CoA changes, but i guess that is not so common to have >>> communications that last longer than the time required to change >>> HA. for scuh communications additional host based RO mechanisms >>> can be used however, i think i more and more agree with Alex >>> here, perhaps we need to figure out what are the exact >>> requirements that we are trying to address personally, i see that >>> general fault tolerance support for multiple distributed HA and >>> using the HA that is closer to the current location of the MR can >>> be tackled with different tools.... i mean, the first case is >>> just a multihomed site that is attached to multiple points in the >>> internet... >>>> second, it is not realistic to assume the mobile router can >>>> figure out which home agent it is closest to based on IP hops, >>>> round trips, etc.. these are not always reliable ways of >>>> measuring distance from each home agent. >>> not sure what difficulties you here, i mean defining a probe >>> protocol between the MR and the HA shouldn't be hard... >> >> It's a probe that draws a conclusion about the intermediary network >> topology, which may vary a lot after the probe... I think in the >> routing wisdom these kinds of probing across large domains are >> considered useless... eliminating the variance of the topology (get >> a more precise idea about who's closer) involves even more probes, >> which adds to the load itself... >> > > well, their current option is AS path length which doesn't seems to > be a very accurate metric in any case. While i agree that probes may > not be accurate, i am not sure how inaccurate they are if you compare > it with AS path length metric. In any case, i guess that as you say, > having more probes can improve the precision, with a cost. OTOH, they > are running BGP today, and this implies quite some traffic, with the > BGP session and the BGP keepalives, so, perhaps it may ok to analyze > how acceptable probes are. What is the risk of probing giving a false indication? It nullifies the entire goal of global HAHA. I wouldn't risk the main requirement of a protocol on this kind of unreliable probe. If the probing were just a hint, in addition to other info, then yes. Alex From nemo-bounces@ietf.org Fri Apr 07 01:42:54 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRjk5-0002ue-HQ; Fri, 07 Apr 2006 01:42:45 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRjk3-0002u5-UX for nemo@ietf.org; Fri, 07 Apr 2006 01:42:43 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRjk3-0006MB-E2 for nemo@ietf.org; Fri, 07 Apr 2006 01:42:43 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 58EC5212CB0; Fri, 7 Apr 2006 08:42:36 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id EA366212C4E; Fri, 7 Apr 2006 08:42:35 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 817E1BDC40; Fri, 7 Apr 2006 08:42:35 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 49CB3BDC3D; Fri, 7 Apr 2006 08:42:35 +0300 (EEST) In-Reply-To: <44352D85.5040006@motorola.com> References: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> <4433E6C1.8050209@motorola.com> <814b89a1f881f0a369230c3d81760162@it.uc3m.es> <44352D85.5040006@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <2877a0e9ca86d7651130007f1c1f8d67@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Fri, 7 Apr 2006 08:42:34 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: a7d2e37451f7f22841e3b6f40c67db0f Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 06/04/2006, a las 18:02, Alexandru Petrescu escribi=F3: >>> In general, I am not at all sure where Mobile IP is needed in a >>> closed (or small) system. >> yes, but from other comments, i understood that they have different >> ISPs in different locations, so it is far from being a closed system. > > If they control all these ISPs it all becomes a "closed" system,=20 > manageable by a single entity. if all the ISPs involved are peering directly this may be easy, but if=20= they are not, they need to start configuring tunnels between them to=20 pass the routing information and the situation becomes rather messy and=20= hard to manage imho (but i agree that is an option... see more below) > If the aviation industry _can_ say to ISP1 "please place a HA here"=20= > to ISP2 "please place a HA there" then it means that they can already=20= > tell ISPs what to do. > > And if they can do that, then they can also tell ISPs "ISPs please=20 > exchange routes about the MR when you sense it in your location". And=20= > if the ISPs exchange these routes then no need of Mobile IP. > but are you assuming that the nemo has its own PI prefix or not? - if it has its own PI prefix, this is basically what has been=20 presented in the slides by boeing, that is what they are currently=20 doing. the problem with this is that each nemo adds an additional entry=20= to the global routing table and they depend on BGP convergence times=20 (which won't be that good if they keep on adding routes i am afraid) - if the nemo do not have its own PI prefix, then it has a PA prefix=20 from one of the ISP, which means that the traffic from the internet to=20= the nemo is always routed through that particular ISP, which in the=20 case that the CN and the nemo are topologically far away from that ISP,=20= then the resulting performance may be pretty bad i guess That is why, i think that a solution that support some form of RO would=20= be good for this case (mixed with having multiple PA prefixes, in order=20= to reduce the need for using the RO solution) > If however the ISPs are reluctant to place mutually trusted HAs, or if=20= > the ISPs are not straighlty connected together (connected without BGP)=20= > then they won't be able to place HAs nor to exchange routes. > well, if they are not directly connected they may dtill be willing to=20 place HAs In addition, it is also possible that the ground infrastructure has=20 internet access, so that the HA can be located in there. This seems=20 much easier that getting each local ISP to peer > To say that if ISPs can place HAs at different places then they can=20 > also exchange routes; I am not sure that is true, i mean selling HA services has much less=20 impact in the ISP netwrok infrastructure than exchenging routes.=20 BEsides, it is possible for the ground infrastrucutre to have a local=20 HA, not located inside the ISP ... > What is the risk of probing giving a false indication? It nullifies=20= > the entire goal of global HAHA. I wouldn't risk the main requirement=20= > of a protocol on this kind of unreliable probe. If the probing were=20= > just a hint, in addition to other info, then yes. > well, another option would be to do use the BGP AS path length as a=20 complementary metric. However, what i don't like about this is to=20 require the MR to run BGP. However, it would be possible to workaround=20= this requirement, by requiring the HAs to acquire BGP information and=20 inform the MR about the distance between the local HA and the COA where=20= the MR is currently located. In this way, each HA would inform the As=20 distance between it and the MR and the MR can choose the prefix of the=20= closest HA as the preferred prefix for the MNN to initiate new=20 communications... do you think this would be more reliable? (i still think that probes would provide more accurate information,=20 but...) regards, marcelo > Alex > > From nemo-bounces@ietf.org Fri Apr 07 06:20:47 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRo4B-0007X7-Oo; Fri, 07 Apr 2006 06:19:47 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRo4A-0007X2-HD for nemo@ietf.org; Fri, 07 Apr 2006 06:19:46 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRo4A-0006qT-6Q for nemo@ietf.org; Fri, 07 Apr 2006 06:19:46 -0400 Received: from il06exr04.mot.com (il06exr04.mot.com [129.188.137.134]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k37AZrwK009076; Fri, 7 Apr 2006 03:35:57 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr04.mot.com (8.13.1/8.13.0) with ESMTP id k37AWnEf018659; Fri, 7 Apr 2006 05:32:50 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 6E52C8637E6; Fri, 7 Apr 2006 12:19:25 +0200 (CEST) Message-ID: <44363CAD.9050406@motorola.com> Date: Fri, 07 Apr 2006 12:19:25 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re: draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <7fbebe6ace150de0260788f182cef53a@it.uc3m.es> <4433E6C1.8050209@motorola.com> <814b89a1f881f0a369230c3d81760162@it.uc3m.es> <44352D85.5040006@motorola.com> <2877a0e9ca86d7651130007f1c1f8d67@it.uc3m.es> In-Reply-To: <2877a0e9ca86d7651130007f1c1f8d67@it.uc3m.es> Content-Type: text/plain; charset=ISO-8859-1; format=flowed X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by motgate8.mot.com id k37AZrwK009076 X-Spam-Score: 0.0 (/) X-Scan-Signature: 67c1ea29f88502ef6a32ccec927970f0 Cc: ml-nemo WG , "Pascal Thubert \(pthubert\)" , "T.J. Kniveton" , "Davis, Terry L" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: >=20 > El 06/04/2006, a las 18:02, Alexandru Petrescu escribi=F3: >=20 >>>> In general, I am not at all sure where Mobile IP is needed in a >>>> closed (or small) system. >>> yes, but from other comments, i understood that they have=20 >>> different ISPs in different locations, so it is far from being a >>> closed system. >>=20 >> If they control all these ISPs it all becomes a "closed" system,=20 >> manageable by a single entity. >=20 > if all the ISPs involved are peering directly this may be easy, >=20 >=20 > but if they are not [peering], they need to start configuring tunnels > between them to pass the routing information I don't think so. If ISPs use tunnels between them to pass the routing information this only helps for hosts in ISP1 and hosts in ISP2 to reach each other. But it doesn't help at all hosts in the Internet other than ISP12 to reach a host1; host1 is a host who moved from ISP1 to ISP2 and for which routes have been exchanged between ISPs over tunnels. CN on the Internet knew the route to host1 goes to ISP1, but it doesn't know it goes to ISP2. So packets from CN1 to host1 will go first to ISP1 and then tunnelled to ISP2. This is exactly the same as if simple Mobile IP were used (no global HAHA). > and the situation becomes rather messy and hard to manage imho (but i > agree that is an option... see more below) >=20 >=20 >> If the aviation industry _can_ say to ISP1 "please place a HA here" >> to ISP2 "please place a HA there" then it means that they can=20 >> already tell ISPs what to do. >>=20 >> And if they can do that, then they can also tell ISPs "ISPs please >> exchange routes about the MR when you sense it in your location". >> And if the ISPs exchange these routes then no need of Mobile IP. >>=20 >=20 > but are you assuming that the nemo has its own PI prefix or not? I don't think edge fringe networks can afford PI (provider-independent) prefixes. These seem to be for providers to further provide sub-prefixes. But it's a vague idea I have. This is exactly a question to deployers. > - if it has its own PI prefix, this is basically what has been=20 > presented in the slides by boeing, that is what they are currently=20 > doing. the problem with this is that each nemo adds an additional=20 > entry to the global routing table and they depend on BGP convergence > times (which won't be that good if they keep on adding routes i am=20 > afraid) I agree and I think they know it too. > - if the nemo do not have its own PI prefix, then it has a PA prefix > from one of the ISP, which means that the traffic from the internet > to the nemo is always routed through that particular ISP, which in=20 > the case that the CN and the nemo are topologically far away from=20 > that ISP, then the resulting performance may be pretty bad i guess I agree with you (as described above). Moreover, exchanging routes over VPNs is nothing short of Mobile IP - global HAHA wouldn't eliminate any HA from the path. > That is why, i think that a solution that support some form of RO=20 > would be good for this case (mixed with having multiple PA prefixes, > in order to reduce the need for using the RO solution) RO have their own issues. But I think RO together with MIP would be a _the_ global mobility solution (on paper at least). >> If however the ISPs are reluctant to place mutually trusted HAs, or >> if the ISPs are not straighlty connected together (connected=20 >> without BGP) then they won't be able to place HAs nor to exchange=20 >> routes. >>=20 >=20 > well, if they are not directly connected they may dtill be willing to > place HAs >=20 > In addition, it is also possible that the ground infrastructure has=20 > internet access, so that the HA can be located in there. This seems=20 > much easier that getting each local ISP to peer >=20 >> To say that if ISPs can place HAs at different places then they can >> also exchange routes; >=20 > I am not sure that is true, i mean selling HA services has much less > impact in the ISP netwrok infrastructure than exchenging routes.=20 > BEsides, it is possible for the ground infrastrucutre to have a local > HA, not located inside the ISP ... >=20 >> What is the risk of probing giving a false indication? It=20 >> nullifies the entire goal of global HAHA. I wouldn't risk the main >> requirement of a protocol on this kind of unreliable probe. If=20 >> the probing were just a hint, in addition to other info, then yes. >>=20 >=20 > well, another option would be to do use the BGP AS path length as a=20 > complementary metric. However, what i don't like about this is to=20 > require the MR to run BGP. However, it would be possible to=20 > workaround this requirement, by requiring the HAs to acquire BGP=20 > information and inform the MR about the distance between the local HA > and the COA where the MR is currently located. In this way, each HA > would inform the As distance between it and the MR and the MR can=20 > choose the prefix of the closest HA as the preferred prefix for the=20 > MNN to initiate new communications... do you think this would be more > reliable? Seems it sounds like two complementary hints then it's probably yes. It can probably be done. But it's probably too early to discuss solution space. Alex From nemo-bounces@ietf.org Fri Apr 07 13:16:16 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRuZA-0000bt-5L; Fri, 07 Apr 2006 13:16:12 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRuZ8-0000a8-1u for nemo@ietf.org; Fri, 07 Apr 2006 13:16:10 -0400 Received: from slb-smtpout-01.boeing.com ([130.76.64.48]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRuZ6-0004pd-M0 for nemo@ietf.org; Fri, 07 Apr 2006 13:16:10 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by slb-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id KAA16019; Fri, 7 Apr 2006 10:15:58 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k37HFwT17020; Fri, 7 Apr 2006 10:15:58 -0700 (PDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 7 Apr 2006 10:15:57 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Fri, 7 Apr 2006 10:15:57 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA6@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: <9a001afec8c802d2b089dc8b4d97a710@it.uc3m.es> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZZRD6nXI9yHU29R+KnFTITGxODGgBHOyVg From: "Dul, Andrew L" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 07 Apr 2006 17:15:57.0529 (UTC) FILETIME=[EF577C90:01C65A66] X-Spam-Score: 0.0 (/) X-Scan-Signature: 0bc60ec82efc80c84b8d02f4b0e4de22 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org =20 > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 =20 > Moreover, i don't understand how injecting a PI prefix would=20 > provide better performance than using a PA block from the=20 > global provider.... i mean, the global provider prefix is=20 > being injected through all the points where the global=20 > provide attaches to the internet, so, the CN will reach the=20 > global provider through the point that is closer to the CN=20 > and then the traffic will be routed through the internal=20 > global ISP network >=20 The path that your traffic takes is wholly dependant on how "well connected" your global provider is... If they peer at every major exchange point and have transit relationships with the major carriers it is unlikely to be very different. However, there are no providers today that "cover the whole globe", so in some cases your traffic may go "the long way" because two providers don't peer with each other. =20 I agree with you that in some cases PI traffic will follow the same path as a PA based prefix. However there are other cases where a PI prefix would be different; for example if you are multihomed or if there are providers who filter deaggregated PA address space. Andrew From nemo-bounces@ietf.org Fri Apr 07 13:45:18 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRv1I-0001kg-EO; Fri, 07 Apr 2006 13:45:16 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRv0e-0000cd-7B for nemo@ietf.org; Fri, 07 Apr 2006 13:44:36 -0400 Received: from slb-smtpout-01.boeing.com ([130.76.64.48]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRuvX-0005mJ-Uf for nemo@ietf.org; Fri, 07 Apr 2006 13:39:20 -0400 Received: from stl-av-01.boeing.com ([192.76.190.6]) by slb-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id KAA14959; Fri, 7 Apr 2006 10:39:17 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by stl-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k37HdGN19804; Fri, 7 Apr 2006 12:39:16 -0500 (CDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 7 Apr 2006 10:39:14 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Fri, 7 Apr 2006 10:39:14 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA8@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: <2877a0e9ca86d7651130007f1c1f8d67@it.uc3m.es> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZaBlRPk8prkzY7QIOZQaTeH6QQdQAWxn8A From: "Dul, Andrew L" To: "marcelo bagnulo braun" , "Alexandru Petrescu" X-OriginalArrivalTime: 07 Apr 2006 17:39:14.0259 (UTC) FILETIME=[2FDB9230:01C65A6A] X-Spam-Score: 0.0 (/) X-Scan-Signature: fb6060cb60c0cea16e3f7219e40a0a81 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > Sent: Thursday, April 06, 2006 10:43 PM > To: Alexandru Petrescu > Cc: ml-nemo WG; Pascal Thubert (pthubert); T.J. Kniveton;=20 > Davis, Terry L; Vijay Devarapalli > Subject: Re: About route aggregation in global HAHA (was=20 > Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter=20 > proposal) >=20 >=20 > El 06/04/2006, a las 18:02, Alexandru Petrescu escribi=F3: >=20 > >>> In general, I am not at all sure where Mobile IP is needed in a=20 > >>> closed (or small) system. > >> yes, but from other comments, i understood that they have=20 > different=20 > >> ISPs in different locations, so it is far from being a=20 > closed system. > > > > If they control all these ISPs it all becomes a "closed" system,=20 > > manageable by a single entity. >=20 > if all the ISPs involved are peering directly this may be=20 > easy, but if they are not, they need to start configuring=20 > tunnels between them to pass the routing information and the=20 > situation becomes rather messy and hard to manage imho (but i=20 > agree that is an option... see more below) Yes, this is one of my operational concerns with global-HAHA =20 > > If the aviation industry _can_ say to ISP1 "please place a=20 > HA here"=20 > > to ISP2 "please place a HA there" then it means that they=20 > can already=20 > > tell ISPs what to do. > > > > And if they can do that, then they can also tell ISPs "ISPs please=20 > > exchange routes about the MR when you sense it in your=20 > location". And=20 > > if the ISPs exchange these routes then no need of Mobile IP. > > >=20 > but are you assuming that the nemo has its own PI prefix or not? >=20 I personally think you have to assume that globally mobile networks have = PI address space. > - if it has its own PI prefix, this is basically what has=20 > been presented in the slides by boeing, that is what they are=20 > currently doing.=20 Just because you have a PI prefix doesn't mean that it has to be 1 = prefix in the global routing table per mobile-network. It is possible = to aggregate. > the problem with this is that each nemo adds=20 > an additional entry to the global routing table and they=20 > depend on BGP convergence times (which won't be that good if=20 > they keep on adding routes i am afraid) It is something we are keeping an eye on, but at this time there are = plenty of other people out there who are putting lots of useless updates = into the routing system... http://www3.ietf.org/proceedings/06mar/slides/grow-3.pdf Andrew From nemo-bounces@ietf.org Fri Apr 07 13:45:33 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRv1Z-0002QO-7Y; Fri, 07 Apr 2006 13:45:33 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRv10-0000pD-Nm for nemo@ietf.org; Fri, 07 Apr 2006 13:44:58 -0400 Received: from blv-smtpout-01.boeing.com ([130.76.32.69]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRumM-0005Kz-Ll for nemo@ietf.org; Fri, 07 Apr 2006 13:29:52 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by blv-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id KAA14673; Fri, 7 Apr 2006 10:29:46 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k37HTkT10742; Fri, 7 Apr 2006 10:29:46 -0700 (PDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 7 Apr 2006 10:29:41 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Fri, 7 Apr 2006 10:29:41 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA7@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZZSnj9KnKIp5k9QHCfsBJgmm6XYQBFsWTw From: "Dul, Andrew L" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 07 Apr 2006 17:29:41.0561 (UTC) FILETIME=[DA80CA90:01C65A68] X-Spam-Score: 0.0 (/) X-Scan-Signature: 0fa76816851382eb71b0a882ccdc29ac Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org =20 > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > Sent: Thursday, April 06, 2006 12:20 AM > To: Dul, Andrew L > Cc: ml-nemo WG; alexandru.petrescu@motorola.com > Subject: Re: About route aggregation in global HAHA (was=20 > Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter=20 > proposal) >=20 >=20 > El 05/04/2006, a las 19:23, Dul, Andrew L escribi=F3: > > > > I think this is a question for me... > > > > I think you can assume that the aircraft will have a=20 > connection to the=20 > > Internet both for passenger services as well as other crew services. >=20 > By crew services are you talking about air traffic and=20 > navigation control traffic? i.e. critical traffic? (in the=20 > sense that the integrity of the airplane depends on it) In this case I wasn't talking about critical traffic. I was thinking = about services for the crew that would allow them to communicate with = their operations center for things like re-ticketing...etc. Critical infrascturcture may likely use IPv6 too in the future, but will = operate on closed networks. They will likely still have mobility = issues, but those are different from the mobility issues to access the = Internet. >=20 > If this is the case, but wouldn't these be routed through=20 > different mechanisms? I mean, i guess you won't share the=20 > same channel between the web browsing of one passenger and=20 > the communication to the control of the airport. >=20 > Imho this distinction is quite important because they are=20 > likely to have very different QoS requirements Yes, critical vs. non-critical will have very different requirements. > > I believe Global-HAHA can be scalable if the individual BGP=20 > prefixes=20 > > are > > not injected into the global BGP table. >=20 > what do you call individual prefixes? > I mean, as i understand it, if you want to use global HAHA with=20 > multiple ISPs, you need to inject the distributed home prefix through=20 > the different ISPs, which imho precludes the routing system=20 > scalability, > so are you considering the distributed home prefix is not an=20 > individual=20 > prefix, or do you envision some way of running global HAHA wihtout=20 > injecting the dsitributed home prefix in bgp or other option that i=20 > missing? >=20 Let say I have lots of aircraft and I assign each of them a /48 out of a = /32. The individual prefixes would be the /48 for each aircraft. For a = simple case with 2 HA's each in a different ISP. If both ISPs announce = the /32 via BGP but carry the /48 addresses internally, then the = Internet doesn't see the /48s and reaches the aircraft via closest ISP = announcing the /32. I believe this is scalable because you are limiting = the /48s to inside of autonomous systems. This scheme requires the two = ISP to cooperate and exchange MN state information between themselves. = If ISP "A" receives a packet for a MN that is currently being served by = ISP "B" then it must forward the packet to ISP "B". Andrew From nemo-bounces@ietf.org Fri Apr 07 14:40:42 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRvsv-0003wa-13; Fri, 07 Apr 2006 14:40:41 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FRvsu-0003op-1E for nemo@ietf.org; Fri, 07 Apr 2006 14:40:40 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FRvss-0000u8-O6 for nemo@ietf.org; Fri, 07 Apr 2006 14:40:40 -0400 Received: from az33exr02.mot.com ([10.64.251.232]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k37Iu12Z009705; Fri, 7 Apr 2006 11:56:54 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by az33exr02.mot.com (8.13.1/8.13.0) with ESMTP id k37IqX13018656; Fri, 7 Apr 2006 13:52:34 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 85636865980; Fri, 7 Apr 2006 20:39:42 +0200 (CEST) Message-ID: <4436B1EE.4040709@motorola.com> Date: Fri, 07 Apr 2006 20:39:42 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "Dul, Andrew L" Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA7@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA7@XCH-NW-7V1.nw.nos.boeing.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 8abaac9e10c826e8252866cbe6766464 Cc: marcelo bagnulo braun , ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Thanks for the example! Dul, Andrew L wrote: > Let say I have lots of aircraft and I assign each of them a /48 out > of a /32. The individual prefixes would be the /48 for each > aircraft. For a simple case with 2 HA's each in a different ISP. If > both ISPs announce the /32 via BGP but carry the /48 addresses > internally, then the Internet doesn't see the /48s and reaches the > aircraft via closest ISP announcing the /32. Interesting. Put this way looks ok scalable, no BGP route injection for mobility. However, three things. One is that 48-32 one can acommodate only 65536 aircraft :-) The second is that CN may be closer to ISP1 than ISP2 (MR under ISP2) and ISP2 may be closer to CN than to ISP1. In which case, the fact that CN sends to ISP1 (because global HAHA) instead of ISP2 is not optimal at all. It should have sent to ISP2 instead. In general, one has some expectancies about how the two ISPs are distanced from each other, but it's more difficult to predict the placement of the CN with respect to ISPs. Third, if the two ISPs are tightly linked together somehow (because they advertise same /32) then it would be sufficient for them to run route updates between them to point exactly where the /48 is, thus avoiding core-BGP injections _and_ Mobile IP tunnelling. Alex From nemo-bounces@ietf.org Mon Apr 10 03:08:58 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FSqW2-0008JA-6a; Mon, 10 Apr 2006 03:08:50 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FSqW0-00089k-Cs for nemo@ietf.org; Mon, 10 Apr 2006 03:08:48 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FSqVm-0002xM-Ai for nemo@ietf.org; Mon, 10 Apr 2006 03:08:34 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FSqVh-0007vn-OU for nemo@ietf.org; Mon, 10 Apr 2006 03:08:30 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id DDFCC212C5D; Mon, 10 Apr 2006 10:07:33 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id A6CE6212C4E; Mon, 10 Apr 2006 10:07:33 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 67FBEBDC40; Mon, 10 Apr 2006 10:07:33 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 341C4BDC38; Mon, 10 Apr 2006 10:07:33 +0300 (EEST) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA7@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA7@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <0e350184fe800f79789245e8d752dab3@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Mon, 10 Apr 2006 10:07:36 +0300 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: -2.6 (--) X-Scan-Signature: 92df29fa99cf13e554b84c8374345c17 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 07/04/2006, a las 20:29, Dul, Andrew L escribi=F3: > >>> I believe Global-HAHA can be scalable if the individual BGP >> prefixes >>> are >>> not injected into the global BGP table. >> >> what do you call individual prefixes? >> I mean, as i understand it, if you want to use global HAHA with >> multiple ISPs, you need to inject the distributed home prefix through >> the different ISPs, which imho precludes the routing system >> scalability, >> so are you considering the distributed home prefix is not an >> individual >> prefix, or do you envision some way of running global HAHA wihtout >> injecting the dsitributed home prefix in bgp or other option that i >> missing? >> > > Let say I have lots of aircraft and I assign each of them a /48 out of=20= > a /32. The individual prefixes would be the /48 for each aircraft. =20= > For a simple case with 2 HA's each in a different ISP. If both ISPs=20= > announce the /32 via BGP but carry the /48 addresses internally, then=20= > the Internet doesn't see the /48s and reaches the aircraft via closest=20= > ISP announcing the /32. I believe this is scalable because you are=20 > limiting the /48s to inside of autonomous systems. This scheme=20 > requires the two ISP to cooperate and exchange MN state information=20 > between themselves. If ISP "A" receives a packet for a MN that is=20 > currently being served by ISP "B" then it must forward the packet to=20= > ISP "B". > still you need one bgp table slot for this /32 which is not a LIR i.e.=20= not PA addressing used here, which is a consumption of the global=20 resources. In other words, the number of routes added by this approach=20= grows linearly with the number of distributed home networks. This=20 definitely better than a solution that adds one bgp entry per moving=20 network, but still consumes global bgp slots Now, this may be ok, but, imho we need to understand whether there are=20= acceptable solutions that do not consume such resource, like for=20 instance other types of RO mechanisms. Clearly and end to end RO=20 mechanism do not consume any bgp slot since it is end to end and uses=20 PA addressing. Now maybe the RO solutions do not fit the aircraft=20 scenario requirements, but i still need to be convinced that this is=20 the case. Hence, i would propose (again) that we first identify which are the=20 requirements for this scenario and then evaluate different solutions. bottom line is why is the charter only considering the global HAHA tool=20= as a possible RO solution and does not takes into account other RO=20 solutions? Regards, marcelo > Andrew > From nemo-bounces@ietf.org Mon Apr 10 03:30:10 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FSqqe-0007kt-V4; Mon, 10 Apr 2006 03:30:08 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FSqoV-0007JT-Ht for nemo@ietf.org; Mon, 10 Apr 2006 03:27:55 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FSqZW-00030m-05 for nemo@ietf.org; Mon, 10 Apr 2006 03:12:27 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 4EDCC212C5D; Mon, 10 Apr 2006 10:12:25 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id E0F34212C4E; Mon, 10 Apr 2006 10:12:24 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id AB372BDC40; Mon, 10 Apr 2006 10:12:24 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 77C4DBDC38; Mon, 10 Apr 2006 10:12:24 +0300 (EEST) In-Reply-To: <4436B1EE.4040709@motorola.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA7@XCH-NW-7V1.nw.nos.boeing.com> <4436B1EE.4040709@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Mon, 10 Apr 2006 10:12:27 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 5a9a1bd6c2d06a21d748b7d0070ddcb8 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 07/04/2006, a las 21:39, Alexandru Petrescu escribi=F3: > Thanks for the example! > > Dul, Andrew L wrote: >> Let say I have lots of aircraft and I assign each of them a /48 out >> of a /32. The individual prefixes would be the /48 for each >> aircraft. For a simple case with 2 HA's each in a different ISP. If >> both ISPs announce the /32 via BGP but carry the /48 addresses >> internally, then the Internet doesn't see the /48s and reaches the >> aircraft via closest ISP announcing the /32. > > Interesting. Put this way looks ok scalable, no BGP route injection=20= > for mobility. > well, we still need to inject one bgp route per distributed home=20 network, right? > ... > > Third, if the two ISPs are tightly linked together somehow (because=20 > they advertise same /32) then it would be sufficient for them to run=20= > route updates between them to point exactly where the /48 is, thus=20 > avoiding core-BGP injections _and_ Mobile IP tunnelling. > good point, basically if the aircraft already knows what ISPs it will be visiting=20 and it has some commercial arrangement with it, it is likely that the=20 ISP can accept routes from the aircraft when it is visiting, so no=20 mobile is required whatsoever I guess that the bottom line is that the trust/commercial model in the=20= case of an aircraft visiting an ISP is different from the one assumed=20 in mobile IP. I mean, i guess we can assume a much more higher level of=20= trust between the aircraft and the visited ISP, so that route injection=20= is an option (we can see that from boeing presentation, i guess) regards, marcelo > Alex > > From nemo-bounces@ietf.org Mon Apr 10 04:06:39 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FSrPx-0007Fi-Mi; Mon, 10 Apr 2006 04:06:37 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FSrPt-0007EA-UQ for nemo@ietf.org; Mon, 10 Apr 2006 04:06:33 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FSqMU-0002iL-JT for nemo@ietf.org; Mon, 10 Apr 2006 02:59:01 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id A430A212C5D; Mon, 10 Apr 2006 09:58:43 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 3D62A212C4E; Mon, 10 Apr 2006 09:58:43 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id F313ABDC40; Mon, 10 Apr 2006 09:58:42 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id BBF31BDC38; Mon, 10 Apr 2006 09:58:42 +0300 (EEST) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA6@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA6@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <9ca3659f5914f73d94c94718e67308c0@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Mon, 10 Apr 2006 09:58:45 +0300 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 9ed51c9d1356100bce94f1ae4ec616a9 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 07/04/2006, a las 20:15, Dul, Andrew L escribi=F3: > > >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] > >> Moreover, i don't understand how injecting a PI prefix would >> provide better performance than using a PA block from the >> global provider.... i mean, the global provider prefix is >> being injected through all the points where the global >> provide attaches to the internet, so, the CN will reach the >> global provider through the point that is closer to the CN >> and then the traffic will be routed through the internal >> global ISP network >> > > The path that your traffic takes is wholly dependant on how "well > connected" your global provider is... If they peer at every major > exchange point and have transit relationships with the major carriers=20= > it > is unlikely to be very different. However, there are no providers=20 > today > that "cover the whole globe", so in some cases your traffic may go = "the > long way" because two providers don't peer with each other. > > I agree with you that in some cases PI traffic will follow the same=20 > path > as a PA based prefix. However there are other cases where a PI prefix > would be different; for example if you are multihomed or if there are > providers who filter deaggregated PA address space. > agree regards, marcelo > Andrew > From naotan@dion.ne.jp Tue Apr 11 02:17:47 2006 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTCCB-0004jB-GF for NEMO-ARCHIVE@LISTS.IETF.ORG; Tue, 11 Apr 2006 02:17:47 -0400 Received: from [221.207.212.163] (helo=ocn.ne.jp) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTCC5-0002NV-DT for NEMO-ARCHIVE@LISTS.IETF.ORG; Tue, 11 Apr 2006 02:17:47 -0400 Received: from fblsbzqld9 (unknown [46.155.54.11]) by smtp3 (Coremail) with SMTP id YeAO2IBQQsxOqWnV.1 for ; Fri, 04 Apr 2003 00:30:01 +0800 (CST) X-Originating-IP: [46.155.54.11] Subject: =?iso-2022-jp?B?GyRCTlc7fjx9Rn4kTiQqQ04kaSQ7GyhC?= From: =?gb2312?B?c3V6dQ==?= To: X-Mailer: Microsoft Outlook Express 6.00.2800.1478 MIME-Version: 1.0 Content-Type: text/plain; Content-Transfer-Encoding: base64 X-Priority: 3 X-MSMail-Priority: Normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 X-Spam-Score: 3.2 (+++) X-Scan-Signature: 7a6398bf8aaeabc7a7bb696b6b0a2aad R0VUGyRCSn1LISRPIVo9UEQlJVslOSVIIVsbKEINChskQjpHRGMlaSUkJXMbKEIxGyRCO340ViM1 S3whQSRHJDkhIxsoQg0KaHR0cDovL3d3dy5tZWd1cmlhaS1tYXgubmV0Lz8yMDIwDQoNCg0KGyRC PXdALU1NJCwbKEJQRhskQj5wSnMkciRiJEgkSyEiNSQkS0Z+JEMkPyVbJTklSCQsJCQkXiQ3JD8k aSEiRD5FRSEmRD4lYRsoQg0KGyRCTyJNbUp9SyEkTz13QC1NTSRLJGgkajBbJEokaiReJDkkLE8i TW0kLEZ+JGo8IUJoISI+cjdvISY2YjNbJHIbKEINChskQiQqN2gkYTI8JDUkJCEjGyhCDQobJEIk PSROOGUiLSEhIi0hISItGyhCDQobJEIiKCQqNmEkLyROPXdALUJwJGQlWyVGJWtFeSRLPVBEJSQ3 JEYkJCQ/JEAkLUJnP00kTjRYNzgkSiQqSVU5ZyQkJHIbKEINChskQiQ3JEYkJCQ/JEAkLSEiPi44 LyQkJHJEPkBcPHU8aCRqJF4kOSEjGyhCDQoNChskQiIoOHJETEhxJEokSSRPJDkkWSRGPXdALTJx MHckLElpQzQkNyReJDkhIxsoQg0KGyRCIihKcz03JE9EPkBcISI9d0AtJCskaTx1PGgkQyRGMjwk NSQkISMbKEINCg0KGyRCIighWiEhNDBBNEw1TkEhISFbJDkkMCRLGyhCR0VUGyRCISohKhsoQg0K aHR0cDovL3d3dy5tZWd1cmlhaS1tYXgubmV0Lz8yMDIwDQoNCg0KDQoNCg0KDQoNCg0KDQobJEIl YSE8JWtJVE1XJEpKfSRPJDMkQSRpIi0bKEINCnBwX2luZm80NTg2QHlhaG9vLmNh From nemo-bounces@ietf.org Tue Apr 11 04:10:41 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTDxL-0005fq-8J; Tue, 11 Apr 2006 04:10:35 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTDxJ-0005fg-Jz for nemo@ietf.org; Tue, 11 Apr 2006 04:10:33 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTDxI-0005yp-94 for nemo@ietf.org; Tue, 11 Apr 2006 04:10:33 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 11 Apr 2006 10:10:32 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3B8AU3A005634 for ; Tue, 11 Apr 2006 10:10:31 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Tue, 11 Apr 2006 10:10:31 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Tue, 11 Apr 2006 10:10:25 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA Thread-Index: AcZZSnj9KnKIp5k9QHCfsBJgmm6XYQBFsWTwALZ0GnA= From: "Pascal Thubert \(pthubert\)" To: X-OriginalArrivalTime: 11 Apr 2006 08:10:31.0181 (UTC) FILETIME=[669213D0:01C65D3F] X-Spam-Score: 0.0 (/) X-Scan-Signature: cab78e1e39c4b328567edb48482b6a69 Subject: [nemo] RE: About route aggregation in global HAHA X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi: We have seen a lot of great arguments in this list in the recent days about whether and how global HAHA could / would be deployed. I need to revisit it all and add things to the draft. Nevertheless, here's what I gathered: - Some assumptions lead to aggregation trouble at BGP level. Those assumptions are based on the way the prefixes are distributed to the Service providers today. Mobile prefixes might not necessarily be distributed in that same fashion, though. They could be shared. - We need to work out the conditions to make the protocol deployable. If this involves revisiting some assumptions about IPv6 deployment, then that part goes beyond the scope of NEMO and has to do with either IPv6 or IPv6 Ops.=20 - The NEMO part for this discussion is enabling Route optimization within the infrastructure. There are a number of use cases to sustain the need. IP mobile telephony (4G?) is an example, where Service Providers do not necessarily wish to have the phone perform the RO themselves. - Global HAHA introduce proxy HAs which enable this route optimization with a trust model in place. I'd say that defining how that works is enough work for NEMO; whereas proposing the deployment model should be a request from us to other WGs. Let me give you an example. Say that 3-4 ISPs from different continents form a mobile ISP consortium.=20 That consortium obtains a short prefix from one of the NICs, call it the mobile prefix. Internally, the consortium shares that addressing space between members, leaving big chunks free for new comers. The rule is that only the big aggregation is redistributed over the Internet BGP, from most POP of members ISPs. But internally, the consortium of service providers runs a second BGP which deals with the subnetting of mobile prefix, over a mesh of tunnels and private links.=20 A similar mesh has formed already in the GPRS core network. On top of that model, Global HAHA introduces a Route Optimization scheme that does not exist between SGSNs, and integrates voice and data. Pascal From nemo-bounces@ietf.org Tue Apr 11 06:50:44 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTGSH-0008Tl-6D; Tue, 11 Apr 2006 06:50:41 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTGSF-0008Tg-UP for nemo@ietf.org; Tue, 11 Apr 2006 06:50:39 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTGSE-0003wq-Eq for nemo@ietf.org; Tue, 11 Apr 2006 06:50:39 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 25385212C63; Tue, 11 Apr 2006 13:50:37 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id E6447212C5F; Tue, 11 Apr 2006 13:50:36 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id B09D7BDC40; Tue, 11 Apr 2006 13:50:36 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 7D791BDC38; Tue, 11 Apr 2006 13:50:36 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> Content-Transfer-Encoding: 7bit From: marcelo bagnulo braun Subject: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 11 Apr 2006 13:50:35 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: d0bdc596f8dd1c226c458f0b4df27a88 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Pascal, I have two points that i would like to address as reply to your note: First, a broader topic about RO and global HAHA as a route optimization mechanism. you stated that: > > - The NEMO part for this discussion is enabling Route optimization > within the infrastructure. There are a number of use cases to sustain > the need. IP mobile telephony (4G?) is an example, where Service > Providers do not necessarily wish to have the phone perform the RO > themselves. > > - Global HAHA introduce proxy HAs which enable this route optimization > with a trust model in place. I'd say that defining how that works is > enough work for NEMO; whereas proposing the deployment model should be > a > request from us to other WGs. > IMHO, we should first decide on which forms of RO for nemo are we going to work on, and then move on to specific solutions. I mean, as specified in the RO taxonomy draft there are several and very different approaches to RO. I think that in order to do the charter for future work, we need first to decide which one of the RO approaches are we going to work on. However, the proposed charter only has Global HAHA as a RO solution for nemo, so it seems to imply that this is the RO that we have selected. I would propose to open the general discussion about which RO approaches are we going to work on first rather than focussing in if we are going to work on global haha or not... Second, a more focussed issue with respect to routing system scalability and business model you stated that: > Let me give you an example. Say that 3-4 ISPs from different continents > form a mobile ISP consortium. > > That consortium obtains a short prefix from one of the NICs, call it > the > mobile prefix. Internally, the consortium shares that addressing space > between members, leaving big chunks free for new comers. The rule is > that only the big aggregation is redistributed over the Internet BGP, > from most POP of members ISPs. But internally, the consortium of > service > providers runs a second BGP which deals with the subnetting of mobile > prefix, over a mesh of tunnels and private links. but this seems to be complettelly different that what other people have been claiming here on the list. Past comments have strongly argued that global mobile network need PI addressing and global HAHA seemed a mechanisms to provide access to this PI prefix that the airline company gets. This is completelly different than the mobile ISP model that you are suggesting. What worries me is that i don't see any mobile ISP in the list claiming that they need global HAHA to deploy their business, but there are end users (air companies) that claim that want PI addressing for deploying this. Regards, marcelo From nemo-bounces@ietf.org Tue Apr 11 07:54:24 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTHRr-0005yD-IR; Tue, 11 Apr 2006 07:54:19 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTHRq-0005y8-2f for nemo@ietf.org; Tue, 11 Apr 2006 07:54:18 -0400 Received: from omgo.iij.ad.jp ([202.232.30.157]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTHRp-00069i-IA for nemo@ietf.org; Tue, 11 Apr 2006 07:54:18 -0400 Received: OTM-MO id k3BBsDqM001444; Tue, 11 Apr 2006 20:54:14 +0900 (JST) Received: OTM-MIX(otm-mix01) id k3BBsDUZ078011; Tue, 11 Apr 2006 20:54:13 +0900 (JST) Received: from localhost (keiichi00.osaka.iij.ad.jp [192.168.64.45]) by jc-smtp.iij.ad.jp (JC-SMTP/jc-smtp) id k3BBsCnE019589; Tue, 11 Apr 2006 20:54:12 +0900 (JST) Date: Tue, 11 Apr 2006 20:56:18 +0900 (JST) Message-Id: <20060411.205618.66928349.keiichi@iijlab.net> To: marcelo@it.uc3m.es Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA From: Keiichi SHIMA In-Reply-To: <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> X-Mailer: Mew version 4.1.50 on Emacs 21.3 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Spam-Score: 0.0 (/) X-Scan-Signature: 3e15cc4fdc61d7bce84032741d11c8e5 Cc: nemo@ietf.org, pthubert@cisco.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hello, From: marcelo bagnulo braun Subject: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 11 Apr 2006 13:50:35 +0300 > > Let me give you an example. Say that 3-4 ISPs from different continents > > form a mobile ISP consortium. > > > > That consortium obtains a short prefix from one of the NICs, call it > > the > > mobile prefix. Internally, the consortium shares that addressing space > > between members, leaving big chunks free for new comers. The rule is > > that only the big aggregation is redistributed over the Internet BGP, > > from most POP of members ISPs. But internally, the consortium of > > service > > providers runs a second BGP which deals with the subnetting of mobile > > prefix, over a mesh of tunnels and private links. > > but this seems to be complettelly different that what other people have > been claiming here on the list. > Past comments have strongly argued that global mobile network need PI > addressing and global HAHA seemed a mechanisms to provide access to > this PI prefix that the airline company gets. I think, in the proposal from Pascal, "a short prefix from one of the NICs" is the equivalent of the PI address. I believe what they need is a network which has a fixed prefix and provides low-latency communication to the rest of the Internet, not PI addresses. If the global HA-HA provides the same features, then there will be no problem. (Although I still don't understand the global hA-HA completely) I'm afraid many people don't understand what the global HA-HA can provide (including me). > This is completelly different than the mobile ISP model that you are > suggesting. > > What worries me is that i don't see any mobile ISP in the list claiming > that they need global HAHA to deploy their business, but there are end > users (air companies) that claim that want PI addressing for deploying > this. That is because many people think only PI addresses can provide what airline companies need. If we understand the global HA-HA can provide the same things, maybe we can compare them, I think. Regards, --- Keiichi SHIMA IIJ Research Laboratory WIDE Project From nemo-bounces@ietf.org Tue Apr 11 08:06:47 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTHdt-0002ei-6u; Tue, 11 Apr 2006 08:06:45 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTHds-0002bX-79 for nemo@ietf.org; Tue, 11 Apr 2006 08:06:44 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTHdr-0006dk-NW for nemo@ietf.org; Tue, 11 Apr 2006 08:06:44 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id AE053212C63; Tue, 11 Apr 2006 15:06:42 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 4C719212C5F; Tue, 11 Apr 2006 15:06:42 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id D93E3BDC40; Tue, 11 Apr 2006 15:06:41 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id A62CBBDC38; Tue, 11 Apr 2006 15:06:41 +0300 (EEST) In-Reply-To: <20060411.205618.66928349.keiichi@iijlab.net> References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <20060411.205618.66928349.keiichi@iijlab.net> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 11 Apr 2006 15:06:40 +0300 To: Keiichi SHIMA X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: f66b12316365a3fe519e75911daf28a8 Cc: nemo@ietf.org, pthubert@cisco.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 11/04/2006, a las 14:56, Keiichi SHIMA escribi=F3: > Hello, > > From: marcelo bagnulo braun > Subject: About the proposed charter and global HAHA (was Re: [nemo]=20 > RE: About route aggregation in global HAHA > Date: Tue, 11 Apr 2006 13:50:35 +0300 > >>> Let me give you an example. Say that 3-4 ISPs from different=20 >>> continents >>> form a mobile ISP consortium. >>> >>> That consortium obtains a short prefix from one of the NICs, call it >>> the >>> mobile prefix. Internally, the consortium shares that addressing=20 >>> space >>> between members, leaving big chunks free for new comers. The rule is >>> that only the big aggregation is redistributed over the Internet = BGP, >>> from most POP of members ISPs. But internally, the consortium of >>> service >>> providers runs a second BGP which deals with the subnetting of = mobile >>> prefix, over a mesh of tunnels and private links. >> >> but this seems to be complettelly different that what other people=20 >> have >> been claiming here on the list. >> Past comments have strongly argued that global mobile network need PI >> addressing and global HAHA seemed a mechanisms to provide access to >> this PI prefix that the airline company gets. > > I think, in the proposal from Pascal, "a short prefix from one of the=20= > NICs" > is the equivalent of the PI address. > not really, because PI means that is not related to the provider. So=20 the number of PI prefixes that you need in the global table grows=20 linearly with the number of end sites that are needing it However, in the case of the PA or the mobile ISP that Pascal is=20 suggesting, many end sites can use the services of a single mobile ISP=20= and then the number of entries required in the global BGP table grows=20 with the number of mobile ISP (in other words, it scales much better=20 than the case of PI prefixes) So, a mobile ISP could serve many airlines and still be announcing a=20 single prefix in the DFZ, while if the PI approach is followed, then=20 the number of prefixes announced in the dfz is proportional to the=20 number of airlines. regards, marcelo > I believe what they need is a network which has a fixed prefix and > provides low-latency communication to the rest of the Internet, not PI > addresses. If the global HA-HA provides the same features, then there > will be no problem. (Although I still don't understand the global > hA-HA completely) > > I'm afraid many people don't understand what the global HA-HA can > provide (including me). > >> This is completelly different than the mobile ISP model that you are >> suggesting. >> >> What worries me is that i don't see any mobile ISP in the list=20 >> claiming >> that they need global HAHA to deploy their business, but there are = end >> users (air companies) that claim that want PI addressing for = deploying >> this. > > That is because many people think only PI addresses can provide what > airline companies need. If we understand the global HA-HA can provide > the same things, maybe we can compare them, I think. > > Regards, > --- > Keiichi SHIMA > IIJ Research Laboratory > WIDE Project > From nemo-bounces@ietf.org Wed Apr 12 09:19:08 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTfFR-000186-PQ; Wed, 12 Apr 2006 09:19:05 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTfFQ-00017t-Sy for nemo@ietf.org; Wed, 12 Apr 2006 09:19:04 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTfFN-0005FS-PZ for nemo@ietf.org; Wed, 12 Apr 2006 09:19:04 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 12 Apr 2006 15:19:01 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3CDIxUE002443; Wed, 12 Apr 2006 15:18:59 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Wed, 12 Apr 2006 15:18:59 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Wed, 12 Apr 2006 15:18:50 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZdVdlJDKeompzsRmm0Evo0FmawuAAugMZw From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 12 Apr 2006 13:18:59.0172 (UTC) FILETIME=[A89B2240:01C65E33] X-Spam-Score: 0.0 (/) X-Scan-Signature: 2ed806e2f53ff1a061ad4f97e00345ac Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Marcelo: >you stated that: > >> >> - The NEMO part for this discussion is enabling Route optimization >> within the infrastructure. There are a number of use cases to sustain >> the need. IP mobile telephony (4G?) is an example, where Service >> Providers do not necessarily wish to have the phone perform the RO >> themselves. >> >> - Global HAHA introduce proxy HAs which enable this route optimization >> with a trust model in place. I'd say that defining how that works is >> enough work for NEMO; whereas proposing the deployment model should be >> a >> request from us to other WGs. >> > >IMHO, we should first decide on which forms of RO for nemo are we going >to work on, and then move on to specific solutions. > >I mean, as specified in the RO taxonomy draft there are several and >very different approaches to RO. I think that in order to do the >charter for future work, we need first to decide which one of the RO >approaches are we going to work on. > >However, the proposed charter only has Global HAHA as a RO solution for >nemo, so it seems to imply that this is the RO that we have selected. >I would propose to open the general discussion about which RO >approaches are we going to work on first rather than focussing in if we >are going to work on global haha or not... > [Pascal]=20 My understanding is that the new charter covers a transition time. Global HAHA, which covers the RO within the infrastructure, is well advanced; it has been supported up to the plenary, so there's a common thinking that we can go ahead and move it forward.=20 Now, it is also my expectation is that NEMO will recharter again in a year or so to solve more RO problems. If you have in mind a MR to MR RO such as covered by MIRON, then I believe that if a group accepts the document, it should definitely be discussed within NEMO at that point. Finally, some RO schemes might not belong to NEMO as a WG. For instance, there's a question whether MANET for NEMO (MANEMO) should be done within NEMO or as a separate WG, because of the specific skills and interests involved. > >Second, a more focussed issue with respect to routing system >scalability and business model > >you stated that: > >> Let me give you an example. Say that 3-4 ISPs from different continents >> form a mobile ISP consortium. >> >> That consortium obtains a short prefix from one of the NICs, call it >> the >> mobile prefix. Internally, the consortium shares that addressing space >> between members, leaving big chunks free for new comers. The rule is >> that only the big aggregation is redistributed over the Internet BGP, >> from most POP of members ISPs. But internally, the consortium of >> service >> providers runs a second BGP which deals with the subnetting of mobile >> prefix, over a mesh of tunnels and private links. > >but this seems to be complettelly different that what other people have >been claiming here on the list. >Past comments have strongly argued that global mobile network need PI >addressing and global HAHA seemed a mechanisms to provide access to >this PI prefix that the airline company gets. > >This is completelly different than the mobile ISP model that you are >suggesting. I was stretching the picture, but the ideas were their. The mobile prefix I proposed might not clearly fit in PA, and is not PI in the geo sense either.=20 If you consider that the consortium becomes a virtual provider, then it is a form of PA. But the prefix is not subnetted, as far as the Internet BGP can see, so the aggregate looses some sense.=20 You might consider that the consortium is provider independent, but the prefix is surely not geo based. In fact, it is ubiquitous (close to you wherever you are). This is why I proposed to go back to IPv6 or IPv6Ops to discuss what that prefix is really and what the rules are for the RIR to allocate it. In any fashion, if there are few large consortiums, then this is surely not a burden for the DFZ. In fact, there would be another DFZ within the consortium to handle that prefix as a sub-internet. GTP proves that providers are willing to congregate when they find that such is their interest. > >What worries me is that i don't see any mobile ISP in the list claiming >that they need global HAHA to deploy their business, but there are end >users (air companies) that claim that want PI addressing for deploying >this. > If I understand well Tony's draft, NEMO is not the right forum for Provider Independent (PI) allocation policy. And that work is still ongoing, somewhere else... As for ISP side, I'm copying Masafumi-san who co authored the NEMO RO PS. Note that the work on global HAHA started with voice ISPs and 4G design. There was a strong requirement to make the phones unaware of RO. Some of the reasons: - privacy: one should not know where the mobile correspondent. In other words, a phone should only see the other phone's Home Address, not its CoA. - phone: RO might cause interoperability problems between generations of phones. You want the phone software to be simple - limited resources - and conservative -protocol ha to remain stable for years - so changes must occur centrally in the ISP side. - radio: client based RO consumes additional resources on the first hop whereas it saves resources globally. When the first hop is a radio and the rest of the way is high speed wire, you might prefer the infrastructure based RO to maintain the benefits of RO while preserving radio resources. - trust model: The global HAHA trust model is similar to classical (phone) roaming. It is something that ISPs can understand and deploy. In particular, it is possible to push policies to the proxy, acting like a LMM anchor.=20 I hope this helps... Pascal From nemo-bounces@ietf.org Wed Apr 12 09:58:39 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTfri-0001xV-TU; Wed, 12 Apr 2006 09:58:38 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTfrh-0001xQ-S9 for nemo@ietf.org; Wed, 12 Apr 2006 09:58:37 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTfrh-0006WL-3Z for nemo@ietf.org; Wed, 12 Apr 2006 09:58:37 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 3CB9A212C63; Wed, 12 Apr 2006 16:58:35 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 048C1212C5F; Wed, 12 Apr 2006 16:58:35 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id C863FBDC40; Wed, 12 Apr 2006 16:58:33 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 9134EBDC38; Wed, 12 Apr 2006 16:58:33 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <5ad0d29dbb9fcbb8dc2b37b095fb3b82@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Wed, 12 Apr 2006 16:58:33 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: cbb41f2dbf0f142369614756642005e3 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Pascal, El 12/04/2006, a las 16:18, Pascal Thubert (pthubert) escribi=F3: > [Pascal] > > My understanding is that the new charter covers a transition time. not sure what you mean, and i am not sure why we need a transition nor=20= a transition time... in any case, i guess we need to discuss in the wg=20= whether this is the right approach... > Global HAHA, which covers the RO within the infrastructure, is well > advanced; what do you mean by "well advanced"? there are several nemo RO that has=20= proposed ID and has implementations and so on... moreover, there are RO solution that don't even require new protocols,=20= so i guess those are really advanced, in the sense that they are=20 possible with current specs... > it has been supported up to the plenary, what plenary? ietf plenary? since when IETf plenary supports solutions?=20= I guess this is the wg to decide, not the plenary... > so there's a common > thinking that we can go ahead and move it forward. > not sure what common thinking you are talking about, but i would really=20= amazed if there is a common thinking in the community to support a PI=20 based solution.... > Now, it is also my expectation is that NEMO will recharter again in a > year or so to solve more RO problems. If you have in mind a MR to MR = RO > such as covered by MIRON, then I believe that if a group accepts the > document, it should definitely be discussed within NEMO at that point. > not sure why do you think that we need two different stages... is this=20= has been discussed in this wg? wouldn't this be part of the=20 rechartering discussion that is having place right now in this wg? > Finally, some RO schemes might not belong to NEMO as a WG. For=20 > instance, > there's a question whether MANET for NEMO (MANEMO) should be done=20 > within > NEMO or as a separate WG, because of the specific skills and interests > involved. > agree > > >> >> Second, a more focussed issue with respect to routing system >> scalability and business model >> >> you stated that: >> >>> Let me give you an example. Say that 3-4 ISPs from different > continents >>> form a mobile ISP consortium. >>> >>> That consortium obtains a short prefix from one of the NICs, call it >>> the >>> mobile prefix. Internally, the consortium shares that addressing > space >>> between members, leaving big chunks free for new comers. The rule is >>> that only the big aggregation is redistributed over the Internet = BGP, >>> from most POP of members ISPs. But internally, the consortium of >>> service >>> providers runs a second BGP which deals with the subnetting of = mobile >>> prefix, over a mesh of tunnels and private links. >> >> but this seems to be complettelly different that what other people=20 >> have >> been claiming here on the list. >> Past comments have strongly argued that global mobile network need PI >> addressing and global HAHA seemed a mechanisms to provide access to >> this PI prefix that the airline company gets. >> >> This is completelly different than the mobile ISP model that you are >> suggesting. > > I was stretching the picture, but the ideas were their. The mobile > prefix I proposed might not clearly fit in PA, and is not PI in the = geo > sense either. > not sure what do you mean by "PI in the geo sense" current PI allocations are unrelated to geo-PI schemes, they are simply=20= independent of any provider but not related to any geography > If you consider that the consortium becomes a virtual provider, then = it > is a form of PA. But the prefix is not subnetted, as far as the=20 > Internet > BGP can see, so the aggregate looses some sense. > > You might consider that the consortium is provider independent, but = the > prefix is surely not geo based. In fact, it is ubiquitous (close to = you > wherever you are). > This is why I proposed to go back to IPv6 or IPv6Ops > to discuss what that prefix is really and what the rules are for the=20= > RIR > to allocate it. this has no relationship at all with geo-PI it is about allocating prefixes directly to end users of allocating=20 them to providers so they can aggregate all the prefixes of their=20 customers in a single bgp announcement. In this case, it is a single end user that is requiring to burn a bgp=20 slot, my question is why would them be allowed to do so? what are their=20= special requirements that make other RO solutions not suitable for=20 them, but yet suitable for the rest of the people? I am not saying that=20= there is no such requirements, but if they are, i think they should be=20= presented to the wg and discussed in order to figure out that there is=20= no alternative solution, that does not imposes additional stress to the=20= global routing tables, that would fulfill the requirements. That is=20 why, imho the first step is to flesh out these special requirements...=20= I mean it is really hard to determine if this si the only acceptable=20 solution if we don't know what the requirements are... > > In any fashion, if there are few large consortiums, then this is = surely > not a burden for the DFZ. yes, right and there are really few people called Marcelo Bagnulo and=20 this is not a valid reason for allocating a PI prefix to those who are=20= called Marcelo Bagnulo. The reason to define a solution that relies in the injection of PI=20 prefixes in the global routing table is that there is no other solution=20= that will be suitable for the requirements imposed. So again the first=20= thing would be to agree in what the requirements are > In fact, there would be another DFZ within the consortium to handle=20 > that > prefix as a sub-internet. i may be fine with they bloating their internal routing tables, but i=20 don't feel the same in the case of the global DFZ > GTP proves that providers are willing to congregate when they find = that > such is their interest. > > >> >> What worries me is that i don't see any mobile ISP in the list=20 >> claiming >> that they need global HAHA to deploy their business, but there are = end >> users (air companies) that claim that want PI addressing for = deploying >> this. >> > If I understand well Tony's draft, NEMO is not the right forum for > Provider Independent (PI) allocation policy. And that work is still > ongoing, somewhere else... > i am not claiming for PI and even less for geo PI actually i am claiming for exactly the oposite, to avoid solutions that=20= require PI addressing to work > As for ISP side, I'm copying Masafumi-san who co authored the NEMO RO > PS. > > Note that the work on global HAHA started with voice ISPs and 4G=20 > design. > There was a strong requirement to make the phones unaware of RO. Some=20= > of > the reasons: > this is a good starting point for the requirements discussion > - privacy: one should not know where the mobile correspondent. In = other > words, a phone should only see the other phone's Home Address, not its > CoA. > i think privacy considerations are an interesting issue and it deserves=20= careful analysis, in particular, we need to understand what information=20= is being protected and what information the user would like to hide=20 (wee the work done in the ALIEN bof)... for instance in the case of=20 global HAHA, the mobile node would be using this special prefix that is=20= reserved to the airline, so it would be disclosing the fact that it is=20= flying somewhere. If PA addresses from the provider are used, it would only be disclosing=20= that it is using a given provider (that depending on the area of=20 coverage of the ISP may be bigger or smaller) so, exactly what is what is what we are trying to hide? > - phone: RO might cause interoperability problems between generations=20= > of > phones. You want the phone software to be simple - limited resources - > and conservative -protocol ha to remain stable for years - so changes > must occur centrally in the ISP side. > but this is true for almost any widespread deployed device... i mean=20 this is true for hosts also... that is why protocols need to be defined=20= in a way that they are extensible and allow future features in a=20 backward compatible way > - radio: client based RO consumes additional resources on the first = hop > whereas it saves resources globally. exactly, that is why the scale so well, because no central resourses=20 are required > When the first hop is a radio and > the rest of the way is high speed wire, you might prefer the > infrastructure based RO to maintain the benefits of RO while = preserving > radio resources. > i agree that this is good in some cases, however, i think that it may=20 be possible to design such a solution without requiring announcing=20 routes globally > - trust model: The global HAHA trust model is similar to classical > (phone) roaming. It is something that ISPs can understand and deploy.=20= > In > particular, it is possible to push policies to the proxy, acting like = a > LMM anchor. yes, but e2e RO is similar to mipv6 RO, which is also an existent model=20= that i guess people are starting to adopt. fwiw, i think this is the type of discussion that is needed before=20 deciding which type of RO solution is required regards, marcelo > > I hope this helps... > > Pascal > From nemo-bounces@ietf.org Wed Apr 12 11:09:19 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTgy6-0000R7-ND; Wed, 12 Apr 2006 11:09:18 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTgy5-0000R2-HG for nemo@ietf.org; Wed, 12 Apr 2006 11:09:17 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTgy1-0000vp-QQ for nemo@ietf.org; Wed, 12 Apr 2006 11:09:17 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 12 Apr 2006 17:09:14 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3CF9CUE008136; Wed, 12 Apr 2006 17:09:12 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Wed, 12 Apr 2006 17:09:11 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Wed, 12 Apr 2006 17:09:03 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC020E9FBF@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZeOTqGYFh6SM88RWyDrzpUyKDQvAAAcWww From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 12 Apr 2006 15:09:11.0990 (UTC) FILETIME=[0E271560:01C65E43] X-Spam-Score: 0.0 (/) X-Scan-Signature: 156eddb66af16eef49a76ae923b15b92 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >> Global HAHA, which covers the RO within the infrastructure, is well >> advanced; > >what do you mean by "well advanced"? there are several nemo RO that has >proposed ID and has implementations and so on... > > >moreover, there are RO solution that don't even require new protocols, >so i guess those are really advanced, in the sense that they are >possible with current specs... [Pascal] An informational draft would be enough, then. Do you have an example? Now, go through the MLs (at least NEMO and MONAMI) and you'll find the numerous discussions on global HAHA. AFAIR, it was also discussed at all IETFs meetings for at least 4/5 sessions - chair help please. This makes it well advanced and explains its advanced position. > >> it has been supported up to the plenary, > >what plenary? ietf plenary? since when IETf plenary supports solutions? >I guess this is the wg to decide, not the plenary... [Pascal] It was presented at the IETF plenary as something wanted. Was that IETF 62 or 63? >not sure what common thinking you are talking about, but i would really >amazed if there is a common thinking in the community to support a PI >based solution.... =20 [Pascal] Please browse the global HAHA draft. You will not find a mention of PI.=20 With global HAHA, only the most aggregated prefix appears in the DFZ (like a PA). And since it is in the best interest of ISPs to form a single (or a minimum set of competing) consortium, then ideally you end up with a single prefix. I thought you agreed on that considering your answer to Keiichi-san. > >> Now, it is also my expectation is that NEMO will recharter again in a >> year or so to solve more RO problems. If you have in mind a MR to MR RO >> such as covered by MIRON, then I believe that if a group accepts the >> document, it should definitely be discussed within NEMO at that point. >> > >not sure why do you think that we need two different stages... is this >has been discussed in this wg? wouldn't this be part of the >rechartering discussion that is having place right now in this wg? > [Pascal] Please browse TJ's proposed charter: Mar 2006 Submit final doc(s) on Analysis of the Solution Space for Route Optimization Mar 2007 Shut down or recharter the WG to solve further identified topics >> >> I was stretching the picture, but the ideas were their. The mobile >> prefix I proposed might not clearly fit in PA, and is not PI in the geo >> sense either. >> > >not sure what do you mean by "PI in the geo sense" >current PI allocations are unrelated to geo-PI schemes, they are simply >independent of any provider but not related to any geography > [Pascal] I might be misinformed but last I checked, IPv4 type PI was not an accepted model in IPv6 and you have to trick to get something equivalent, proving that you interface with a number of other ISPs.=20 My reference about geographic PI addresses is: http://www.ietf.org/internet-drafts/draft-hain-ipv6-pi-addr-use-09.txt=20 >> If you consider that the consortium becomes a virtual provider, then it >> is a form of PA. But the prefix is not subnetted, as far as the >> Internet >> BGP can see, so the aggregate looses some sense. >> >> You might consider that the consortium is provider independent, but the >> prefix is surely not geo based. In fact, it is ubiquitous (close to you >> wherever you are). >> This is why I proposed to go back to IPv6 or IPv6Ops >> to discuss what that prefix is really and what the rules are for the >> RIR >> to allocate it. > >this has no relationship at all with geo-PI > >it is about allocating prefixes directly to end users of allocating >them to providers so they can aggregate all the prefixes of their >customers in a single bgp announcement. > [Pascal] The prefix of the phone (Mobile Router) is obtained from its ISP.=20 It can be expected that it comes from a subnet that the ISP allocated to a region where the user lives. In turn, that subnet part of the share that this ISP got from the mobile prefix of the consortium, which is the only entry in the DFZ for the whole thing.=20 >In this case, it is a single end user that is requiring to burn a bgp >slot, my question is why would them be allowed to do so? what are their >special requirements that make other RO solutions not suitable for >them, but yet suitable for the rest of the people? I am not saying that >there is no such requirements, but if they are, i think they should be >presented to the wg and discussed in order to figure out that there is >no alternative solution, that does not imposes additional stress to the >global routing tables, that would fulfill the requirements. That is >why, imho the first step is to flesh out these special requirements... >I mean it is really hard to determine if this si the only acceptable >solution if we don't know what the requirements are... [Pascal] Well, I tried to interrupt that discussion which was creating its own problems and then could not solve them, by giving an example of what a mobile prefix could be.=20 > >> >> In any fashion, if there are few large consortiums, then this is surely >> not a burden for the DFZ. > >yes, right and there are really few people called Marcelo Bagnulo and >this is not a valid reason for allocating a PI prefix to those who are >called Marcelo Bagnulo. >The reason to define a solution that relies in the injection of PI >prefixes in the global routing table is that there is no other solution >that will be suitable for the requirements imposed. So again the first >thing would be to agree in what the requirements are > [Pascal] sorry, we are completely off sync. Global HAHA is not injecting Marcelo into the internet.=20 The requirement imposed is to advertise a short prefix from multiple points in the internet. That does not mean PI, does it? The traditional problem with PI is that the prefix for every site (every corp and so on) ends up in the DFZ, causing an explosion of the BGP tables of the core. >From reading Tony's draft, you might realize that this is not necessarily true anymore. Now, If there were only a handful of prefixes, even if you rated them PI for some semantic reason, that would not be an issue. Global HAHA is designed for a handful of short prefixes injected ubiquitously into the Internet. Any of the global HAHA POPs is an entry to a sub-internet where another routing hierarchy takes place, like within a VPN. MIP proxy HAs, located at the POP, enable POP to POP RO, and the routing between the POPs takes place within the internal network, which is a mesh of private lines and tunnels. Pascal From nemo-bounces@ietf.org Wed Apr 12 12:12:44 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FThxK-00064P-Ho; Wed, 12 Apr 2006 12:12:34 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FThwE-0005E6-P2 for nemo@ietf.org; Wed, 12 Apr 2006 12:11:26 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FThqM-0002xD-IA for nemo@ietf.org; Wed, 12 Apr 2006 12:05:24 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 12 Apr 2006 18:05:23 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3CG5IUE025184; Wed, 12 Apr 2006 18:05:20 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Wed, 12 Apr 2006 18:05:18 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Wed, 12 Apr 2006 18:05:09 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC020E9FF8@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: draft-thubert-nemo-global-haha-01.txt Thread-Index: AcZYmk3is9SvJbI2QI6wJiM6ak7TLgFr9+oQ From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" , "Ryuji Wakikawa" X-OriginalArrivalTime: 12 Apr 2006 16:05:18.0579 (UTC) FILETIME=[E4CBCC30:01C65E4A] X-Spam-Score: 0.0 (/) X-Scan-Signature: d6b246023072368de71562c0ab503126 Cc: ml-nemo WG Subject: [nemo] RE: draft-thubert-nemo-global-haha-01.txt X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi again Marcelo >section 5.1.1 states > > In return, each site advertises the aggregation that encompasses the > Distributed Home Network aggregation back into the service provider > network, to the outside world (the internet). > >this seems to me to imply that the home network prefix is announced >through BGP to the rest of the internet... i mean if you want to >announce something to the rest of the internet it needs to be done with >BGP right? > [Pascal] I understand that this wording is disturbing you and why. I agree I need to fix the text; I will try to reflect the mobile aggregation we discussed based on our long yet positive thread :) Pascal From nemo-bounces@ietf.org Wed Apr 12 12:18:33 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTi37-0000or-OL; Wed, 12 Apr 2006 12:18:33 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FTi36-0000om-4n for nemo@ietf.org; Wed, 12 Apr 2006 12:18:32 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FTi34-0003XI-RD for nemo@ietf.org; Wed, 12 Apr 2006 12:18:32 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 12 Apr 2006 18:18:31 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3CGISUE028302; Wed, 12 Apr 2006 18:18:29 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Wed, 12 Apr 2006 18:18:28 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Wed, 12 Apr 2006 18:17:56 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC020EA005@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZdXrFSo/aLRzrQQsCdHP9aCSulpgA7IUPQ From: "Pascal Thubert \(pthubert\)" To: "Keiichi SHIMA" , X-OriginalArrivalTime: 12 Apr 2006 16:18:28.0860 (UTC) FILETIME=[BBD71BC0:01C65E4C] X-Spam-Score: 0.0 (/) X-Scan-Signature: 9182cfff02fae4f1b6e9349e01d62f32 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Keiichi san >I'm afraid many people don't understand what the global HA-HA can >provide (including me). [Pascal] At the core, global HAHA proposes a NEMO RO solution within the infrastructure, intended for 4G phones, planes, etc... It enables the distribution of HA (getting rid of the Home Link) and introduces the concept of proxy HA. Proxy HAs can be widely distributed. A MR binds to the closest proxy (proxy binding) which succeeds if the proxy can bind to the MR HA (primary binding). When two MRs need to establish a RO path, this is actually performed by the proxies (secondary binding) between each other, on behalf of the MRs. Proxies discover each other using a form of NHRP, and establish a tunnel. At that point, since there is a trust model within the infrastructure, a proxy HA can inject a route to the MNP of its MR over the proxy to proxy tunnel, enabling RO. RO can happen within an ISP network (as described in that section Marcelo pointed out) or within an aggregation that is shared between ISPs, as discussed within the recent thread. Pascal From nemo-bounces@ietf.org Thu Apr 13 21:20:02 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUCyO-0007vz-Pz; Thu, 13 Apr 2006 21:19:44 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUCyO-0007vu-4V for nemo@ietf.org; Thu, 13 Apr 2006 21:19:44 -0400 Received: from mandala.kddilabs.jp ([192.26.91.6]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FUCyL-0003uf-9H for nemo@ietf.org; Thu, 13 Apr 2006 21:19:42 -0400 Received: from localhost (localhost [127.0.0.1]) by mandala.kddilabs.jp (Postfix) with ESMTP id CEB4BEC82F; Fri, 14 Apr 2006 10:19:38 +0900 (JST) Received: from neutrino.hsc.kddilabs.jp (neutrino.hsc.kddilabs.jp [2001:200:601:200::60]) by mandala.kddilabs.jp (Postfix) with ESMTP id 1EFFDEC83E; Fri, 14 Apr 2006 10:19:38 +0900 (JST) Received: from [IPv6:2001:200:601:200:9048:7e3e:16d4:513c] (unknown [IPv6:2001:200:601:200:9048:7e3e:16d4:513c]) by neutrino.hsc.kddilabs.jp (Postfix) with ESMTP id 78EC52E02E; Fri, 14 Apr 2006 10:14:55 +0900 (JST) Message-ID: <443EF89D.2020508@kddilabs.jp> Date: Fri, 14 Apr 2006 10:19:25 +0900 From: Masafumi Watari Organization: KDDI R&D Laboratories Inc. User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "Pascal Thubert (pthubert)" Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new X-Spam-Score: 0.0 (/) X-Scan-Signature: 7e439b86d3292ef5adf93b694a43a576 Cc: marcelo bagnulo braun , nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi, On 2006/04/12 22:18, Pascal Thubert (pthubert) wrote: > Hi Marcelo: > >> you stated that: >> >>> - The NEMO part for this discussion is enabling Route optimization >>> within the infrastructure. There are a number of use cases to sustain >>> the need. IP mobile telephony (4G?) is an example, where Service >>> Providers do not necessarily wish to have the phone perform the RO >>> themselves. >>> >>> - Global HAHA introduce proxy HAs which enable this route > optimization >>> with a trust model in place. I'd say that defining how that works is >>> enough work for NEMO; whereas proposing the deployment model should > be >>> a >>> request from us to other WGs. >>> >> IMHO, we should first decide on which forms of RO for nemo are we going >> to work on, and then move on to specific solutions. >> >> I mean, as specified in the RO taxonomy draft there are several and >> very different approaches to RO. I think that in order to do the >> charter for future work, we need first to decide which one of the RO >> approaches are we going to work on. >> >> However, the proposed charter only has Global HAHA as a RO solution for >> nemo, so it seems to imply that this is the RO that we have selected. >> I would propose to open the general discussion about which RO >> approaches are we going to work on first rather than focussing in if we >> are going to work on global haha or not... The charter is still open for discussion :) The only reason for Global HAHA being mentioned in the charter I believe is because it's been presented several times in different WGs, though originally it wasn't about RO but more on HA redundancy I believe(local HAHA?). > [Pascal] > > My understanding is that the new charter covers a transition time. > Global HAHA, which covers the RO within the infrastructure, is well > advanced; it has been supported up to the plenary, so there's a common > thinking that we can go ahead and move it forward. I don't know about the plenary, but in the sense that Global HAHA seems deployable compared to other solution, I guess it is advanced. But note however that other RO solutions have not had their chance to present their work because the WG was not chartered to work on it. > Now, it is also my expectation is that NEMO will recharter again in a > year or so to solve more RO problems. If you have in mind a MR to MR RO > such as covered by MIRON, then I believe that if a group accepts the > document, it should definitely be discussed within NEMO at that point. > > Finally, some RO schemes might not belong to NEMO as a WG. For instance, > there's a question whether MANET for NEMO (MANEMO) should be done within > NEMO or as a separate WG, because of the specific skills and interests > involved. > > > >> Second, a more focussed issue with respect to routing system >> scalability and business model >> >> you stated that: >> >>> Let me give you an example. Say that 3-4 ISPs from different > continents >>> form a mobile ISP consortium. >>> >>> That consortium obtains a short prefix from one of the NICs, call it >>> the >>> mobile prefix. Internally, the consortium shares that addressing > space >>> between members, leaving big chunks free for new comers. The rule is >>> that only the big aggregation is redistributed over the Internet BGP, >>> from most POP of members ISPs. But internally, the consortium of >>> service >>> providers runs a second BGP which deals with the subnetting of mobile >>> prefix, over a mesh of tunnels and private links. >> but this seems to be complettelly different that what other people have >> been claiming here on the list. >> Past comments have strongly argued that global mobile network need PI >> addressing and global HAHA seemed a mechanisms to provide access to >> this PI prefix that the airline company gets. >> >> This is completelly different than the mobile ISP model that you are >> suggesting. > > I was stretching the picture, but the ideas were their. The mobile > prefix I proposed might not clearly fit in PA, and is not PI in the geo > sense either. > > If you consider that the consortium becomes a virtual provider, then it > is a form of PA. But the prefix is not subnetted, as far as the Internet > BGP can see, so the aggregate looses some sense. > > You might consider that the consortium is provider independent, but the > prefix is surely not geo based. In fact, it is ubiquitous (close to you > wherever you are). This is why I proposed to go back to IPv6 or IPv6Ops > to discuss what that prefix is really and what the rules are for the RIR > to allocate it. > > In any fashion, if there are few large consortiums, then this is surely > not a burden for the DFZ. > In fact, there would be another DFZ within the consortium to handle that > prefix as a sub-internet. > GTP proves that providers are willing to congregate when they find that > such is their interest. > > >> What worries me is that i don't see any mobile ISP in the list claiming >> that they need global HAHA to deploy their business, but there are end >> users (air companies) that claim that want PI addressing for deploying >> this. >> > If I understand well Tony's draft, NEMO is not the right forum for > Provider Independent (PI) allocation policy. And that work is still > ongoing, somewhere else... > > As for ISP side, I'm copying Masafumi-san who co authored the NEMO RO > PS. We are now migrating our network based on MMD and NGN, and within the next two years all of our mobile and fixed network would be based on IP. Once that's done, I would imagine that something like global HAHA would be needed. But if global HAHA is for NEMO and not MIP6, I don't see a strong need at this point in time. > Note that the work on global HAHA started with voice ISPs and 4G design. > There was a strong requirement to make the phones unaware of RO. Some of > the reasons: > > - privacy: one should not know where the mobile correspondent. In other > words, a phone should only see the other phone's Home Address, not its > CoA. > > - phone: RO might cause interoperability problems between generations of > phones. You want the phone software to be simple - limited resources - > and conservative -protocol ha to remain stable for years - so changes > must occur centrally in the ISP side. It's also true that operators need to have control of their service too. watari > - radio: client based RO consumes additional resources on the first hop > whereas it saves resources globally. When the first hop is a radio and > the rest of the way is high speed wire, you might prefer the > infrastructure based RO to maintain the benefits of RO while preserving > radio resources. > > - trust model: The global HAHA trust model is similar to classical > (phone) roaming. It is something that ISPs can understand and deploy. In > particular, it is possible to push policies to the proxy, acting like a > LMM anchor. > > I hope this helps... > > Pascal > From nemo-bounces@ietf.org Thu Apr 13 22:08:38 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUDja-0002HT-9P; Thu, 13 Apr 2006 22:08:30 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUDjZ-000248-52 for nemo@ietf.org; Thu, 13 Apr 2006 22:08:29 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FUDU6-0004kV-UA for nemo@ietf.org; Thu, 13 Apr 2006 21:52:31 -0400 Received: from mandala.kddilabs.jp ([192.26.91.6]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FUDJN-0002dX-Uu for nemo@ietf.org; Thu, 13 Apr 2006 21:41:28 -0400 Received: from localhost (localhost [127.0.0.1]) by mandala.kddilabs.jp (Postfix) with ESMTP id 2BD21EC88F; Fri, 14 Apr 2006 10:41:23 +0900 (JST) Received: from neutrino.hsc.kddilabs.jp (neutrino.hsc.kddilabs.jp [2001:200:601:200::60]) by mandala.kddilabs.jp (Postfix) with ESMTP id C8A90EC88A; Fri, 14 Apr 2006 10:41:22 +0900 (JST) Received: from [IPv6:2001:200:601:200:9048:7e3e:16d4:513c] (unknown [IPv6:2001:200:601:200:9048:7e3e:16d4:513c]) by neutrino.hsc.kddilabs.jp (Postfix) with ESMTP id 1C3332E02E; Fri, 14 Apr 2006 10:36:40 +0900 (JST) Message-ID: <443EFDB6.2090302@kddilabs.jp> Date: Fri, 14 Apr 2006 10:41:10 +0900 From: Masafumi Watari Organization: KDDI R&D Laboratories Inc. User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "Dul, Andrew L" Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA8@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AAA8@XCH-NW-7V1.nw.nos.boeing.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: quoted-printable X-Virus-Scanned: by amavisd-new X-Spam-Score: -2.6 (--) X-Scan-Signature: 4b800b1eab964a31702fa68f1ff0e955 Cc: marcelo bagnulo braun , ml-nemo WG , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org On 2006/04/08 2:39, Dul, Andrew L wrote: >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 >> Sent: Thursday, April 06, 2006 10:43 PM >> To: Alexandru Petrescu >> Cc: ml-nemo WG; Pascal Thubert (pthubert); T.J. Kniveton;=20 >> Davis, Terry L; Vijay Devarapalli >> Subject: Re: About route aggregation in global HAHA (was=20 >> Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter=20 >> proposal) >> >> >> El 06/04/2006, a las 18:02, Alexandru Petrescu escribi=F3: >> >>>>> In general, I am not at all sure where Mobile IP is needed in a=20 >>>>> closed (or small) system. >>>> yes, but from other comments, i understood that they have=20 >> different=20 >>>> ISPs in different locations, so it is far from being a=20 >> closed system. >>> If they control all these ISPs it all becomes a "closed" system,=20 >>> manageable by a single entity. >> if all the ISPs involved are peering directly this may be=20 >> easy, but if they are not, they need to start configuring=20 >> tunnels between them to pass the routing information and the=20 >> situation becomes rather messy and hard to manage imho (but i=20 >> agree that is an option... see more below) >=20 > Yes, this is one of my operational concerns with global-HAHA > =20 >>> If the aviation industry _can_ say to ISP1 "please place a=20 >> HA here"=20 >>> to ISP2 "please place a HA there" then it means that they=20 >> can already=20 >>> tell ISPs what to do. >>> >>> And if they can do that, then they can also tell ISPs "ISPs please=20 >>> exchange routes about the MR when you sense it in your=20 >> location". And=20 >>> if the ISPs exchange these routes then no need of Mobile IP. >>> >> but are you assuming that the nemo has its own PI prefix or not? >> >=20 > I personally think you have to assume that globally mobile networks hav= e PI address space. >=20 >> - if it has its own PI prefix, this is basically what has=20 >> been presented in the slides by boeing, that is what they are=20 >> currently doing.=20 >=20 > Just because you have a PI prefix doesn't mean that it has to be 1 pref= ix in the global routing table per mobile-network. It is possible to agg= regate. >=20 >> the problem with this is that each nemo adds=20 >> an additional entry to the global routing table and they=20 >> depend on BGP convergence times (which won't be that good if=20 >> they keep on adding routes i am afraid) >=20 > It is something we are keeping an eye on, but at this time there are pl= enty of other people out there who are putting lots of useless updates in= to the routing system... >=20 > http://www3.ietf.org/proceedings/06mar/slides/grow-3.pdf Which is not acceptable, adding even greater concerns when more sites start multihoming, is what I heard... watari From nemo-bounces@ietf.org Fri Apr 14 13:49:50 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUSQQ-00087E-BY; Fri, 14 Apr 2006 13:49:42 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUSQP-000879-Le for nemo@ietf.org; Fri, 14 Apr 2006 13:49:41 -0400 Received: from blv-smtpout-01.boeing.com ([130.76.32.69]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FUSQP-0001TE-9Q for nemo@ietf.org; Fri, 14 Apr 2006 13:49:41 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by blv-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id KAA10868; Fri, 14 Apr 2006 10:49:33 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k3EHnXT13432; Fri, 14 Apr 2006 10:49:33 -0700 (PDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 14 Apr 2006 10:49:30 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Fri, 14 Apr 2006 10:49:30 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AADF@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Thread-Index: AcZcbiA+h592IJ3RRjqcHSMAEKyQ5ADeKtsg From: "Dul, Andrew L" To: "marcelo bagnulo braun" , "Alexandru Petrescu" X-OriginalArrivalTime: 14 Apr 2006 17:49:30.0799 (UTC) FILETIME=[C83C5BF0:01C65FEB] X-Spam-Score: 0.0 (/) X-Scan-Signature: 6cca30437e2d04f45110f2ff8dc1b1d5 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > El 07/04/2006, a las 21:39, Alexandru Petrescu escribi=F3: >=20 > > Thanks for the example! > > > > Dul, Andrew L wrote: > >> Let say I have lots of aircraft and I assign each of them=20 > a /48 out=20 > >> of a /32. The individual prefixes would be the /48 for each=20 > >> aircraft. For a simple case with 2 HA's each in a=20 > different ISP. If=20 > >> both ISPs announce the /32 via BGP but carry the /48 addresses=20 > >> internally, then the Internet doesn't see the /48s and reaches the=20 > >> aircraft via closest ISP announcing the /32. > > > > Interesting. Put this way looks ok scalable, no BGP route=20 > injection=20 > > for mobility. > > >=20 > well, we still need to inject one bgp route per distributed=20 > home network, right? I think I'm getting confused on the terminology. I tried going back to = the nemo-terminology draft but that didn't help... :( We need to inject at least 1 BGP route. Theoretically all mobile nodes = could be aggregated into a single route, however that isn't likely. The = best case that I see happening is an aggregated prefix per = "mobile-network operator". Where a "mobile network operator" is = responsible for the provisioning an providing service to a large number = of mobile platforms. =20 Make sense? >=20 > > > ... > > > > Third, if the two ISPs are tightly linked together somehow (because=20 > > they advertise same /32) then it would be sufficient for=20 > them to run=20 > > route updates between them to point exactly where the /48 is, thus=20 > > avoiding core-BGP injections _and_ Mobile IP tunnelling. > > Advertising BGP routes between providers certainly is an option, but the = complexity can really grow when there are more than two providers. My = concern about that is that you are now seperating the mobility = signalling between the NEMO and BGP. It seems like it would be easier = operationally if your mobility signalling was in one protocol not two.=20 >=20 > good point, >=20 > basically if the aircraft already knows what ISPs it will be=20 > visiting and it has some commercial arrangement with it, it=20 > is likely that the ISP can accept routes from the aircraft=20 > when it is visiting, so no mobile is required whatsoever >=20 > I guess that the bottom line is that the trust/commercial=20 > model in the case of an aircraft visiting an ISP is different=20 > from the one assumed in mobile IP. I mean, i guess we can=20 > assume a much more higher level of trust between the aircraft=20 > and the visited ISP, so that route injection is an option (we=20 > can see that from boeing presentation, i guess) >=20 I can see there being a higher level of trust as you describe. Route = injection could be an option. I don't necessarily think that is the = best option though. Route injection similar to what we are doing today = requires a high level of integration between ISPs. =20 Andrew =20 From nemo-bounces@ietf.org Fri Apr 14 15:30:51 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUU09-0005fQ-4c; Fri, 14 Apr 2006 15:30:41 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FUU08-0005fL-3b for nemo@ietf.org; Fri, 14 Apr 2006 15:30:40 -0400 Received: from blv-smtpout-01.boeing.com ([130.76.32.69]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FUU04-0005Ls-CX for nemo@ietf.org; Fri, 14 Apr 2006 15:30:40 -0400 Received: from stl-av-01.boeing.com ([192.76.190.6]) by blv-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id MAA28585; Fri, 14 Apr 2006 12:30:33 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by stl-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k3EJUWN13068; Fri, 14 Apr 2006 14:30:32 -0500 (CDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 14 Apr 2006 12:30:23 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="----_=_NextPart_001_01C65FF9.DF73011A" Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Date: Fri, 14 Apr 2006 12:30:22 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AAE2@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC020E9FBF@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Thread-Index: AcZeOTqGYFh6SM88RWyDrzpUyKDQvAAAcWwwAG0ZI4A= From: "Dul, Andrew L" To: "Pascal Thubert \(pthubert\)" , "marcelo bagnulo braun" X-OriginalArrivalTime: 14 Apr 2006 19:30:23.0665 (UTC) FILETIME=[E0066A10:01C65FF9] X-Spam-Score: 0.0 (/) X-Scan-Signature: cad4f6755f654f51ca3500ae415e5b7d Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org This is a multi-part message in MIME format. ------_=_NextPart_001_01C65FF9.DF73011A Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable > -----Original Message----- > From: Pascal Thubert (pthubert) [mailto:pthubert@cisco.com]=20 >=20 > > > >> it has been supported up to the plenary, > > > >what plenary? ietf plenary? since when IETf plenary supports=20 > solutions? > >I guess this is the wg to decide, not the plenary... >=20 > [Pascal] It was presented at the IETF plenary as something=20 > wanted. Was that IETF 62 or 63? >=20 >=20 I think people are referring to the presentation I made at IETF 62. A copy of the presentation was unfortunately omitted from the proceedings. I've posted a copy here for people who are interested. http://www.quark.net/docs/Global_IP_Mobility_IETF62.pdf =20 The point of the presentation was to present the solution we had deployed for ipv4 network mobility for our service and to spark conversations about how network mobility might be done in the future.=20 I've also recently written a whitepaper which describes our current solution as well as some additional information. I've attached a copy to this message. Andrew ------_=_NextPart_001_01C65FF9.DF73011A Content-Type: application/octet-stream; name="Global IP Network Mobility using Border Gateway Protocol.pdf" Content-Transfer-Encoding: base64 Content-Description: Global IP Network Mobility using Border Gateway Protocol.pdf Content-Disposition: attachment; filename="Global IP Network Mobility using Border Gateway Protocol.pdf" JVBERi0xLjYNJeLjz9MNCjM2NyAwIG9iaiA8PC9MaW5lYXJpemVkIDEvTCAzNDQyMDcvTyAzNzIv RSAxNjY0NDcvTiA5L1QgMzM2ODAzL0ggWyAxMzM5IDQ1Ml0+Pg1lbmRvYmoNICAgICAgICAgICAg DQp4cmVmDQozNjcgNTANCjAwMDAwMDAwMTYgMDAwMDAgbg0KMDAwMDAwMTk3MiAwMDAwMCBuDQow MDAwMDAyMTEwIDAwMDAwIG4NCjAwMDAwMDI0MTcgMDAwMDAgbg0KMDAwMDAwMjQ2MCAwMDAwMCBu DQowMDAwMDAyNjE5IDAwMDAwIG4NCjAwMDAwMDMxODAgMDAwMDAgbg0KMDAwMDAwMzYxMSAwMDAw MCBuDQowMDAwMDA0MTI4IDAwMDAwIG4NCjAwMDAwMDQ2NzAgMDAwMDAgbg0KMDAwMDAwNTI2MiAw MDAwMCBuDQowMDAwMDA1NTMxIDAwMDAwIG4NCjAwMDAwMDU3NzkgMDAwMDAgbg0KMDAwMDAwNjA1 MyAwMDAwMCBuDQowMDAwMDA2MzA2IDAwMDAwIG4NCjAwMDAwMDY0MDggMDAwMDAgbg0KMDAwMDAw ODIxNCAwMDAwMCBuDQowMDAwMDA5NzcxIDAwMDAwIG4NCjAwMDAwMTA2ODUgMDAwMDAgbg0KMDAw MDAxMTg3MSAwMDAwMCBuDQowMDAwMDEzMjAwIDAwMDAwIG4NCjAwMDAwMTQ0MzMgMDAwMDAgbg0K MDAwMDAxNDU1NyAwMDAwMCBuDQowMDAwMDE0NjAxIDAwMDAwIG4NCjAwMDAwMTU4NDYgMDAwMDAg bg0KMDAwMDAxNzU0MiAwMDAwMCBuDQowMDAwMDYwOTcwIDAwMDAwIG4NCjAwMDAwOTE1MTEgMDAw MDAgbg0KMDAwMDEyNzE0NCAwMDAwMCBuDQowMDAwMTUxMzQzIDAwMDAwIG4NCjAwMDAxNTE5MDAg MDAwMDAgbg0KMDAwMDE1MjI0MyAwMDAwMCBuDQowMDAwMTUyOTYyIDAwMDAwIG4NCjAwMDAxNTMy NzcgMDAwMDAgbg0KMDAwMDE1MzgyNSAwMDAwMCBuDQowMDAwMTU0MDQxIDAwMDAwIG4NCjAwMDAx NTQ2NzkgMDAwMDAgbg0KMDAwMDE1NTA4OCAwMDAwMCBuDQowMDAwMTU1MjU2IDAwMDAwIG4NCjAw MDAxNTU3NzQgMDAwMDAgbg0KMDAwMDE1NjA5OCAwMDAwMCBuDQowMDAwMTU2ODc2IDAwMDAwIG4N CjAwMDAxNTc4MzUgMDAwMDAgbg0KMDAwMDE1ODIxNSAwMDAwMCBuDQowMDAwMTU5MTI3IDAwMDAw IG4NCjAwMDAxNTk0MTkgMDAwMDAgbg0KMDAwMDE2NjAxMSAwMDAwMCBuDQowMDAwMTY2MjUyIDAw MDAwIG4NCjAwMDAwMDE3OTEgMDAwMDAgbg0KMDAwMDAwMTMzOSAwMDAwMCBuDQp0cmFpbGVyDQo8 PC9TaXplIDQxNy9QcmV2IDMzNjc5MS9YUmVmU3RtIDE3OTEvUm9vdCAzNjkgMCBSL0VuY3J5cHQg MzY4IDAgUi9JbmZvIDE1MyAwIFIvSURbPDcyQUYyRUM0NzczMUVBRDE5MTUxN0QyQjMxNDFFRjIz Pjw2NEU1MDQ4NTZBM0I1NjQ2OUFERTAzRjk1RDFCRjI5Nz5dPj4NCnN0YXJ0eHJlZg0KMA0KJSVF T0YNCiAgICAgICAgICAgICAgICAgICAgICANCjQxNiAwIG9iajw8L0xlbmd0aCAzNTIvQiA0MDkv QyAzODcvRmlsdGVyL0ZsYXRlRGVjb2RlL0kgNDMzL0wgMzcxL1MgMjQ5Pj5zdHJlYW0NCtlcUtsi u6aJheG7geLwnsFEprOXRoMBIw5KoqK61GEx7KU6GQKVSaGTWcJmvMAwJvasC7PgFmLsh9FqM3Yn WphPobIwixYMJ2EIhbGzx2vZ7rg8ZrpuBFdzyP3SmK0sAjNXy3PDMV0PegHBs7dfyst+6bXbzfIG 30lc8EpPGfGJDcXXG62GtyHVDS1J7TLJSCKVx7bI8OgbIC/Th0/0KUfLOKqCDwwXjOvkEXIljVEn 8bMHTkR8nmkTVQ4sv4t/7D6rfJEOyvYcZz53n+5htnWl3dmSEhM6pu0EqCE76SAP3HxghxnPYXBP cPMEfvTj8oWCxqujqyN19tAbOzI1t8u5+FcxWN0rOUYCgx8o6jKbe0rNu+Vm4iIk+cdB4X1Nl9E/ SapBluQZx0+BUpO9ErV79/0pFoiUMgM9wuL3vP+riE1ij5hfQJ764w2hMhu/g9gsLAyN+3NuXjka pGB8TOgNCmVuZHN0cmVhbQ1lbmRvYmoNNDE1IDAgb2JqPDwvTGVuZ3RoIDMwL0ZpbHRlci9GbGF0 ZURlY29kZS9XWzEgMSAxXS9JbmRleFsxNTQgMjEzXS9EZWNvZGVQYXJtczw8L0NvbHVtbnMgMy9Q cmVkaWN0b3IgMTI+Pi9TaXplIDM2Ny9UeXBlL1hSZWY+PnN0cmVhbQ0KeNpiYupjYGJgYBzFgwUz zh0Ng6EdHwABBgA5zwRJDQplbmRzdHJlYW0NZW5kb2JqDTM2OCAwIG9iajw8L0xlbmd0aCAxMjgv RmlsdGVyL1N0YW5kYXJkL08oUNTJ+JlF9lwo9v71n7HqDLd6QE/bTRv/hDgCXniJFG9eKS9QIC0x MzQwL1IgMy9VKCTyTxHZkuyWXChMXmpsInTmAAAAAAAAAAAAAAAAAAAAACkvViAyPj4NZW5kb2Jq DTM2OSAwIG9iajw8L01hcmtJbmZvPDwvTGV0dGVyc3BhY2VGbGFncyAwL01hcmtlZCB0cnVlPj4v TmFtZXMgMzcwIDAgUi9NZXRhZGF0YSAxNTIgMCBSL1BpZWNlSW5mbzw8L01hcmtlZFBERjw8L0xh c3RNb2RpZmllZCgHPFTduqIiGOw1MKHXIfAvKT4+Pj4vUGFnZXMgMTQ4IDAgUi9QYWdlTGF5b3V0 L09uZUNvbHVtbi9PQ1Byb3BlcnRpZXM8PC9PQ0dzWzM3MSAwIFJdPj4vU3RydWN0VHJlZVJvb3Qg MTU0IDAgUi9UeXBlL0NhdGFsb2cvTGFzdE1vZGlmaWVkKAc8VN26oiIY7DUwodch8C8pL1BhZ2VM YWJlbHMgMTQ2IDAgUj4+DWVuZG9iag0zNzAgMCBvYmo8PC9FbWJlZGRlZEZpbGVzIDE0OSAwIFI+ Pg1lbmRvYmoNMzcxIDAgb2JqPDwvVXNhZ2U8PC9DcmVhdG9ySW5mbzw8L1N1YnR5cGUvQXJ0d29y ay9DcmVhdG9yKE56LBntoxBstiuZn1SeP9UxQOQPjj8g4LA9+2p2QDkpPj4vUGFnZUVsZW1lbnQ8 PC9TdWJ0eXBlL0hGPj4+Pi9OYW1lKEd8PxLqsCIjiRu6oCkvVHlwZS9PQ0c+Pg1lbmRvYmoNMzcy IDAgb2JqPDwvQ3JvcEJveFswIDAgNjEyIDc5Ml0vUGFyZW50IDE0OCAwIFIvU3RydWN0UGFyZW50 cyAwL0NvbnRlbnRzWzM4MiAwIFIgMzgzIDAgUiAzODQgMCBSIDM4NSAwIFIgMzg2IDAgUiAzODcg MCBSIDM5MCAwIFIgMzkxIDAgUl0vUm90YXRlIDAvTWVkaWFCb3hbMCAwIDYxMiA3OTJdL1Jlc291 cmNlczw8L1hPYmplY3Q8PC9JbTEwIDM5NiAwIFIvSW0xMSAzOTcgMCBSL0ltMTIgMzk4IDAgUi9J bTEzIDM5OSAwIFIvSW0xNCA0MDAgMCBSL0ltMTUgNDAxIDAgUi9JbTAgNDAyIDAgUi9JbTEgNDAz IDAgUi9JbTIgNDA0IDAgUi9JbTMgNDA1IDAgUi9JbTQgNDA2IDAgUi9JbTUgNDA3IDAgUi9JbTYg NDA4IDAgUi9JbTcgNDA5IDAgUi9JbTggNDEwIDAgUi9JbTkgNDExIDAgUj4+L0ZvbnQ8PC9UVDAg MzczIDAgUi9UVDEgMzc0IDAgUi9UVDIgMzc1IDAgUi9UVDMgMzc2IDAgUi9DMl8wIDM4OCAwIFI+ Pi9Qcm9jU2V0Wy9QREYvVGV4dC9JbWFnZUJdL1Byb3BlcnRpZXM8PC9NQzAgMzg5IDAgUj4+L0V4 dEdTdGF0ZTw8L0dTMCAzODEgMCBSPj4+Pi9UeXBlL1BhZ2U+Pg1lbmRvYmoNMzczIDAgb2JqPDwv U3VidHlwZS9UcnVlVHlwZS9Gb250RGVzY3JpcHRvciAzODAgMCBSL0xhc3RDaGFyIDEyMS9XaWR0 aHNbMjc4IDAgMCA1NTYgMCAwIDAgMCAzMzMgMzMzIDAgMCAwIDMzMyAyNzggMCA1NTYgNTU2IDU1 NiA1NTYgMCAwIDAgMCA1NTYgNTU2IDMzMyAwIDAgMCAwIDAgMCA3MjIgNzIyIDAgNzIyIDY2NyAw IDc3OCAwIDI3OCAwIDAgNjExIDgzMyA3MjIgNzc4IDY2NyAwIDAgMCA2MTEgMCAwIDk0NCAwIDAg MCAwIDAgMCAwIDAgMCA1NTYgNjExIDU1NiA2MTEgNTU2IDAgNjExIDAgMjc4IDAgNTU2IDI3OCA4 ODkgNjExIDYxMSA2MTEgMCAzODkgNTU2IDMzMyA2MTEgMCA3NzggNTU2IDU1Nl0vQmFzZUZvbnQv RkFNTkVBK0FyaWFsLUJvbGRNVC9GaXJzdENoYXIgMzIvRW5jb2RpbmcvV2luQW5zaUVuY29kaW5n L1R5cGUvRm9udD4+DWVuZG9iag0zNzQgMCBvYmo8PC9TdWJ0eXBlL1RydWVUeXBlL0ZvbnREZXNj cmlwdG9yIDM3OCAwIFIvTGFzdENoYXIgMTUwL1dpZHRoc1syNzggMCAzNTUgMCAwIDAgMCAwIDMz MyAzMzMgMCAwIDAgMzMzIDI3OCAyNzggNTU2IDU1NiA1NTYgNTU2IDU1NiAwIDU1NiA1NTYgNTU2 IDU1NiAwIDAgMCAwIDAgMCAxMDE1IDY2NyA2NjcgNzIyIDcyMiA2NjcgMCA3NzggNzIyIDI3OCAw IDY2NyA1NTYgODMzIDcyMiAwIDY2NyAwIDcyMiA2NjcgNjExIDAgMCA5NDQgNjY3IDY2NyA2MTEg MCAwIDAgMCAwIDAgNTU2IDU1NiA1MDAgNTU2IDU1NiAyNzggNTU2IDU1NiAyMjIgMCA1MDAgMjIy IDgzMyA1NTYgNTU2IDU1NiAwIDMzMyA1MDAgMjc4IDU1NiA1MDAgNzIyIDUwMCA1MDAgMCAwIDAg MCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMjIyIDAgMCAwIDU1Nl0v QmFzZUZvbnQvRkFNTkZCK0FyaWFsTVQvRmlyc3RDaGFyIDMyL0VuY29kaW5nL1dpbkFuc2lFbmNv ZGluZy9UeXBlL0ZvbnQ+Pg1lbmRvYmoNMzc1IDAgb2JqPDwvU3VidHlwZS9UcnVlVHlwZS9Gb250 RGVzY3JpcHRvciAzNzkgMCBSL0xhc3RDaGFyIDE0OC9XaWR0aHNbMjUwIDAgMCAwIDAgMCAwIDAg MzMzIDMzMyAwIDAgMjUwIDMzMyAyNTAgMjc4IDUwMCA1MDAgNTAwIDUwMCA1MDAgNTAwIDUwMCA1 MDAgNTAwIDUwMCAzMzMgMCAwIDAgMCAwIDAgNzIyIDY2NyA3MjIgNzIyIDY2NyA2MTEgNzc4IDc3 OCAzODkgMCA3NzggNjY3IDk0NCA3MjIgNzc4IDYxMSA3NzggNzIyIDU1NiA2NjcgNzIyIDcyMiAx MDAwIDAgNzIyIDAgMCAwIDAgMCAwIDAgNTAwIDU1NiA0NDQgNTU2IDQ0NCAzMzMgNTAwIDU1NiAy NzggMCA1NTYgMjc4IDgzMyA1NTYgNTAwIDU1NiA1NTYgNDQ0IDM4OSAzMzMgNTU2IDUwMCA3MjIg NTAwIDUwMCA0NDQgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAw IDMzMyA1MDAgNTAwXS9CYXNlRm9udC9GQU1OR0YrVGltZXNOZXdSb21hblBTLUJvbGRNVC9GaXJz dENoYXIgMzIvRW5jb2RpbmcvV2luQW5zaUVuY29kaW5nL1R5cGUvRm9udD4+DWVuZG9iag0zNzYg MCBvYmo8PC9TdWJ0eXBlL1RydWVUeXBlL0ZvbnREZXNjcmlwdG9yIDM3NyAwIFIvTGFzdENoYXIg MTY5L1dpZHRoc1syNTAgMCAwIDAgMCAwIDAgMTgwIDMzMyAzMzMgMCAwIDI1MCAzMzMgMjUwIDI3 OCA1MDAgNTAwIDUwMCA1MDAgNTAwIDUwMCA1MDAgNTAwIDUwMCA1MDAgMjc4IDI3OCAwIDU2NCAw IDQ0NCAwIDcyMiA2NjcgNjY3IDcyMiA2MTEgNTU2IDcyMiA3MjIgMzMzIDM4OSA3MjIgNjExIDg4 OSA3MjIgNzIyIDU1NiAwIDY2NyA1NTYgNjExIDcyMiA3MjIgOTQ0IDAgMCAwIDAgMCAwIDAgNTAw IDAgNDQ0IDUwMCA0NDQgNTAwIDQ0NCAzMzMgNTAwIDUwMCAyNzggMjc4IDUwMCAyNzggNzc4IDUw MCA1MDAgNTAwIDUwMCAzMzMgMzg5IDI3OCA1MDAgNTAwIDcyMiA1MDAgNTAwIDQ0NCAwIDAgMCAw IDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMzMzIDQ0NCA0NDQgMCA1MDAg MCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgMCAwIDAgNzYwXS9CYXNlRm9udC9GQU1OQ08r VGltZXNOZXdSb21hblBTTVQvRmlyc3RDaGFyIDMyL0VuY29kaW5nL1dpbkFuc2lFbmNvZGluZy9U eXBlL0ZvbnQ+Pg1lbmRvYmoNMzc3IDAgb2JqPDwvU3RlbVYgODIvRm9udE5hbWUvRkFNTkNPK1Rp bWVzTmV3Um9tYW5QU01UL0ZvbnRTdHJldGNoL05vcm1hbC9Gb250RmlsZTIgMzkyIDAgUi9Gb250 V2VpZ2h0IDQwMC9GbGFncyAzNC9EZXNjZW50IC0yMTYvRm9udEJCb3hbLTU2OCAtMzA3IDIwMDAg MTAwN10vQXNjZW50IDg5MS9Gb250RmFtaWx5KIWBo73KOY1/K1xcTZ5hu2ApL0NhcEhlaWdodCA2 NTYvWEhlaWdodCAtNTQ2L1R5cGUvRm9udERlc2NyaXB0b3IvSXRhbGljQW5nbGUgMD4+DWVuZG9i ag0zNzggMCBvYmo8PC9TdGVtViA4OC9Gb250TmFtZS9GQU1ORkIrQXJpYWxNVC9Gb250U3RyZXRj aC9Ob3JtYWwvRm9udEZpbGUyIDM5MyAwIFIvRm9udFdlaWdodCA0MDAvRmxhZ3MgMzIvRGVzY2Vu dCAtMjExL0ZvbnRCQm94Wy02NjUgLTMyNSAyMDAwIDEwMDZdL0FzY2VudCA5MDUvRm9udEZhbWls eSgTOlxczsYpL0NhcEhlaWdodCA3MTgvWEhlaWdodCA1MTUvVHlwZS9Gb250RGVzY3JpcHRvci9J dGFsaWNBbmdsZSAwPj4NZW5kb2JqDTM3OSAwIG9iajw8L1N0ZW1WIDEzNi9Gb250TmFtZS9GQU1O R0YrVGltZXNOZXdSb21hblBTLUJvbGRNVC9Gb250U3RyZXRjaC9Ob3JtYWwvRm9udEZpbGUyIDM5 NCAwIFIvRm9udFdlaWdodCA3MDAvRmxhZ3MgMzQvRGVzY2VudCAtMjE2L0ZvbnRCQm94Wy01NTgg LTMwNyAyMDAwIDEwMjZdL0FzY2VudCA4OTEvRm9udEZhbWlseShSPvynp4vLZWdq66AdvO0pL0Nh cEhlaWdodCA2NTYvWEhlaWdodCAtNTQ2L1R5cGUvRm9udERlc2NyaXB0b3IvSXRhbGljQW5nbGUg MD4+DWVuZG9iag0zODAgMCBvYmo8PC9TdGVtViAxMzgvRm9udE5hbWUvRkFNTkVBK0FyaWFsLUJv bGRNVC9Gb250U3RyZXRjaC9Ob3JtYWwvRm9udEZpbGUyIDM5NSAwIFIvRm9udFdlaWdodCA3MDAv RmxhZ3MgMzIvRGVzY2VudCAtMjExL0ZvbnRCQm94Wy02MjggLTM3NiAyMDAwIDEwMTBdL0FzY2Vu dCA5MDUvRm9udEZhbWlseSiBTmw+3ikvQ2FwSGVpZ2h0IDcxOC9YSGVpZ2h0IDUxNS9UeXBlL0Zv bnREZXNjcmlwdG9yL0l0YWxpY0FuZ2xlIDA+Pg1lbmRvYmoNMzgxIDAgb2JqPDwvT1BNIDEvT1Ag ZmFsc2UvQkcyL0RlZmF1bHQvb3AgZmFsc2UvVHlwZS9FeHRHU3RhdGUvU0EgZmFsc2UvVUNSMi9E ZWZhdWx0L1NNIDAuMDI+Pg1lbmRvYmoNMzgyIDAgb2JqPDwvTGVuZ3RoIDE3MzUvRmlsdGVyL0Zs YXRlRGVjb2RlPj5zdHJlYW0NCgyBh6xscVRd4zKuskL7RUoH+XL1eWOpiVwm10SIdAqcYPvxrFl2 ClQmucGEVaNlWZ8gE/S6mA2+nYPuchhJ0QLUhcLzaMzGPqZAz1N8UYp+OizFRQQ/Ydbo5kM1g2e/ EkGZvDHLFtxHPVUJRqoTFUD/RooaXwJLXNyj0jnIwPV3y6/IQ9gQwxZapdwYoFGS4vaI00PHQuZg g/2tbQpOC2upeq+PTSz5B5D2P42VbVU3xuQFRuCKYjl7FozzeWqJmj6i370ixorcFfszOAdOG7wB cnVQfK7E7qwmvOteRFjcfZ/V2v8Uivs/+XW3pVthwubmBMtRPu+Cfu71hYCpDe3/kIp0uuOHokvi u+QhfiDyEYwuS4DzIc2U5KkvOiVJd+Gf10PXUZFbQgQyqqUj/TacrwtVei7W99o29ELRDeaiGgf9 bWc1l6EMrNU87IwBULNTr+gYj9LhbiNLWFPsA3swUS+eR7P6XDFS7vF1g0mZdVJ7uka8f+fFVvyI ppzREHK4nSePx6fVzNR7JqO+e/4eHDN9Tsnb5JJAIUDgKc2gMLt/001aXzexa+lpCnkcXdgK+h0I 220AXR/KF8wvKiGxEDJUU/3UT03UQVBbs/vNSZBdKaiVf255KNdeh4dzSZGlcwQAqAFTSgEeSTFb /68mMtWJgTFvD8kbfDVjM9OGyksxs0GYiR8utvB9jKxmjyNt0F6LETMqteRVr6IUo/9Omgu4pdSk f1W9wv6MikjesqlPVZboCpiTq5HVcoG5un2HHNrJ1E8gfi+2h1g9PDN0H3RC3/w6ZRIaCKK7rnWa TNFjWw1yrmSSA2whTkgng7fbpk4UkJjyIFOrX8sYA68NLGwosg75Y7cVo8ItUjDGxBeMq1a0yd8g 6RQdoDMB6+l3WOMfZGt+UWAsUD6Xjg7UMu2uBbVofV+x7bHRW4jOqhI1XGiP8sv3yhyElQbcPaZG bGjpA8ghId6gghDCcFXZH3fTCWORojX8vZ6oY+vGquuR8S3FjX1W7UkTZ87/JWSW6rrXMfFz9t/h X0wvdd1hPdobDhTgXX+xFTeqehWKRsX/+5E+q4KEWQiCTNhZEYY8WdHZwEwk42jzDmRPLhDjRf4Q Wlr+Z80m2V9IOF5loxo929RqkPJuRjXQl2U0wsOiY4apez8Lr872bD0QMgto3/mX5VkOWebTKfrg TFY5Io+lywemorSApLYXP+WufpP18DS9BTLwTYS75KqKmMTrt98HnA4D1/fCEYtu92wTA1HLVw7w VdGxpitxlrnU+kKfj2YI8Pxl6fGueBgA2x/jqTlvzGLgnj1hKKUz2u1W89ujnXcwDod0FtDzDfBZ 9aj/ObWJS/sPgAybP4Mn0BuI2+pGFtwPpxMuEqTnRX6qhdfDKNS6c2mxo/FEuJ+48iZb/5BlbF2Y VwE5JPQTOJSebtXSxpBRKPP91TUJgJXQlY3oZK1KzyNZ+LGbqmXQmh8EiKpUdA+vVP3QMhcyFC0d JaurbXTwPJkQe1phSIYD6HYeR1G9CRxp08sSRpBs/umkwMPREIq0netJ+CQtlmItJPL/bIb3dZcj OKu4k2wtHPWfQjAiynQfPBrJO4gCr0k4M1RQVtLfUiMVTtoZCZ/UgTqJg6IGLC1iM8CLbFavg5Ih XfnQbxoYwI5+p+cfAEfK1hOVq8XTIQ2mkCMPAcUEKUq4nmOKVh7n/x+fYvQziOitugSisj32ywAo e27rtsKH5nEwUripSu7eNqtIq7VUvKaj0bo6pZ6g5mG4y56PxzYpvzcwG9EzRQmFTxfA6bNHLlb7 +BC2LmhCSGnfc6Hzo4UxGzP/au5d3iCak75KohFWsuUGSN8R7j65cSBf79wd9gQxYAJBsUMt3H/a iJlnYbgvZJwKFOb+mcpjDF2F6T9xvJ8w88YBiGqNfwV2y9ZRuS+173XGpMkKnMX8uaoM5DgfP6EZ C/3GLHY1dJe3XPr9eeanWRrxdhuBr5+rYuumxhi7ErMbQh1c2c8PeqM1oM0QlExuet5Kerj6RvYj vHb8/1YiqV5SLLhsvaWSS0e7arH7VBTOlY3CKBZFUhj/X+skNccP6X2pXAeYMammNl38aL1n7yQB GdPjxLmBQ8K98oeNhWHHqP748YRroxY7swcdGagSVp3Ow02aRjOTxFPT8Ify0osujGThxrjT9oPe mSjkXdDrFWi+HK1KxBAhet62Yu+PV4NR5gu1B89cgnzPWyguWvuCxA6tARhv4XL7cGMuehJTcZ6n 9m6SrBDrpaWn3qVDpQeBglJJQJFQBlRmrI+vLyz3PpXSOjeVJW+1p6jK7MANCmVuZHN0cmVhbQ1l bmRvYmoNMzgzIDAgb2JqPDwvTGVuZ3RoIDE0ODYvRmlsdGVyL0ZsYXRlRGVjb2RlPj5zdHJlYW0N CqZjX3uptxGB4LmVq9ev5aT0tOVwLbU8QOT0z54NibC0SRjPiWl6tTwL65pBd8pWV5OhEnqgcRlh RtNkS43BQ9UB7j9JcY9R0jlGH65sudxb68RsFIMQJpa286qsQ18WyEDASw2QHkh3PAcI0nbrClnQ hVnotZ7hpNrIhqouBR+fxGumKNlSCHtl5GO3gwR2nlzpa3gS/yyJK8cyRI9k2zlQvoKQZXbfm+Gx djJV03f7Hyzyy84v19PzXh0zp8mD9JsLwwz45sjLFH56RWy3zjmMd3X+oOCP8m41qnqedgbUFMQU GvEnBDniUp6Yv7AQe13ux9cxc19jeLBwHuhr+psKszYua7L+PORuByc7QKQ+bgrurr7gTOd1a2g0 kXgnF3IoT8k5lC+eRCWGAfKrpNbSs/giwzSc1247PHPGSPZLuYhvX8Md9v+gEZlDPst1NCuJ56bS tQt85C3U3vy8pOy9tPAN15wjGFuyb54e77t87mPl4xhDyi6M/EvKxcmN20F6rw++A4ocwd/PsvNA qho04ddfwGHMr2vKYmjiP2C/z93MdQ3b4WE9jZIpVfwmPV0owGBIV45HpddtDvB6Uo9dji+a1VUm k0JTlA0fyZsKZm0HymSCYvtts6AEtK+MkAidzjx4oI8y9OBzh0/OVbpR8q2rZ+XBJMCrZBL6FT8F udikcTFt9vXQ6qkZBYnNOPJ+YV98NCNRXUiLYRvOus/yH9Q1hC0hC8rL9C8bweNtH1qHKMzCG6p6 s6VNqWaO+QeoHujb/0CftzW72cDuhhfXtcABs47/Fi0ESy4FSN1ExIologNKEVzHJdhDNytj25gf C7ueA/P/MnQhtUH7UG+s6ihtzCH5mVUYikwTK2zUZk828KGcVfrg67fNrTMqAneKrZCl+CRHUdox Ph0KpxfDjV+zvv0V9UJQi0ofxrcFPUit/bHUGLp4vWf9rcZF4HJD2kLRid5yeTIlhqIbicYapSoY uDvdILprVE5JxwbFrkco0nUw8WfvKjvTQYhrLyc1fb8iuI3yOoBYgamFBUoiaLtyJY1IBzJeJNC5 VqoJuo7y4n2h28mw6HbF86HwgU8hc94cNuPJJsUJXBcKyufYG0SdvP3VODKlT0YwwXHUXHUQDf/+ JJNw2fYh1rWSdoS9bsRjYueyfYiziJPkczm2mXlEYk0n7YAsXacJZz443U+nPU2M3L8ZqdGwQfuj SdLpVIE7zD/sA4xzQASoGvBtHKHqu8mlL0LzOizN/2F59UcTpLpN9pZsKyKZRTgF+k3XDDYpMg8V bgDarHACL9wzGxCBv8FIpoOeWATw/mGm1MbFZcfdOKI3XiBGRgmOmO/Y4wox1u8WolwsFCDUwDs9 YS6HQ6+Cz+00T6/W+vVrh0qN2DsweEdpMY6Dr5J5jOJyPBKLvEAwszG/BEHoYgmvGKZcxE1CRCtS NJ2UX3HlT64Sfny1eZN7JaDlQGDeLaxJnsR5bFw9r00v3P489LEv7zQUNKafjMMP/HaRLZfkVIB5 Ry2cqGwS/scpxN1oskYb3xI9HI11os0MVqwqagqsVutTB5msYN/UG1SFiNl+W7dljwTOeDdVu/xL q+fh58bZaDbEaXGpXE0ZTLCeKDoDHnclvldhIc8uSPVS+bA5VHhZw+P+gQX5Nznk+qBU8009O6VQ qjxOKyPo1QSTON2pUZ2fUwIJsAvDgIe/DkCQZ4a+9N4MvfCtznCMYjtGRhhbdd07XWTAqHaiY5+Z uBDj4VY6IATqHCsd9ALz21VuFLXbWxodPyJopM/pDF4NFeMTPBWfwp25viRJbLhqAOZ7BgJ8DIa3 OThJm/SChSZ69PCFXv0dw94qjR/Gmd6crCKQZJvT2yku/hy7wyheEf29E9LibTzsKIDqbxWrM/Ol MpGyOU/jvP/jmsJ7VHxaKTZrhiW8mz6cVidlvoUilGmM+XnKUoqaGLPS+k83pQvQs2+sOr7Mjgwg 6Wt7hhgNCmVuZHN0cmVhbQ1lbmRvYmoNMzg0IDAgb2JqPDwvTGVuZ3RoIDg0NC9GaWx0ZXIvRmxh dGVEZWNvZGU+PnN0cmVhbQ0KlQnE+g1pfQpeQQnfxrLEXl6CztpYJqfYE6+3LWBpN9CPAQi2/Erx F2iuz2t/qBJn1HP5qhdBm5ugOOsEa8c1FKZ1BghAsF69SYQ9Xmluzf5GvNnE9E4lCEAms+NYbqpF vC3CcgkS8xdyaBjQ+pwWU/xgwARixWtJ/81tWBshwkJhtuLdE3IP3OKrkQuMs6X2y2rIDd0t8uU0 pNJvvZYvCnDQG5WMTTSBIbDAPJnh6//NbQFT+RChFCmIq9/66u9H54PLB2xqfiEmcldpQ/IpYXrF khM6zrS+NghhQPypMYvm0ohjlcul9gskXJBCeezdRQQlrQE2wrB6JiqvrvfhRdceZVW6kRjEgJP7 riN8Y8V+4lJ9tz+RdgMtueToVXpbUKB+vR2T51JI6gcOuETVxn69qszHSPQ+vOj1uvsVLKyf3xBa FAFbcPnVfLz2Z+lbikmXNmFcNQDuokzrluM4VLAc1W8h4S00hHz6yt+RGLVL1im0sDpnD8oCZdMW /imxHcJQzwUVlF9PaOLkDsFLg+dbu2qVpuBf8KnTmOaDj6dlKywPf63CBGnzJ60Y1qjx5y8VSedq EFxxABJqx8v3M79JNevN7W8EusR5pVsbYPi66buuIDw5Dt8yOvr5UovGKac0t8ngFD68ws39dRNB YcvM1okUp6GIjt9j7T1KfjSjc39qi+LzjohR7vQV9JI2/DXtkYJnGVUAbuzTCVXtlAh5sXWZtw17 7Spn4/w4GsmP2GYsqDMjJWbhYwLG9zSS5lIF1wjKOIPhLAKWK3OxwTbsetlX3OpMzJisOlye7+kT 0zYIw2fm3U1hNff/wo3VCT4VEhb1n8Mlq8x3ydqWqE9BO8FovQuhoKv6piCKbUgxbBW3FZMprXnZ 8iua+QT7vn/6GbxKHitqygNNiJHY+TfKLGW6lrJPFxa0CQeFubnJi+MRlSljLPdyNOxTE5ZDSALj Zjy0d6W4zcp8DeRD5RWov4NAKr9dArF3FJI1ySyWjtWQvAroiMf/mBoaqaRDqMkPHYrmA5bpGWtt z9feGrSxjUub18KtT9srG+ws98yrEGy70QnUNSqg+oOFtm4oYmpRJpsbryahA+9lvJx0snO/aBAD RhfrN0tS1g0KZW5kc3RyZWFtDWVuZG9iag0zODUgMCBvYmo8PC9MZW5ndGggMTExNS9GaWx0ZXIv RmxhdGVEZWNvZGU+PnN0cmVhbQ0KFSRgcwgQEw/j/Dwgvaphyoa6JFBNIcS6n6Pj09mydkWp5FKv zUg6DxId9s3LEC7qBA0L+TRd2P2D8GhCSBRWdKCrh9GBL4R3QHm8EooJ3BvCWEXDWJKtDSEY495F tW88Q0B3F4gmmJ16K7pihdvUON/co3haiRJ6DComrio4/bFYQceHiTWtpzpyokQ0YZgdkfEIUaFj QzGptuF3lC7lvKV2i0VZHdGzZWaltR5WyEmgmaTAOJqBFO01alRxboEb89mBfG+WCcHdTNlc7TBR wHtYkIuCDcu2IAIz91QgcyqRW4CI7f0/r4i8ELuRRV01ER8unL+paMUutQoYaSKStJMV/i/i61ye yPbZSzrkJma0r8LWugz6yuLQu9gmKw519NkCPddLmFOBy8ScfJyExyqPC9Nqp1Ioy6x2KltzPYJw 3o1ATQfLeZh9FxkKK7Hocl4oOXMnsK2f93US2ay4RdrHHTrfdKwDnFxHsWSqIQr+0wKIJIsk99Hf uLBITav9sfOpq8AWi/9JVLWQWrznwLjuaOBGix6hfXE9K9nRfNo40PYi8btaBmse+DlgtawrRRte TJlHFGhSsIP576XMQMtucf9VfYBVTliWCoGqgwL9sj0+OjvsDhrAXDkYcvdsVH8CIyfFFXu7rHKg ustGU6KRts+6cXIIMCNjQkWAbW6VRceizCcBDfJ9Wm4ahYmHZOHHHZF/UU7xalJr/mDwiYwwhAiq lj1/6kqVIhLwr6X0nWXXSg10mvLTioQegFOdcg2TcZjGw6sfq/dBkPkreH7FUeo2iFcAVJn7Ogeb QMgBSTdhSAkr67mSVW8tUt9HKzvVGlpHV5yoKKyWjeOH7aig+sp+PEe6LKzXwNK1GvYx+b8FCvSo 14CVxqX3wa/ySx4lpXl6YBm2OBcfrOkrk7+lMZdnLS0DqmKjtKzenh5gEYnvKxH0Vzexth09t52e aYR5kn4ebgERV363BeWp1nqrgCH2MCvHydQDk7+sekeQyhxHCT+hLntoFQQQEcoJq37Wqb0Q99bJ itswSxGuNSGKPPZo7J4uXCL29mIviO14Nfzh1AknrXrdE27jQIIJzlF8gFwQPqTtJ/0EqMLJkX5U Wv1PCrIGJf8W2DICeqZMelFcbGNhkCy9JfYuS8dBWDk1kRslYXKRfUnII9ezHZtAzVVoyhzXb7D8 /wYSZrW9tkZFMVZ6hhAMhXrNk0QKJgXKCfhM4aqcH/OmTm13K/aFFuFT6q31XfvkoaDx9xAW8Oml 6RZCUmeTNSdR5hTlO+dtRgTv9bcs6hK9Kd0h87mJ60bZN/9fHd+UThNN/tVad4z3dltxTxBVhXmm wlwHioBQ5jOwnowc3fKaBsnge4dU5UH80FWyxTXcVIRkS3evGOGGEtdNsZ3ZS4nniPiW72v/RL03 pfTnFgK9hAyp91bHtvtUPmyVPgMWboJ0IeXGwyyizuQTIdmXFq2AelpaIVOfIe6nF0tVJCYNCmVu ZHN0cmVhbQ1lbmRvYmoNMzg2IDAgb2JqPDwvTGVuZ3RoIDEyNTgvRmlsdGVyL0ZsYXRlRGVjb2Rl Pj5zdHJlYW0NCttSJqlQRtbDGwJLp44QPCa93Hh3PyxaQq+DETDpHMm5cbkF6oM3JQzs0ynCQ17W t1gGGELrpgL80ZlFIa+E3GLSdnqpfxg9q8mNPtz96eHQZ/EzeuYzHBFd0d0Bv0RRchdIADfxVpBC mDvc9yzhkukBvU7/vF4qDrDdbOM9MD2H9GPsxzpDFjXXsyf8cC4w+zc2eDl4qi9hjlEgTPXY5ulb Qz0VNGb9V9uFeGM2bbAWMeVUaw2nATa95LsvMv5/pcllT4n+KH9e+yFt5H8xKeY2azZWUYnkIlHZ y0C5zACw411oq4Yfv5Fe902spLIMEreY5kI+CL+6at/ip57Gf8U0984Co4w7buH83JpjFVS5pFEj 2eL14BELedmqp3wIlfLMVs2Wtm2tmzj1PjMeUdmvHHBOwUhQjvxfdM2/7uvHBxAhu9KZOgYcLXqJ DW7TSQKuCmjyGtf5ahuQS9J+B5d5XJZzKMvDTQ8iK5ISXgM3fnGDJTGsJxCo6IKt65y0I1hz0yQM OtgXmfI834EZWkWJsIKOKPf3yMum30DCOvmsK5P9wKFSxXH6NX17xuXK2mPXfripZ0QbOqv4FBrg zOi+er7b/WJGrw75iQepH++fnBGpU7UINj8QDTMvnseZYU+HNpgIjX87WmWcy/SacyHMC2CFJCrq AU7u2WlRRMIjbPK/pYGCMucfow51yebXuLfx/wKokzeCqujOf09t9POp2nwQEAfgO6GFG+rKuCVR CdUCpGzfNG0h02C81V/zfvyj5Y067CyIdMThJkSUsJKjQiDcku2WH571XWsGu6bSv2hG5kLJVCWd Yaz+L6hJ29WxQfdOUrsOk+CToXZo1pcMP714JwJxPBh8ztYPuD4n6q0U59Fb3y4/AYmuiEPwraBh dK6OgFE0+9FC67IxfPhoOAVqO5wXaA2vlJDV1z+fRCQlLJFNTx9aRDs/B/oUmzRJPGHv+ShVH1EU QY798xb2eKfNdvsK4RZcEXjPyOWbmJe0fOAOArToi5lrfNJeUZJVeXYzgLFHumMlJtRhmAYPPanZ xzAKsw9KrhIVSEb4aMJBw3t916Xg4JixvHW6SAC0YKnfhFk9YayokxUtNzac9EAUQQxE3D1TB8hK /NDvgXe6giuGM/KWK1byH/L0fwLmjl+1AoOxcwzTxVT888WpBORRLI5oRmD846w5EzleyDP/zESx zd/UEqWrmt8PvnovOByeok4M/c4HaQp7ZNg2p4o8on0qtLmaTjH/tMBNiMe7tM3VGxrFMRwQjW8/ WRY8oOWqHCJQtN+x+7LSuLs5Oe6ZZpgCtBZRmwiYqGM4MiPTSlESLSxqIvRhm51hd1oCk+cJBX+L LovxqjJ/HO4mbxr8pFEr4ccgIOHx8QPBMWylxzIcsUirloTZlYgEffCYqzQ7utgIbLqKrMrBRXMs JKh9I7YSlwdR3Nzr7wk/2LCkv8zlwd+UCtsdAvvs3goi4yqoaW/JNk1h2sfkQ5j+1D04RSQC4C3e BWZtcdyoJUH12INFXfLDsZqCbDQRgbXSNvXO0Gc0MliKyH60hmdzi7GfhgS4Pvaqk5k9CuHvTmYJ 11CdU1wH6kDiiu5WwvoNkslZgGiMM1FHtFJQ8+DCU5DWU4Xcgejbu6X09FX+9j3KJIXrjmFOG/XE wcd9ps4/FgZTN5puDcUNCmVuZHN0cmVhbQ1lbmRvYmoNMzg3IDAgb2JqPDwvTGVuZ3RoIDExNjIv RmlsdGVyL0ZsYXRlRGVjb2RlPj5zdHJlYW0NCjOXmLqGh4cEEu6z7N7sSgAhD5vG4C+oG5MRUaYP Ho1wdKmBnUvP0pgfysDOog8G7BADYlfIZXfP6vDRSwR2oaFfuStckVYa8TIWA2Go7SgHGSRzztft PlHRn6d6OZ4VMnY3NNRn9FXYLT44ayRkvdJ9xriU3hBtiBKsEyTGsLme2YicDniHdiqf90T39MI9 +LhFBpMtOT/goHE9O6Gpa8GEL0tSdW9XVwy3fFS+cYIcyfcUdO2nXnx1vHwSWuKqIUqKJ6z4NqZ+ KuXSDad/ja7qBMafZszk6cS0/zDHGOcxTRZZEPNXERIjxSMarBTDO8snkx4CF0mNLHCnQvdSoc22 fL5cZ86yETP+uPXI83+lelDY7gWULDtch9YujIKMk8OsJTSeW14XLm2QoG7p3fiYj41o1vPrqNkR BiasRbr2E9kA3BTIUqNM6uGU0r/l3t7L/joR4a0MkYq+r9VKsUklCiNaK0TTYKxo+qiOVfP4cm5f 26UK9RSA/F2FUi83j234U4ALPS9M5rJXWiCcJ5QxlikmWVxhiZ51RzeYJa9gRcYcCw7+FTlEOTqi USr4VP/SZyGuSNcMt2owUVOQ2KgJeB3CzMHlP19eOyNCmGQW8fmWvABjxgKldc2E5nNLXYSc96qP Nz4dKnBETTuOXarj78dQUBW0w9VTbyfEruELlXhi5fVVpXxc9F9pJGfOfOmpxD6sa+En/aHUA80h 2/u3uxhxgcUmf7CqfnjZjyjH1Cew8F0m/npvGKlwv1Gj03CNI9xf3YHGyUlwu+JM4W6sWgGkmln7 XnQZaTrapo7y8o336Z9kddKGSy+3N+vPQWi5lftTjPgOMQ/Qqgjemlpd/7IwKQhXSSr34r2CCBjB PGD1SszB1NmNWe9RVwt41w1dICCTN55ft7idfbIiA2JYtuhJxucQsxFCBAiIxLiGKaj2bUNibA0a UMd5WFWj8tGGMiJ/1tQE5Q0lV+DYZG7z93LT53DxDaWQ2+WCeaMABSxXVtjW8z95cFdpoIAhtQr2 mMwTYAKaZYHgXjeh/i7MjVj1ma9aOAgv/FWQAnKnK+TnuwP0kxcfZgVIxGgn+M3a80JqIWoUZlKU ojLph2OSWThF1UMFzMvYATUweLqjM0mbFZgilnZYnuA7rUPIsdOw4w5VQD5m6M21NNXKuwp5fGgc rmTEs1eWkLnsNd5t3Qg3u1CMbR6H8/LcTy0KAn6qOsN6ANzYK8ioS7CXSzJTqfQ7ABan3roiF0BH fKga+UeitiiStBMZ2zfiLhfayCXht2L+rirp9u39Urk40ynPKMe+thYI5N0gB1Um9AR3N2dK4MtF +hUOkluTXGGZ9kq+go7JZD+/RarG3yFCjxp9ef3mPIeEQLKYB92W8uEp5s/G2O72WmOygtK0Tzis qUWjo9EmzceTmW3vMOrrtyzAKICSpYx0ue3yKs5mFvgeR1IElDttGSgOAAle6vnr5iXFfVfn8OaM 2zDQH42Zq06wpQ8G2s/UvqRBLnMeS8wvfpnT1sao6kAhiVnd/Yl9XJaK3LsFHy4AzIQNCmVuZHN0 cmVhbQ1lbmRvYmoNMzg4IDAgb2JqPDwvU3VidHlwZS9UeXBlMC9EZXNjZW5kYW50Rm9udHNbNDE0 IDAgUl0vQmFzZUZvbnQvRkFNT0VCK1dpbmdkaW5ncy1SZWd1bGFyL0VuY29kaW5nL0lkZW50aXR5 LUgvVHlwZS9Gb250Pj4NZW5kb2JqDTM4OSAwIG9iajw8L09DR3MgMzcxIDAgUi9UeXBlL09DTUQ+ Pg1lbmRvYmoNMzkwIDAgb2JqPDwvTGVuZ3RoIDExNzQvRmlsdGVyL0ZsYXRlRGVjb2RlPj5zdHJl YW0NCvrZs2jO6+uo7AxBvktwihi/31Iuyrfy4yPohasgVgZ11M4JSnWKMBx5mImsJzYK4Oq44eLj yRRVKdcPsPYj/4VgaMcKxuedjfXrolRmzJBr8c972kaHriG0VgYm8Byjk7kubwHs3Q2/omcsjhtw AGGF2iPA68419/9CO9jSN+yTDUqc20P3qiA2Jzo0Jq9H+++b5rgNpcxVFd51PbN7Hu/GIyVgin6s 6HPD569AxuzsJLWMvSEiWysXFLdzYKNizk8+VqeULxdvKkArE5AB/HlKKJI/+cYikVSxavaloM9K /ApG03MYaYFrYyq1Advzs4B4thh61G0lewL2Dl6LeKCW1cvuyoE7gGdHBWpeZCXvX2ZKuyT4sDCN l19d55FG+ZmWU0s2NDLfVxNLhN4ICtmzcnKZ229ihlHZQ+1jBhwsJ8D0V/RWQVa0XUCBuGlVX5gR yfaK1vF0a460Z7Gbk5zAFl0Z2JGrJDHYPksT3hHok3zERBBsVKy/pvfGqxCNAHn6pIfN71Y9CAoR 5Xa+qr8Q4EnoIjNv8lLAdrph3y6Klh11CFrs4g4UrgHAvlyi+BCXpwm4kIkVB3HiB108RMsMfCHU tS77yTF2foQBNNYJnhm8uwZO68wm4vuWjBSAzZlaTJFvK1exNbOZ61Vxz1ga7O4AJtP3Xul419nh qF3ya9iq/DOpPcraEb2vyGigJQ5YtlLZTp5yKjAotSukGdIrJ82/Gs5Pe3SWp4YW2vXfEXQierRR 1OiKjjAyVVYVa1k11udy0Pt5HIzOoReDnz6DZeJIPVBSfUluuqp7Y0xdoH7MY83SzxZW3G84ZUak NUxnIi20awBC6fdAAtZZh6Z2kuu83L6M8ciiwMkpolQ7rzIGOw3uLlYsLnTNQ/Ena5MrF+cRbuiY tjnBNmBYWQNEwd7QNs+eiMGCR8+pj0UzJj/ZDRDooKYp2UhiKTYlJRl40DFwQkUYoqXcgYdRN4We fe+6rghX5wwtcSa5TBhRF+6Db9uPgzATQM5G1tk9/cIZExUSsKOAq2fCMwEx+e8RhZf85FLeEIpy jsAEhUdgRegh3nvuxPkCUtgdT2D8SA7MqW//KYdsQFUJ88xZv3+rj71NYU/4i3DRX9VyT5+FgRjw +3/NNgHZEakt/P1DA43CGZDY9SDbjOtfmrjSUEhhkgX5ppjAyZJdUpUFJ0oMt2/E/Hcd51RAKpEO rUczcav0AXvn2fhOEF6JvhPFZxY1Mas4fX3ovahXGQPQEOKAqM0ioriRV77TkDZALaSDsj8QytqB 0avicVjJnFCbElnxhY2B4DsK5DBHp12SwmAhpkBmZWmIMn2+1j82BtkMLepX9UBtXbXDgPz+oHmb Z4lLPKXZ4GKBMZpi1LxDF6XLfuxGPivEiNsYRJpJm4p5/4I08W3djhwcqLL58i+jWEJ/vbSLE++q e7qrdn7knBOYFBRu8XJq0+0aDqq+uNPBhita3DHw7UXniXEoNwvhNvC1p3vm6xnjx1qpVOrvI9md YYLoGLwhh8GfIX3Rr2QPqeY/q8T+TNYtlFaWLujLxDETxaJAABoNCmVuZHN0cmVhbQ1lbmRvYmoN MzkxIDAgb2JqPDwvTGVuZ3RoIDE2MjUvRmlsdGVyL0ZsYXRlRGVjb2RlPj5zdHJlYW0NCiyyybja x6ozf6eC/E/AUexiVypI7AdX6IDjqUqtc5NK/pYkQ69e+D3dStn00sWKmhRpa/B159lkAvYPZO7T wrWTUbYEYSk3xU7HA1yCB8SyXMircYJpUY4tk/VlUyeuFwJ+Lldnh2DAd+HO7sldYuqwEE23Sut+ lapXzHx7d5fJm1jwpjcTl+ruAnYVODtw+5VdNqZ6JAj1OaEmmK5nhHeYivUNTi457jbVtM+tbXDr pzj0HxjKZ9lHPm6Xy2Ah5/OnoyDHI8H6040IXSsIVUHjJjUopyGR/7EkDen1oCFm+B/dBSOGWjks ssEiP4HXbRnn16jrAkYjNRqg2O/5u0grYregj0XVzggLev7MuzMzR/EFVqqfS199x/wT2LsmY/x3 /+BpnZHlEYHiGTP2fXvEk/6uyXVyhk8NWICdNHEVafbaT6alFiaClafwrlN+OXkmHvCt+cGG5T4K DCAkupHFbsD/ELH/Mfvo9ROsvGkWXTVEdX54xfhApIdViL0HfNdIyXiGrQetti+UWDZDJ9Gz56mG 4i9SK3oDUIsjRG5fZKg/f04MMGYjojTdjNoRGi0LCXzWsoZm6tQdGkyRf0NDTyG3/3ajGWtiG1m2 mtxELqPoJRoP006OXL/XJCrpxdoDYT97C4ytHq/zwS+XfdecZKXjU6HsCrxYaxXmVe45aPaoTPQT ZnqKrzDu0hPAaZe/p1/9FmDjwtDy8QA/gtk96mnVbYz/LULIECI/9rq3P3+srawBK/GoxYlOAknr HGMgDsK5hTn7+ze4vu1gIZRfh+Gi8pJGclNEl4f9iUOtxz0+e2Q+h+MxvoILx44MItCeG2rZkfhN SR7eD3aFEN9x73Qi2LnQtk/N6mhQAWP43VTiX2LAa6kEPYzee3hQNu2xMaDq5OBKCRvpAJeI84Q8 HdE2+UrsVWfNmfTwThJA8nYyRWi7VW42WuVyV+jnMSFb9aWAm+PESBVphX/ao6diM71VA6oIOBTS 0bjjfWCGjD+VLmpwgSMoAxH1LEl+jHnfjVDBps00ZY+KqIrjcS3GLhrSWJNI1P+v3wrYwBX/NStA EJCdklwBpDnQDjQEWjCJEJUP6GCDlPtS1/H0/p3c4vDxSBTq38FI0wwn/lpN1PKD1an68cySN9pi RoWLUBoqey/U/pTg14cJEoHDqd3J81eDmdqNLo/GmlWX8UEAuJYhHHUNpUK0vbXoVHA1kb8Gm929 8ulR/E7XowRmvQJv1hLVrBxF6wCDi4fQ2QX6HShj2XAQ9/ULDslE6TSohYf8iEm5QZD7IZfcLvYx NEq0mGUj1c42f7/GiiJDD1h+HkFtmErTENbLd5KfLxicTpdmdKdDv28fEqWYVAb/1kYkaFkTKbV4 P/QscbFdvHzTaA5WqRfJAn+USBBe9lSP6CJu+jlxd1nBW2BfFPfazoSta8vE1ZCD3TSfrAr8TC47 5wTSW2pG6y/Ln1pDEzBcziu6qqQZrqoyKkQnPqPZoX+sAAWY1DFQR0cCCbameXz9nsyU1gyqiD2h rRwSIZqK+tMe3guoFj5gGcrgPwjrMLw60nN2XHLZA33oZGDy1vcRdaOa2Pp3dOrRXJnsharPsUvK TV7gO7Oh5S5e77xFgDC1z36hxRP1kPN/pT7JVtNia5tRGopfKsduvDXWV4326CcjPV3RiK0OH+Y3 7dZYIvAJQj/xJauF1ZOtheEo5fgDyTu0/2cP/qy4N3cAuLDbX+v9/+a/yQ2dlsMyfdrhVvLINWSI gy0PG2XU1C35iFBPffiC5aLtO1uCDXTDp/4esnWgeWwJJhOpjlKUfQnPvOUhxFvbts4zkOk0af7N Ij8757O4EFvtrg+JWl4DUNjSKFbAp6kcMXcRc5Sc/e04CXPPgIK7+UMh8NxYTgO/+/Io3GaQkF8W ti6bDHlk2J4crfGaxQpCa8axICuBfYOY+hT7k6VZOcpK0+RcVCXPn2kkNvo4O57Q+AvodQy2tqQT gMsdSYcO7L67GiMQEGFDxj8RJuM6FfFkgKHCdOMM4v/+89oFJXdD7/WZE501+N9unZMfLAaCXdoQ KWFAR2m9OiRW+Ci43bgh4Gu799k2y/1bqWbfT6S+2J6cCRScHa/Q92hr6YC7kYDvAPVC8xiYjT+q 7UjJuwRy/GTS6v+yhxA+KxTI66l1ZJyVDQplbmRzdHJlYW0NZW5kb2JqDTM5MiAwIG9iajw8L0xl bmd0aCA0MzM0Mi9GaWx0ZXIvRmxhdGVEZWNvZGUvTGVuZ3RoMSA2OTkwND4+c3RyZWFtDQoi0XSg zzXmvq1k4YHBiBmk6jlaId5bZnFUXHOmMkNXVLT8+JIpVY8NFuE7STqjPouCIP/BdBVTS1k26WP5 g+SLN3wPOVrZ4G4sD0wvHfaiC9A3Esw2y2grhfcy1an1q+Xdzvlx9mcX3ZeFxXq2bele0RgFnMhk 3fsm8mLRzttRPrLDm2f0dkvnTnl4cnLpNONdNOgoSJQ75W74lmJ3j67NJ7sSiPmsAByIFDjHWT2g /+9GsFzBePRfu/2eVKk30Q0e4er1GmmpFOxBDCVHJUeQC0OLpW48HVdRpXrtbfvpBQLCzDbFETLM 2bIZQhksUA59pkSAmAt7W9E476FVI0XYDrehRefUEPUdY9TvKS1QhqiLmm9YqImCh1eQv413mvsA U/WTAYGGoKJAPbRT7wN4jSGNwmm5DJomMXZz54meNyjZP6M9abzMYXr/ezbkXoEWCITvdRlnznkB sP2Sdl+yKa197/9g8ZGRn2dsnUsfLWH2MVIzH6YJbfUBIKEXohKmpRK6Swehz35MHNnrbnzQr1fH AfvE3Xzi7UxGfYky2MF+lTRqrcISXX/SSjyuf2P47mV9Mcw1fUUeCjA/yLyyWlEsBaxWwSHTRgU6 V4QwklYSQu+5jP3ftKuOhJucdVIa8LoVfOsnSu1r9dIGgkGRuCpiF/db8VdEtivLiu3miQUTTUj5 CmIOI/T/YGflfQPHIYV86NRVoKy/l802/AnzZXxdv9RRbuQmAh7ASPIoh+EGilcdU6Vrrpd97eVH 9WgUGDZwh3m29/YI9d3lwLiux4sGSO27bbyemLOgTbmTEX2ALLld35atz5ld/cYfx4L7BciPcit5 /FT+D7pIi0g+Q09ckFizCSZYSEsRkI1/2PuTXOTs15uF1QH03dbKW2OcLGT95vniu0JkaWSHOOxJ m16eTBY69vCS6d9brYL2uHjAQsuGbqZy6CXy2ZNspttBuxJjoX03ox1zuPB6p0FM7LG5Tmiekhgk 6GpK0cSdquL4V2SFD+HBj/PSzKy3VM4W6NL6VVAQSh+VSbC+s3UNp3E4TcmlJvciFBPnknJL1oLX PBdBBUBb0MFjxKiZnM69lKeaXHwET4dGRxxHtPQ+i9MCEFtQpiHCGZhvtKvDJ438hvevbhreXrX4 kVoSC/C4xbH6U802trU5y3oxb88DCJ7+nMVKcMVmBxYrexaxXEorpRaGbtGWk1IhH1lvobyxPtA/ GYXv4qUDH7MpS/+hCPePQjpeZ1a1zAOlWg+UMlHgUeFwFMM9dK32+eeCtozLt8m/EEqlp95ci2QV 9L0Ad91TXLz0snC45J7nm5dIokN9IqSJixGsWPOGhMIgmwSOySk3aHmavI8/tXmqYgZ4c9YUrBtr Bkxm+YYBpGMk/yRapdrbL6gOvO/5AyzDBGv7JPL/+l/MBnUPbnjVrNXgrsIM2zBlW4NZnB+3T9tc 22vBlToy8ncJligW5kCjOJTRhJVZpKQF/j100vrKE0Io13LdeouA9jzvQ+lQ/C2c1xbPFC51Qg7N 4DfVW+tnM1H2J3fxDmSukoxvZfsrBn93AMaf2JCPjUQksI140yVj9a3qZiBznt2gTZwYVevPEd6x f/hJ9iraXNI5Am3jCqI5DJ3eLgDte0Zy5b7k3QM0L+CZ3/CbAFut9Z/I07EQigaGjjPchWbHWj1m DVnZGZPWS+ziKBl0oOig4qZiiubi6cedauV+MVs1fP9of8B2mzMHh55dI336E3z+e+PhRrzTBlTL vmdLHiB8il7azz9IWszF0LV7xfzw3gV4zJlVLEuP639bkiBE37KoAISvumWVO7svJOA8tql3Oar/ FaDxarjLmx4h7pCVcepTD6iLrSL+JWClCJzoVJQSClVgqJgy/55cUsunXSuNZfAEkh62A5+mOETy MkpWvdTS0XGoR6TF95ydQNPEYB1L8SpK8PhEeFhWGKANiOcuL+C5hfP+VQnw5VC6tHt05Z0OPdX+ ynrXNeI6v2jTRBBijmTOFjYs6LTiwPOEPRXYRhJ72uCQI+YWQvNcIju5EUUdNSmTk52SvvTkM0fv HTk/anCDQlw6O55UX3Ap7HZBFuM3G3iYttcJpfBBQMPvASA2FpYyUKo4e6yKz8dsCXoS5544ralt 4emMgOWWmHRIqNybCGkXfhdCMafnjS3F/Kvug+zysU713qoNzkO+2donOwYXEYzYqV3Wa11eI8/2 A5M8WlLuK2l2LHEDWqxirwn0h4V+/U6IpDogQdMR3tiXQN5/V1xkEWEKhYzkIgfqlIzTIs9lprz0 BXC0kyT51zJAEXkFeY7wKYSiWhFoutafLr0Ka/mhr3E+V3kMDLyh5u6kDUF7PJJBKwVXQDuIXt98 4OocrFA2b3QMgR5Ua3RfjijQEEy623eUCEi4grjmWgTTvoAeDlgoI1+82cwq/s1T9XdthneB8lja SLdeQAA31PzHDLqTHx/S71GSNp1HWWQEbHM0cvU7Z7D2nQ/Dd8ZgxjY1zUgb0w2MlU9Kstmxr2VP Xzaku9x6KEupxHsPNDeJ85ixhlatX66W9VVirvclqRQTQTOtmoI3DQSWCuZuz3J91vJs39gCUfcp 7LuSBSR5Ez6ErDpnqkxuOxFPW38Q8xeBX35nQvaMNNgaUgpcpl15LFQZxldVgutaIozzjmzPfjcN jWGpNteMxg8O4j0MmpRsj7BXWHvLWxvlKZTurbdCBWMXsc6Il5//EsQ7I9wm9R9fR/N3S3y1ubtF vVk2x4Ay9txnbTaTPKCh03MDSOtsmBaWEYOp0Y4egnROzOHeO7daUMrAN6qHSqvouvr0K8JjaDvA vQ0hX8IvIzgUnT6uQN0SdtStGJcor71677mA6tyBKacHYD+yGQtnsxs0HoRETIHrS0a9RLwrHt5K dFVYlhNyjAVOMBiUxmjRfSMviYOraDo0JM953wdT5aPcRRmcjxZYSDZuNAodYcLOInUmM/8PoKsw 4+rd9atuGr9+DcikbbGAuo6ZdjkZS3TWOzMA6IXjoHFrH8tFtmbqnTIw5LKFszCHa6Q6OmMrbhZU z9BIAi58VkkbPGLJQDkJgQCz5vL4maPEw8xznuDkOksz49FLRg7j0H/NVq3+5v6hkpb7ehHDFn5O iU6ao1Ti6z18Q5+YAszIcWTQH+5Gszvf5X80DoviLLEEy6tQDFOesY7SKFrkz+pY/tHW8wmZoxTk 5/cOqmLqT71By0fC2m2BC5NkapCk+WWCIC9MlwtUV8M2gqZOfryyYLskHmiXWrvnREb3jwzMD5uY 6b8IikxH16zKatO9lPcZzMpPmuGAuBZIa1Cc1Cx9bHwihtJQMJIVbeE7ICCRqMWDYPFiZliqRnEq iFKd5wDkvueLH4FSebAIm6qGwNHsfB7QGdQjKdDKjSbjbCsC2LR5JR1ROTGuspoyP1sPvKqSB710 sHWsa2RQJPBQ2dbOKOOc/vwOo1yaP3f/biI5DY2hg4ks29StAGJrLtffWPiUxrdT3uVGpJ33n9ci j5Uw0+OkNu0pFgOGX5q53K0i0x9kgrsbOxV+Ym/Vm2aR9KCEftEO4WY6Rv+WobVtj5s4DtwLAiXy /HUILcv3MDCVNCULAlvXP7GxtXeBc1Nv/5LJAbaYuCa8PZn90J3i6dHngnkXpxnA+ILdfcBuGonp QGrxUftMWKjf+61/VoOD4kg4fUftMFmSonxs5ehfXkhK5Rsg/Tuc+CYuYLJTibDhhn2kVup7wOt7 bYG9Qf+ShU4edTWuH/wc0+ph6VYZsk0upnd5YCZzaK8P10K0GaW6wABCwuupgghTIqHjUucWFpjP WcWD//8cSfKl7KgKsQl11RFZmAu1j0+pPFgt4ibDn3HXQnXkQzn06aTL9htLlZU0Aw5G2NXwyHb/ 3h4zJJpaMI4/RFrqnLhnEwL308mqGhXnnXKlQzOWyGYoRaQ38OhhA/fNhQ2r8juuwu7oaQDTZdj2 jq5k6/iYISy/Ycb0OU/a/H+yLlCAv+3MIFMkWdtg5Xd5HBS6cHz+mIewGum1wpRh05uHXeN7+7eU x9fSEW/4aqrMZrxshVHomHb2XhOcpuDMueuQ9UaArgjtgpFLOEFZUSRNhudcUzNVnrIskNaF6EYN /vTKSn9b7utK8MGMeaxTIhkSPM6WybRbUSAvwLL6x2mepiyCke7mtDjk11wjkkd0RUAPrgRcJN3I sHU29ocYDqE5bnO50XQH1u5Eu7FaSUwbvo7HtXe07fNF4/ueDF4wRAmAnFQ2dCOsng/H9ugyUGjx lvfpWou0Z5Qznizy9XxEUyfYihcd+IHyBujrpPLENWKcdpXXXR+eBQV2frftZPQjWz2hy34dw3vs 5bGaouAuvM9sAfxszokRSUkSDecpH7PAhLoZU7wnAZXiteaVVy7TPucp4f6B6vYdxa3NuNJL2TFd LzPE9pfvENiT8nvciUAnsUutqPQGAFC4eM5H3muHdpV9s1D46RPChfw8M5fXLPpZpjxV2MbESnvl dUoja68w9Gr/dyI1LkFeWrqsDZ6MkSKN4L1J2UmdzxwcfTMgZ63FKNI4ilLlKQyilFkumwrO7rLy JtlMQHZ6R+9sN7h4FqRaGzWNtf9nY05F3Gwdqn8XlJkFX9RQq2n45s/a8Fp6lnqrhtvllweSKYRc Bs4iyslZOz9w3MbMwVKwTlKXRcAKG14qA0lVAM68agpdnq7z7QKvsxcgvA/BPWbAy/sagELpjjpP BLUxtXs5CD2IAb6h6MkZxYRcUcgmHaVmTkKPzAt8DKxW/TcGFWX1/eV1H8BjE9/3lbPjad/YTxpj Z1oQ1Fb6ZCPZwKnfs06yacqcJj7TB3H8UKdnb66qhunlmRrFBxSuQrNO4uF9/nrv8kQGo6MzZ/Dq R0XlG17Lb2QIxuiodwV100JakQb7yK/2u1/Lj+R1DrHFxpzaE7NpT9EJE4FIuC1sVD9KORnw/rCI BndYb7bDj0L1lY/dUchLWsSMp0QJiMHnSYn6rcCzXUZsiLlPeL0iJk5yDw0R+TIG6yBZLuxryMLU Cxn1vbgut5eE/gl2dCVRWNvbtgob5JeZqCxpS2BOK/7uJppKb+YuPwH3Xipjk23a0Frhqic+CwuD wMGptfYH2EzNHSF4G9MJNBbAbokSqSOmikZDKHueI3YQVqFPphHH35kjgULTXMr4vWSIBmzTefCM Fvaa8bwIfA1efxINaEBFC8aRyDL6zIBIhTHJgDvcbp1FWva8zolV4hUH9jPTfZeMCEG7OXYZhj2p hVQnCyMK95cVtYxS8a1rgogV3/4X1uOubc9kaNkDdUvKVx+b8r+OeTk7Os/B5pDcXRYfThf+qCN7 4oS9sQw1fZVDa5wmJuOu+dOMNOR+zcbVS3DG4Iywy8BL2VynpksHgg5zCC/G8VmofMnoLsRBGtxX WatE3bIOXj4spdDt8rjEz0JLUBsMClZ5KX9Z3OQKRHlrLhZOt3SLrqLWxoDfk62y++mfFXO/pUMQ a53QlmUyOo31TqRBdAduR0Ukg/F6yuNX3XR3DIhnND7UX2SbfYVPpsU44ubBbRtCBxumLrnihuC6 e0JkIUUi68FKYKrN5OjgeXBmFp6HwvJ0l/qevxIXiDSTsu1l0ESVotpTqU7pxjst/+qWCLzAAZi7 TqQStM8Id/ZLJRZvz7E1nKDK3DIEkFn2sGXatxdFcPkGrslayUl59BI0iBL3rvIOw2xQGykWFGqx GfIWA4gCc0hSQuRyJ5GbNDIJv/+7O/dnzSpCcr/v2rLnwUG3CE+zsNpqKAH8oIfOb9Xd7EgpxMsa RBhyABdpRhBHFVaTPmvEMg7aWElvYigfUbryWm6J/4pZGuX2z7Zr8sI6XZNdaErcP2x431dspNVU zs6jGKXSPDKK2w+76DWjf59V97T5fwYoSk2K+cvV526KA9G7DQQ4Lq9GWfrm7AQkOYno7qtPVkgP VKNoq0VNUMkNHr/9SVSTnMn0Kn9h1lwL6hwo6oUvHDOcYVc1ZqUjcz7XgGhti4jpC2+au66ZT7kl VWGQ+7Xz/sFFZHkvJWvttzwdC0uU0iPjIt7L8ptqa17iP3U7pu6hPBCA9nkIxcOxFLcphUmOf9H1 gaif2y6AjLOf4mkpxb0eVejNWyFO9FCFy4vulBE9t8ZQju8sxwygqSbNbpbqWdbsVx2k349yuptH s65bue9oTgRFXnJ2rKuYgs9RZ5IU+oLzCduSkwp7mQpORaKxGuw80rZDrQr57rKZQIcxNDkjV/d+ agFpNwKgQotkjucvsyvwbyRpPf5rNJANQcdCSqUcHMOA39WmcuHm7JUJa9b6+XE6OzjpdL2yGGeg IrYXG9ascqxbB5JpdBkieyYnIc1xO5bDWntp7X4tNMro6cNfXx0iELu5hCjMaHRDXsEKiSRlqEI8 WnH+h227qKHYWlLYb1WwvCqR5oVQPIBu6D6dxMmmoL78OeBESpUAjAItuw6S9dKBzsC++EC6KcES pPmdoS/SO9Gzxp5sGa86isCXBhavWFdsGRwJpLyhKtZaE1tIfIGb5l2UDpK3jddb2SxFAdclSfHv +qHwRsDYVLBHsewnPWQ6LvKYG83uCMNcIj7kESxvY0HyqlzigjilSLazfLNAT4zAqBw0RK5NH/FR XnaFGCFOT8NtnmAIGvbOzi5zjw82hpar34GsVZGZ89oxpvOM5Uu5YmYHLtZ3e3F9FUkDWSA4Hy2q 1blYLqvAUonduj+ZoaIvZwg7BkQCMFwYAMJsTNcBjVCDsRfKq6Okd9BB3O+EQ82qm8aevDopuJOb b4KwTQxJWUwFhOBCOIYoIJiLVfYaNiESFP9k4H88F++KiowudOf/6OHpCiYvCGblIDrfDr4xAK/E tq4YfqSsnx+VYudSYW+s6zga0irbg2IRvJahSYDE4vHaAS9cUs8jU6F105xK/25n7vA+2rW1CtgQ BeOK8EXQ+JB0mScZqRo0hi7OK4895CML6u+qxrz/NYJkFW5s2MUmo2hFitRuqLDNDXLx6oerbV3C 8IN0C7+k5aZlgZkm4hYDwuhP7sSuREQP9Rt56K5+34Dn6O9vpKRfgxAjGR0FXFInCzLllcbprB7O iBLAyL24MgkKOkP/WX/ABQVlbVMdad7lUcLm4Hv0WQBnKCoqACqHYiAwPMaz7P6E3ATU55cCHrux 6l9rj+6DmqOkURLlfFQRY0LZEIoAE5iLaKTu8Vzi+IoLdmjZ4wMZtpeEvODFzvBWKmtBxSKZHTby 49rcsJr4wzNzNFOTaVFEvMYOIbSJY+4GXqvvHdMm7ySQjf8xlIQbt+8zjBKLuyp4+YoyGTDwbZMg W1gPwW5afu9gKQ3P+32OxB8iPtMN97c3Slvg6yRz8iaq9siIjxB7jeXlFTPz4MjgOhjojOessp0B LXEq2KCmI9irc93anG60AZOIk86RE53khX3rMUcur5XbpWEifJJsxa14qWFGSd5o22Ii/WeUgaGD RpeolcrO49bZXw6N9OnbU/LdGOiZU2KLb4EXfQZdcPYrxn3Knsdvf50t2hR/UKXQb2SZL5IoPvvs w2PmHcpcPEG0POmmW5z0f4A49zI9b5jQ5Q9Eu6jyDXSypLN4uOg9xxACa2XVZ/pgF4RKZNibB6z5 L7tU/yrQIWz38m5TsdP61MAgO/BYMvZKcdKBRbmnORKK30MZi3HxqE4bDSPJfTN49BRDUcpWBTlM LidEuU4CGUlGmamc6MaGPlFeReTiw8KB87Xw6u78gF5ioYDcqcVri1skI1Yk0xpzrOwLcjm2VKmS auNrzxsnHaMb2h2rxF7iOrKB0F2cKKSdbrb09zLNrk0V2bn2jH0+8NqwDCRmzujScxDH8Y6Rtav2 WJwu7WuS5rg4AlK3O1/FgGbRAjNIG7hmEdf72V/MQbpsLdcBFwhdGxHvFSRG9AER6Tzrc6KG7vhI 4rRiE5EYKz7KGpg3olusn6FeYUExnDuncWe1+CgxlwJ4MEoqGEzU7CsYNeR1ntuhV5C1A1OxOCFc yGZWjEKrBUeFOFY9DEuaVO0x13QOUc5TmBfw7eIF9wKgm/Cv5PulgnDm76+LtcsMt/5z0RWnOsEj 6InShknFM3dDKIJMMWGfKcTeCdgJjMWE/QOknqHpdg/O0zmS7sgLFDg+cwudLjTb/MuPmwWPCQje /zxKcw4UzTrOxGZbliIpk6qRRFDxtbT7NA+dVcCXlJTR0XVHliuqoAQ39uVGY31V15tmyoodR8GK yn1sZjeVJbIo4Uz30ZXTcMYAxDYQt2PvUzvZj+sVLmaxZTYnxGZhg3gfNpD5+BBoYVZQEP0b8fVR P9MveHpfd0/P/wTf/bpWHtIfxa0QYDYP3ppaPeXY6u0VRCQBTzUU5x+hP2VcE0G83SouxPLEKyTo m0Gs2hZe7WexJ/xf7kkR4d8mr9RnHktCilomxrmDqpzG5m1vloLeL/fz+kfIto66+SzbKcg2lPRC ICHYdhCxksk567/UObuxtQfWyByPaJqmgTI4GMLGFA6OCo7IiDHKad4lxPKbPpKZHlxI0JlAb7nW XgzE1JRpPlXTt8v7XHOnMHzdyCwCQzkmXef+QDUUC3o5AnmwNEj6fHIPiTXKMYFmHBzuL07nKDRe c1NR6Mc0Ur7d9NjWVoPUlYLMlYW3hxEfTImThebQrgs3K1ovErN+kYs2uFIHJW1nPz71q7DvDo5n E73ACi3pmsIZ0+296qYBmdbvJJR19ZIt+xQtBSSCPN+4xY269XXWkFxHTDsBHKMabqA/ZMkzh67b chEnh0mcv9SNLsNe2rsvUWvp5N1xM8KiwxWooLSE69X/b63Iwzt5mG0ip9Oja2q/2zGfsa5fTOW4 AoOoQSa2Zzcua4hh4eqp/vaGbJ55zHXnt8gwos3lvibcCCZbAYyEuz9W9OKLl8ebpPGwU9wTRzHf 1FbAzuoO3HJIJrEXmBXlLYHe7lSaVZEmhUedgAUSux93JwwVziYKycmeMrGkH5pCH9XT/zsJBYY4 v6+UHpOL+d7h3qWpwEPJayEu+jDVRrfutG34Lbuay31Cmr8RV19wFBw/Z37z6oDwWZ5cOkX7nttw bGkLSS4RZoNzn01+auY6QjiDo28mZhSR+6Xo2Pul5GiKVXqtn4KdwJrsTPYWavg1gUydPtRu28cH l39pkCMvUVaAs0vYegFAwJcwP4EyR2o9TT/aj5YpiyD0CcsYfCcTqns0Re+0ekIIPprniJup6Oz1 Lr84PK6zEqjXS/nEN5BXXzvDKp6slcqfR4EK5A+dF8/+TgqPzXEw/26k7J9PDeQku6NA3DRiCK/9 LgMCzXnFumrYVRDOtiLPxw9IKOsB8Iw9a7c0j36+Q3JsTPGhI5q2ZZnGlJRUnLGq2Kg/mmufNPtF lxIJWiK9S4LjTS35WRHeOY4B8aDnE4fRwFNroK8uJj6GdJjqah2jJu1R9o+TGFOPpnX/RjD75Ejy 3YKS9v83ftKXXUOWoUjTJa3fNbYjY/G1ANlJWFlBnJ1qS4AiQcFhaOvrbv3niOR1Haf2rELi3fMf mbUkmKrdE4w+PjXhWHGKQPA8S8NGFdhfR+NjgohWyJo6KvulYh1jBgg1MjJn4CMW1yNtWwTKDYNV 12Suh3ypI4GKQVwgfISoEg8XimEM+f43UlVERQt0YybCb9zpmMx21sgcXHGi88aiw7Ulik17ro30 yWcd+8CbmcsRvCGnpB/GVNBpGhcSZ/eFSGVuUsMzsCAUUKm8vEwMlRpU0IIZGN9iune1Zmc9XH6p 2y6Oa6BrvpVpVoS0IJEDZnpujyjXI90n4uzOEl2+EhY46vpR07m9YFPGSCzvOjAvSMnjCHc9vhjj Qe4NnS92O7ay6egdxhgCGeat8m7MYUvqyUhkvTi0v/T4JrZ30mmxp34deniLwFV1iNxGuDtPPdhB O4YqsJxxRcoZwwGZf5LnwjqH15qgP3gIxVeKBqfVqcVprO5JLk+COtZ3cr4qVHWS93eVii8/4uz/ UM3vHwoFVpHMHpG6OJnvPdBSTvQvLBLZqc6NGxvLw3URkDnhRrx0i5Z7glcOCYjv3LwhfZ7W+GMZ P3PwI0O1xa/4uHq1BoCzuOP4J/EfzE01aya/iphkWin0U8r2UMCDAvaMPESD6/YH+uUoexmG5rdW yL5WXZ8jpcPHGcSrm9b4lLddtLs20gvOVjOvSG/40oub/5+GaV+zsoLDfflBg8iy5trbielI/WJr WF5bzJv8DwmkspkjJLjchYJP6xxWBmgyGen9v73kFpvJGKlmXPQBhZ85CFspIjwyFJoKQi7G4W8+ ZJI8H2UeiqDqHq5Z0g7EFEqfg93FQdIAkFGc/fRzQOZfmjQs92VEdZfkOqu0OhIY0IHuh0mf3zOs nwuOGdQUqGdCvPM9cT2M4YhlqYt5k2kO72kveLEvo9no4jw7NbZsBz5gvrdAi5STzofbxKS3DWjx aeqGuKdMGYuwOZsCaq1LXetUmuzDRS5E/MkMhe0QgMVkjVl8YJ5p46AKdm88Hx4vtkUa9NyEcoiY 0INmPHmlNOgiw35VoqjvBzFAbfW37xy7OSF/cY/ZHo1RgSfMVvtNueCLcj1Ld7aAvGg+eK/PSO0z 8btdlwAUnU1Ty34z9piFgOoc1w3JGuk32l+z9/+EvWiL6GTAcMke85PJpdFJSzo3g+UHUI/MMKXd vMzLwJf20gXamKdp0fwWMSPoAN5BPU+IJObVcjclqVcm+gb6XzPsEvhNaOiFJj9Fdwrpdh7uG4V6 2Rfd6QDORwU0d/Muj0FdUb13d3gtJY8vFi04ypdPni6Rm9Ivk5JMKgLAa6SaRnDb/1z4rMM6a5iY es4IZNaCgv2OOo+qufOTQrm9rDmWfbMInqnwPY8U1RCUMV744VZdQ9KE98oTHJo9oSv1+o5dnPCv CtrVcrc89GHGG94BuUEzX7mpS7duJyKh/3MGuYrUwizOe/83in0gvuREWv2RsjkPpGvGXTcqWN1K w9zMVuOrIF7c3VHvvLVRFrXqgqeh7Jefnfoozyl/ISIP2Pnr4UGJ7Zg6HIi7gacuXp6SBueEDblU rEUpSFpSTGD9mSYgycsmcSw+ACXBwnxDhiPfESUwejO25mFd6L+iGrwQPQ6wXJtABMRD9D8CQyGU quQJ7ZUUqN3e5S99u0HJsFTi91eeiAunSZ2bcbTHNjTOI7kxzXpgDeroTXqxqVBPBvy7hhW26l14 /Fsycz6J9xW12yf7Bi1iXLkJ2Li9l5V3W24xwYdFLarHv1BLG60iCKtuJgUyk0bz+R5dgZWaat20 rMJlZ9BymLVUY8RJFoChXD8zQxBxQH2vkFDaf9d2KJsSceQhjoIocWdAaPJ2hpLQtw+UzhKXE0P0 ea9OezjC5HXaTmhJN/CqpdWrb3nU+9NmseJwXnGbvQjcrsOCDIqNZz6ww7SzpvD8HsjKALF9RBf/ xbolLmAxVf/zaMLg52BaLrlB8Jxn8sdi+yVzS5oaETiq9YIZtEgwpZFjIasEGvrmgSGaNR7bsHOV GF05LQWqTibKUid2gS5GnJk5OLtTmdG7nBbAXD+grQrWlRNGHPZS5Nmds9qd5yAfSX7tOiWXOUbm XzseKmiqxkf0q1/n0KAzrF2eDWlf7SNGI/F+LA56TP/frbdWRiR8sBJnBu73/kct/9Qn606iMm3a S0iY6mgHowbGth0yZYefJw6Kpxc1xvJfFf6btTHFh3KnVdSvxbhx3ckv3gm0e0DVQ4hl1D+g6mL/ /Oja0FANvuzuVwG2jrcIfjsmk+Ga6/QRiWfUA2OGY9sS8/z8JojLvUdpoSaD9mVxFNnQd/SStKKL BMf6sKE4t50plWEtcBV9UpX6G32TUDzyzep+ZPJUZ/nzZlugzokdbGjstkxlkuXR4XHk+HvUcLTD d6pxnS2iZJKu/+ubUS4W+leNit2T40zkshlGW7ROCqH8V602iTaFRu/iMdlVTmXzqdcCZqbsg+cg 5nu/gB9re0CLN7dT1IIL4Xmpa39fINp6wNeXXuthZPW6JAUFpurFaT3dpimRlDjDFHKIW7cZmM6D wzzBdSLKvKUr/tc7P/sJb40A+WZ8NIzkeF/FbNsQ/6bAd2TS0cfqORVkVFNZCQFBvoFekw/CuSlx CwKEVXpUL0/WhBZTRBpzAG0TUddP38fuZOTAYDir6l84Qf6j+jQjDUg4Vw09ZnqH/jU3pW7A5Isa 3rLrmYQ3obgNso5gD+zJK1F+i9kUQXRczbCUkPAdeW8D5WmpIk3UYS/bWsq30K4Bx2N4PmNl3vok ljuF67cCg2rKzJqLUrlVU+o5MwE+JhprXr0qnaZvQOS/UkxJyLUcqRI1s8Fl1gMSlaw73TOJwtsS xqcpGgppYnUiLFIwwde6Bdm1zIwg0eAn8ntvjDgDaZQwjSuKTKt5IfRruLSe54kgYoPnCGg4Wzd6 Y3w9XgCs9XN32edtBOCvGbPlrWltN2WBAQhk4P8JPxKVSa5+CR1x29gHUYtbeMuHIL3lC2/0H1nP UFJsc/F7slQBXSG9m8ruYQP/02IBefBZQr562t1VbYmo0UAv9SiP8zwT4+3gvOifc2ubZGrgVYKB 3dz164IgjkZwdgLTwLwNHzyEKX17qR+U6dbv9RlJwy89/A7brg289A4iCdKTT/wL8CeBamrOqHxT Zx5SXLMoldUrr9KfCj/nGP1Ui4wsYSfgnfgbU/pUAZyF3w6Otxu7Lz1EFu1Mz1FEWlXfE8zvbwXq IfoYxrIURwlrwKShTevoRaykLUGpY4+UK82FFunToGzT3Zu8/5WzNAwgP+OthAttS2VimN+mfEnK z1p7/vx9y1wyrTWVYyGmh23AzWnCMdeqfzUIB6eiJAMmW4gvm7kyni4qXfjYKT8hM+63FIYox3Gl 9oqzEsiEF2onJf3/o+y2Lmo44FWOHT9S9hdzy/7diht0PcvSvMlJ9xMWJUJUIThgIR//5tAa15UP uMln7F+ZqEvZ6bBO9Smgc67rtHtiC/cCA/ijdO6hnNECH6oBxX8+ws6HMWtkpIyl0IvpX4yM1236 DvreC48s3QA65xFbtc2yVdanNg2uD1cM3ChuckPVgACo0ykazeqDglZnVn+8qH84p9LMt3eIzW3x yWpQXAVreagDIfGVH/BUNwuZ6cz7pQnqzU78upLRRoBvOvoW20wVH8tcHAFLPDtZkDnxp5m7LOnr ZEGUSR3V5ko76lnsVd3TPGzeMgy5NVpR4XAxhIx+OlW7CQiwZ8H8Hk/e8QjNTdRp5r2LkTjtFKzi d/GSeHnfZDZUJPBosUYG6i/zsKQhUU8CxRMMI8xgyMwTcXMA1KRb43nQ3jVKXX+Pn7XM8rJ2EBCs c2rEhql718JPDRtvxQV5QyKF7ye3f1p9OT9rAd/HE3syNamKnCpF6JSUvYOasipvKyv6ldNp1vi4 mI+mcAn10fKnXPUnWdPgKpm6WmnLmnu5L+ITbZXVdsyGeJwsN51T5Gxzln264sc6WGbahxtJpnxf tanFX+ay2DPYsD3BlNzjRPPUEzNT5D9betXalhKMqNnmgQdppN013add+LNYEfFdzYi5ECLR0Gqv xzITP/VrqHkfCnMUOKJamoO3wkSeqSUUlFpSLGodQ5+vqKQ9QcHm48jKIBfk72c3mRKgy+DH7DAN SgvagLRegdi0mEfFw8MG5UNaQDHywmpjQd46aW2dTtnI5jWDFJSdZ//b6V0ks1m95dRNz3qxf94Z ZI3ryAfZybJrhBgCPHG67WS+95qV5KY3zKDcAmpgA26Fj1YiZouHdDZURRENWf88pa2JHjL4VXjC wVTw5hv7XyIS7N1y4GxcEjYwxRgtMuUU9/SufiHmlAEgUgQYTYcKKIPY1ehXZjzIVW+OrDmFCLRG Napr0DgYMkjOSIYKkrvquSR3cJ5o9pbESGBPJXwKXuhsqL3jccFE7riHbgpmKe8qpjBhPZQEcFeX DnYUPoIbl82pKvWDKKpF+eiwdptZ9YgmwVx42Qi1SMK7QIxvEKBaKVF3Zh7BbAa5JAXElQghhSrv pqTnwwmgOYtgOZPO1C3hXxwX7QVM+wzt3qT6vOx/xEKuHpM1kb3pJtDN3HOOQUdj+uKcRw2eTeHS 9u8V0R6YWvcAQIIbXXIh+L6pGah279WpCmcOQ57t4/kN6cDdww04iuZq0QXcqV9ZcuxdSI87aCKQ ML0nLtM/pI9ivGAuMaJQzpL9z7upSpVkKbRdsx7Wd8vBm0svguZiPE7Edg96nEs9Zi5jr/aTencU myEpFRSDYoPg9ehd8W5BGPAeQsEaZgLyughfJbqgRxzDGkcUevyNAK+YdR4apP+ifCBpHkVi+3m4 rnVgeHHTI2aXJE52kDcoCZ55dBAXhvUZUZ4BxZZ76Leo/LrQ9V9fqBaWGZmAeaZfSAbSY5krW2ez qGKGWUQ2hWBiTEdBsJjNj7Nz+ZOB9XF7LY2FpFHYv+ByOVHHiiko37gNbQAkWryqlma8MS9866ao Mb6vYCNl49xNEsaod+83LPkiPcwNUc5Y9BHDTajXeHsNtZzSJkLs8Fm7cEFVe+iT74675bqjmwuF VFtccCDe6wKvpW8vbg+SiZfHV1i9MaUR0X+j7LHfE0JydK6voqFvP2iFv+ar0Ql/huqLmDhfUlzl o6TlBNELDECpunLU7itGzb8ol9NAMzWREWJf4i3FM4W3PtVVJ/bLBnHj0+o6s4xze2f0XARXJA9s BIIRNDXXLzDnSeyRC20wQfNqFH0kT0y9QNY3fA9C2Jr503iEKmWbG9p30Yh64CuLGwJYgfcG2Bys u27QL3XuWIMG7T7zmwk9hWYtfOQNchkPukTTCEEfIE3j5I/96f9jh0p8zWAV8mDzK6rW28zpC8yX j8swyHKmqwdBjriPbF6q+unV1edSHi7ze5wxIVBE0E6W7K/OMdT7zZefyBntFhiYGz1UGPDCf58L LKrxI/mQaaH3LNT9FUc/tF7fitRu0+JdANHauDU+0kwhThqs+HvvSIezcY9B7tf8fAaS+dvq2Ewk iDcMP2g+2bwI0ihUNU2YHLBF9JVuWJFYnq1muPDPFSrCKeRGN9SRFAiLT4rlJIU6RssUEMC4cQ4d IcBQ7PvIn8jNiwwTTOhBwYxmjnn/Cq78Flch7RW7upwYd6MqZ4yn1F0MQAXqQV7G91i5wh7RM1fW k4wetJ0YmeTjUY+5UOenrvCeTL1mKPMuhRK4HuhcG3TKuZBRMKYDhuF/EQUzm++XeUw8N2vnhOTU PyW1JNJuVX5r4KwC3KkWZjNy98wFw6DVBtUYgz/63XLRy8EKCfBiTU6ASnvgogOiA6gKwaV7DwLo Ay36wtAot8IrZzZYKmv18bfs6NicWZddwgDOfJ/feUlPilJzXLMgHPES7mxwc8MTRBZcRYmbkHeB fIM1fmCODJQMnANLC9K+xiZnBMKdbR8TtMEIHrQr2Vy7eeGCwr+3xyPs31kCS/e65qM3GKunoedV /iK11MFt7L9wX1WgIVvutEJ7ieSyuZO7tViQZM96p1vBNak/kPlYpcoFBQcO989eCm/zabs4MZ2A SvdO0+anpE//52yuo9C13Vy4BRvu5Ey6Q17lxE2UMyqT1/BBUhIia1IdLMrD85ln12PheAzDX27n 9Vul9cR3FOSg3TbE6Ix5emUgCAhcTrLjqRjhNyL6cWUbEW+RBc5kd+BSOnedltfm+fyeFeRmxG6V WfPq3bXDMiQB+AXNEQ4FV8EbM5+QlKtb3Aa7fvh3y/Ih0+iFognz9HEZ4jIJIooP21cSdBS1PtYm mq7+pW0eJxEfJDKCIpN3/eKXtkEoI1aDJ4ccpuBdVm7ZWdE3+K8YYFvZsCMgqZ/1KZYtsANv62ci auhiDxgM1Qq5THmLakRTwtOMZ0ZawYiN3GI5M+RraPUR278c9/aEQqiBatk7+NXthVVRU4hNRxsm DmI8g9bnuqVzEuRY0uj1jTqyeDNqivIL1VbCKddVwqLmtwJiF6l3UEn8DhBOpHcXy391wT0omTF/ hCAKXQUndYMWmCkiNNuVPNpzSQzEJSqYuxtbD4nRWIk/D8VQ73hf+dHF+L4vW2r+7TDe0FyhlPjj fwGKresey6bY0GXPliwFzbiw48BXU+N6BfkaGlOXanQlBtE3yKwFGp801m3MgEaCsAnu1/WmPR6h ylpgEJFnVIL4H/x1O1odR2ybEeKNPZaaR0c/ln6yM15Kw3vXK0xPjkeGNvxnEO4e3vgvgyZZAvPT 8tAnJYAeNuQu43CFOonJTtggtrLietMzSyXaqRTUrerRyDv5G3zMVUImP7jvqrD/feLir9F/lkvz d0PXflX7tWJ0R0zdNTUtK9s1Uit3hI3itjq7p1PL+n9TRJ+VobO9WUJky4GACkcMTKJbvIXQcVpi E+kf0mDB1wVvLGKzbw1GZ49AxeKQPE0jO7FLH8mh/z3RRRLTMWmzeQBGGju1JZcOnVJ/VVrG3SgY jqrd54QHB7PKX+p1JcRW8O5n/i8TqPOuXQe+ZqADrFnMY8htms9o0565SDugfv+6P0DVJntjr6ar ZJRikb3gUBux7ScEXyeGCgp1Vuazdl9A8emv7SGkWZcOpMor5UYZfJYYCJDIA7pFmNdGaZLXzmKK /IXepjfalqOioD6HoLdRocwtMYOkxLGL0J7iWoLVfMBIStc8Tw5MCb3c0BDiTVlOh5fKJfCYXBm/ +1xGm5rtQ/5DhzglszInzPEflmkUH+PfXlH6HNh2lkk6grQe4ysjR54Zh1rlnEsyOE4ZaxRV0Bvn 9hmTpIE4Amzd9BCw5I3RbX727oBC4PsymF/ESGRTBI1cggQKoDg5XaEKqoISf5nMYV51L/M4mrf+ Sop5x9QWUhMjPWu2XchCsMvzcBWGzYcd5sLT/HOHJTSibKJ7RU+Z8tiVPSS02RFKfbCh4ReVxwzK L6OXUvzA+IVKmlapg761e4XNIV/1MbEV02RkklKHwHuv+cILrSNyWT7k5b4Pk6XI+LcI7DQ8qtzk Yni+HSTgfNqxzx4rWNDFeEFldUM90yJlGxGfoQ8PA+V4Hhkl86D3sMpl7fKfqboqLj4FeAPzGS+i qzOFwn/R3Jhpa+8+U0oqZSxDsPgUolqmcfpZ0ysdC+C9DgiAAQRuM2ysIRNRwUwRxnJuIXqk3r0B vIzMp9VrLOeMrrW2IMgx82tesOBnF4CK+LqzbD4S0CZ+pO2oS3t71SCSFOnwqdgRorDzj+xxUhce PIjV91XldBb9XWEeS70IH2XUTvnrlXYLr9+8ep8IZbVfAAjKPI4UzGkh8gC5+V6EtPD2ghEN9Kh5 S572Y8eTYNOiJVc0ueQrVc9yUFxegtpIUTSy8LuAo8BVUKlRmX0nF8BGNebzgu+SVijU2WCXe9KN p7v1bR4O4mzDg/JWImYmbPBoSeTavyR05nUqOWdmgBa4Np8BdvMehH1Na3LwFn8U8eR/I+jfqPAV BDg5CsaPRFL0RgzAo8j2tEliOTP5oeD/acmeaWBgAQoI7w3Ko2FgDfpIaeA9eRVTtX2yFUud3jxB tylcNlNABtVZ9WsFGUNwrzmyJnMpsr3R5s/7Vz2GzRLNTHsbUocCpxI9lDiF5d7sLmGP/mGb+ZTv gN1+ApbZrY1h5e8DH8nZSVPvj6cljWAN2+a6ypweyOMuRSrXpfXj0cg2iPLhJxaiG6aTCOm956qO qdXWGZQHzSN5dCrou4yQrsVYa9XiXmlLPp6qiw0XNBqRLnNXzFc8C2MP5YHx4vMJE8DdNJKVwgVl 1KM82hTUtMifp5MJcEFnaIID7PRvDPqFROA+IhuDMRUUN/yV54QNr4J2dnGsAhC2yEfoFy+fBpxY s+tZ6EACGxWadH5rd7Nmy4HPLCB459s5L/qXmP/JzpXzcN+LrocHJmhI8VcJV38K6Olzi+NBs5B5 ec/RLaihYnabdTyoARtw33y3c+t86iJV11SVLP3dRETPRxitUQYiEJawU4tUD6Wj9EXFu6Iy+h7S xJ7IYN3e5ERIuFChb6MWxtDmcojZVTA1djRNzert+Utvugmzxo5ldkhRznIMAjL0a76QGuhiNNUQ uzyCeiUGGMTw63JjlVgc5FquZjedf89F21F7aT3jbz6fN5EC3P5SGrmw7X99UKmDlsVHDDrD5jOW H9hw013dr9n3gfz6OmpsFAPquZxPO0qbLGCP6SrSk/ZEeAMYzpwmn3GPiKLVDc0MTd06aaqraycw y7QR52YbrAB662D4iGtRdW3FiH51uw2BD7diL8aRdIpBLwCIzXr05RFfIO4qPsyA3s0Z7LkVFvSb izQ0RDtQIMq53+IrUUxLy2/YMWfCe4gXTfSzhJcWGxAypr5KM6Y5i4iG7TT5C15EhdX9TP+GTyEB nwdhBPSF0aZT/Ff8c3CJxdX9CmecN4tRNpZ5+jsF7MgUlXl7458JRv0nxgwkWJ2Kp17wOpvB3SG1 oDNqUvYHls1SM5X+Z5kxOpPqBP3eCBUnSrRBCyjaIi2Hh3iluvbJ2X0HW1LAduNAp4+2i+2XCOnD XvL/BGEQTfJMTAsOpR77jsJHVQnQUGtYZzdINzRjYDwujTCrNhKhERa8uGkTX9HM0YrxLcY/2QXv ej48mV8wBs5FAK7oaYadJBPdei0EyMP0z+dIy9CdxPY076bTchVABJXnbBOgnoTlilLzuWIl9OI4 IzJ/HfPxJu/3Addk6f2yNQA581g0fd539MnUwxRr2GiCKpVNkQYJpBV2xj4ZmBgDAhLvP5cO5qMd dNAtzFkjXaRJuE2MexjjoW8DQE4ykOIV/4glrtBZj54mF6Gz7mB2RLs98Pr7qmAv6QNsU3boDZlK s05A9YoFbFK+oLEtxx8j63uVWJ2eio3joXf33k5cDNOBvzUycVMXZpUUFI1wVEKUg8wWRjdyOVbe Wp4+taEHDo2F4vTv29kwhXWW+96nmWuzq9BmwgtUZ6bXSlUERZ1hQu1kmyPgYScszKmc9WhIjBxN wGP+jOddRKWAeY3S1GJfxwUj6LnEWD7X/Ahl0CERgp9aUIcDfQBBlwfwAk/b5kfw4e0m+dI3wky7 cqY4fDutKZpBpXpBed2l5F5tN8IiEh0To8fhfCKIBSXMA2pBytsOAuDr3cCI54oeZQ4EmN8I4L8I vut7nti4vMMHi3je8ZgPtMG/QmoI3Ewgp9ziBdJI2u2m4EWMKULc9dMXvr4LIX2h3/pW2UnR5TSy DPneBYntHLDTe2kLzuduZYHmuEx8JxyqVuFb6thvw+SGotZy/73tlYS90WMNMhI4ZMLj6vKq8nPP nAj+Cjj95n38CbCZXP8dG5VASTgWKLkJ51m+byM64D6n89CR9EBywHHoVx8pPxq0hmNMXMX44WlN dQoBxgVvuhTUtgbnQUBUCTJ3hnbX0grKXRjBWN6wMokku2Jm0ap42Nurf6yoVOH6O9X7/QEJk8cB Tp2YSdJYliPgB3qzMiu7USg625Ao7XpK0azSe1buXNSfKR2oGCKpomINzwJegSrcskvvNq8X4Gbn NL86nrPmorzdsr87kGewzd7gJQarN009K7wOElA0XanFtaRjn1EXNorIxpz7JyQMzvJVsgoNYr8N CoHUzLtvm/4GzAacwddFcIWrtoVTibK0qV5KIeZM6fV9pM40sPU4tOFafR24tL2nd/eQ5PchAGvE 0FtYOnMXe9Kn1qfKi3Gl0gs2YyTBdxobf+TZcnYf+598OjzJNRWi77QYiRySWzuDkwIxsYDuU5qk QY4M0w4cXC1w5qVezlmAs2TIgEMiiDE+nUQkJl76B4hKui4YDm5ldgG2XGqtZv5j9GW5vLipfMQn cfWBNNCbCY0DqYYhoREy+chQQutSY4WN6ALi7/cMInYawj7RxBntfKnSZakYtPPMug8/Zdjx8mKL REXgWMf1QfWsKpfEa+Y2LHpAy7dxtZAwcEV/6BoYhgFB5vpsTXyaiL3GvIqjPaC33CiuOc/hcfp0 vxCkHHe4oOTETFY5W9z9s1Ti45DcaEsq1bFQ/GIIB3ev29V1enQNzyEV8Un52fg91PThK9b7nNog crAF42Y4RH6hWCupnLcJqdlOSK4CDoH0+YvXMGqWyMnG0RbXs2k70PDc9f05C75edwI6dUObFKIU 2jv+Ta1S/UtQusR5uZc4QCSb7v97DSD8qMmYyShpKL9TV06C7qk3nVHf6myA7tFdgmr8L41TVyK0 WmPIXrJSc4tpRGkHR/XP0bY0arlPzjf1z9qulukv5PsgLkNDA5hoX8RbjqWbi/9vX5m2pfhWpLD+ K26dw3ewb5jh2q2CS3OZbDkL1ZJUC1KmfPWvCBAhCVxBJQkyawW0i7X+Ljbks/js8+gPk/KOfTCE Y+AwJe0B0W5B2I0D2Wy2vv7Al+GGHAMMHzFA5t9m/oD4wBTRRhKIcihSH/ajvjhxxnVOmXARWQre vAg/xO33CsYzwjOb/vs6dZL2f8CkMRAuDjMk7uHAKZLb8MymqqcIRWIrlHAlZGdoP6dmUOp07aEd 9kZ9T+XbDi0PYMy3nkjKSHAnEjYP7mctujiYPTll54vi5rWucTuvTH7CVxwzEJus1gvUKQq1Qjvc lXQT4aRDpUbyPPLQNdV06hGZQ7zVi9kBeaLJ49MeMTR0+yveYl+ziJMRvq1zqY4xhr5rzHWG+7ve 08a9hjHGHLbxBlHNbwW1D1/OyY/U681CngrZ/MUp9kQSZVxkRQU8vpiM1OYjFFUHeh9/nHsMJKuP CYOawMiTAe6kZmrnfm3JfNpLMjEAAFKuMti5pvuK2qCLlL9CSFLPyBPU8+C70Q4ZvXLdwvjlBop6 PorZAuCDuvYZ5OJQmIqHW+s8gCAdvuc45xVwOFkmsKaEdYtlP0tvlpkQZJ7UNZY0gWD0PouPa7Cr ya4kBzkJdtM9MIO0DaCHHSIAzyf2FGERzYzfJuMkZo0GC0iHrJiGGDJH7sX0hxV4ax9WtrbTRntr /g+ssu2hoxQzlKSyYlS1SrGa6wa3spB2UXT0RakmhLBC1SoxNLOEle1XIvexTUTDBCAoc7lyJI0B 8eiBhW4on0yBf6A5t4mhPZzMzuaE4n7FTZZvBJv9yp2e5ohNe0sS0ntr/2iPk1dintVACGmzgn7Z pkAB+eKKLw2zB64lhm85A9fIE1kkR6UobscldC2opmaGPhjO5HECNWIBjTAvi6Lw/sOAk5n5zH22 ojreCgf4BpsF79UayVHCed8F7EnxejkIbyAM6QVB3zW17BLoxXmt843t/8YrqqKYGR3miQTa060d V1emfVqLBV01xBR6k8iHmNOxK/TXakTT4EsFzoBkx9P4/udHlyx6PmFkiX9cnpmCN8SpR19fbNQi pLAxXTuQ0lUyn1CzLxEhT2s3rCognMhR2vgGhjW/IJTaplC2GMUXT1rBbHlm+ckcqe3YGsBvdffF vr9qi7ffvRR5r3m3eVMZ24JntGM8cyILcsJZyrumRG0fr312CKy0RoC8RRYWkGIdJzlKHNVOxZAw Uekt26UDrqTFk6EA+3WmXWHN3lR14qh7iHCD0N3P4biEGlQzzwZj9+79o/BVWMC/F898zjhFPHQo uahUX8nGkFmpk8F3eNndc/sFdLW/bPzG6bJVUGPsuk8CaWw4JDopWIYHqvRDrnKqtgYlLohNyyOx f+mmZVNvsrR7YnxREJ8dIrmkoJbcNQT3yu/Isj6kKH4i4TTL4JfDX/3jUa9DVoadGNoNys5upxsW Js1Pqd56VoDRrCgS6cG5K5LhNdAmO3q1rgEaxH/RSZtfCfkGXAeIeClnU9hlbnQd17qkgutb8hr8 fS5F8b68ME3lPMBrjRLbpZb+4GGj/w5eFG/1rjj6u+2D5Sn3N0t4i2gYKRKBnuoVCaqQyys3VxI7 FVPLhHke4NvK+V1cgjtmTU3CBunqw3EH4G1YFkAfOMMTRm4H4jy9Ym4Ygm9C/Bn36X0jn9/PbK8c f/Kf/J5EjIkkrewdYeQHaN0C72XJR7CIy9O3xo/pCPD1gDGvXI4okxNF0FJchYKYKI+aNY+O2nPy 6AV5S8pKbFqxejQZJExCUhX7tRWW026TTc34/kDIGJGtI84+IfDHdvz1tHtBX0BbVXxfNWCwYdm/ i5e4PZk/POAWI46ibliFycY2SgPTluyzo7P4ya+EGrZJV3APbDpRUz2wKrj4fSXca6plZ+xciVTQ f4qNod8UVP6yCxhoQ5rvij7oImMxGqyDO76zTLlqhUXNJNkbErVE2QWslSVi6MhaVAtrfRaJeodl QfnOXgYQAKkfJqqhcLG6IDlWjO+v0zFZl6/BROXJqReWOIAdQMYZuV3aA539W3ntPbxIAaGjuvX4 qYElgX8Jo+q4eYk3uTLvRsCBWt90+H9ZNkNIwPGhW9kE30rhvgEBSDaKrak6e59iyWWYIb256Muo SiDJ0Qjj8I6Pi92HvG1o2+I+WyRB7awB8B3SngruTjCmorviWV58dubCr1ZQF13mnzgSY2NcM5JH hAW8zanL98b7ftCH0GtRwo6Fw9jYElex03bKRVYKKvdh4zpHsH0JYXe1b6QNdH5bGjkP5vrdjWsE 92bW3PyFDhj3t3C6OErBvqwCMVUtT2O8bfriA3Gnd52kWy2Rk1dPmiE1Cu3B1RMMto13lnxAw+Jp 6f6AgZIseETJ+RjiovPlNRUS/fmSwpQD/KTaiUNaNbRfnqdfkhOkqK7ebHlUHj+WmQ6+JOyyI5U5 3NlJvCLSjRD2aWvN0HJBX5YJOyv2oiatMR9rhHc6ANFbikX+wUrnjGYxpMQMRAMXrtVzSMdXsuAc mI5DGeWBYugwdQb0d7oEDc7rHIhzNe2Lm3mvcBi3tKVxiyn1R8lGFW5Z8+8OqzIGGpa5cmPvQcly 3DN9nmyInyLzaQv8ybPOKQ9IaoPdisrT+bYjLTjNldAVjYe89WN/hW+qpIAPExI4cCEm1/UcKeao fALS7XrANOELUh/ZPdUuw+rWEYdZZG6JztRP3/KwrjusDAZj47XcXy/Gt/nOmYfnsolwDIhuhJVi l9QNFkMg0ScFYcuWRrlKkJf9mLaGMDAJsVupUr1F8r3pmR3acRkUK6KwH8O3StocjKd/NRI8o0oZ H27XKF7pINIRzMs2t+aZJi1zMijpwmfUl7gxK2iIAEKrNuqsGf++4Yfxp5dgmlqYFHtDXFyvLGA1 JsEwIy2u1nm/svWePNkjhdkGQnibx9mGUw90y6ewOzBYG7o8uBMyXmcgcNoLsCrS5basM6wiIZRW MRQjUnWEWNjCNNsjaf7ZRjR96OoQ8jREZaaK+K2xM0yVGHt3EwP4bcRTPqGIwdkMERH6ZGlpN09W SRqe4tXoLLFKwfozaxWlAQS/Ea6ujWOHK7yCIZPYWE4NIHDYCtgq3ZpwaZTaqHmwwQrqSGmGQJaU ZI9G3U7eICxzwweFUU6/Ul/EnUt+z4J+iQiftXYoxA88/6MjWWYiwHfO9AbxQUcAIN3ykr4pjBeI Gd5K0ncC+wijTe90LTt0KTyhQaq0PGmNlJPutgBBEaHRk+I80d4XniVL5hMvR3LTvMm/nHTWS4sf q+C2U3PkErHkDEwUBH6EXN2nfP83kLzSz7k4WyoCPaMCDDHEOpq28jh2X7ni7oxyz+ElEcPxGbB9 CWI+c4JMtWKIsKIH8Ws3ssZSz5xkii19tM4LH7P1hRoEw7XmWD6/ToxDCIUITTixZdLAxiTtWDbv n+McELQ4g6W8LAEz5CgrW9OKr/tlM0ArGx2FWoNDfP6eI8N7HCCFBGRL5HHTbpVEhacvq+mHLTZq whGjrkbwsZc3up9UifVGtM6a4biEXbASF/6ecI3kEhvSuOhcbRMxwsI0grhhwX8n0nvS9FJyhwwt Nsv6QgJqdV2Rx99twflKNiUDM/hB0FEciBlPxzoEVnUKFcju04TyI+EAPPu/Tu4M6z/CZiCDK3mW XzDrh4b5J6cQmRAmS8lQamOA6BlAC1B0lHUrI2PGGh2YlfhEfq94R/JyRxZAO8MLilV2LKnPxWxt s0CQ7DDkjTcUurDopG+Lm1tyupwaJ7E+LiOQIcFIcpQDxTVWk7cjb0Nisa15I6Zwtnw0V7efefRq f2Ad5cUIePTwiHADqS7dL0hMzyIQICD3E8xpFvmx3VP6UCN+mtp3BouGVW8fHzLkGrGWqbU+T5OO SRch4eaKb98FPUImpSBkgQwnKkTaTbEHfJH1eGUfSwvpWNL8gpThsIYDYbhrPIygsXqxryBFZQ2j wt3BLicxnPoGMmjlUpgpAVkzNgGJiUI/FhJFWMGAsaYjjEOEnZl7Wr2UMw5ZMX6V9WdkMg5Izkb4 9qr90BvE62JEuKwxzGZ1RMDSWC3DmzgpdK8gUviHyHn+HX9AY13r4vpr/fNGl9HXIMeYuSeQzJL6 KAZ2Nqz8yyo94MG6dQ+M/5wViD4aMiKaVpQHgQVBsEar/gNSOPJEb4sYbxIBwTYph49SECuQA0Zu PzelA1w7vzBhfBg+g49HyTxj4P9thbUaXjIi90tUbTOnWMo8d60Y/03crJH6Wg3BBBCHuTQ7zj2C m8YcYcvqqWfcvaPyuHFRt3Bpi10t9+inf1P5kW7i1lFGMCxGaDc+/BaM7OfZaam7bgYpVILI2H6g or1m94JqRqhhCrNJEPGK2P52BMymtD5c54jrrLH7imBU3d3vqcGDeoJHBRQPpxWw5qCIWRJxBIY+ TcTXwE6Gs6UH5eFRJfZjsFfo2xKk7JoGAb1JONupOqdUaxD3TOiAJgbu/BUM+HEA0CA3fJ6efu3z rDg91pgnE8GJFdu1hUgh5dvUq02aBsFxR/HZ4nivUtv8A+OWgYGsElBqyShwxpy26ydqPjzztwdM 6MfUDe9qQe9LS8RYU6t87KA64NtCwjpybBA9v6agyyytAxjiKXivRqZpWQ5SXWzGyXr2SfF21Ph5 0d5RjpFdJeQOhv+SLp3p19Ovya4wrt0Fyu5/WjZLP9OjM1Pi6rntE3PfaPWVyZdUe1AokRNYHFUq 7JvD6uyE+qmB4vdm9jWS9yPXY5yf4TGLDVbF4fBw5A2ne59ZDr8RFUuQrNaYttPSV0pT62wLObd0 f8qLxN3AGynZu96iX+y2TByCbWDZWZjVIskJrEem6KeIq9oOKpaEPL0cP/3VvuiFVj9hGFXn148o gMhm3wb9SBQ9Z5aZoHOloa3H16Ghn5uhY0cXJ4CIbWTYEXqd1MSnIbzUEn4W5QdGr2gURC5xhxiO PisVDmzyxYcoirNVfHe3flxaO4EGjuvvQuFhBTuN5SGv3oupwcPAoc+PpOUb22w2vZgzb189SXOn yM/jOzM2FlceNf534bG3BCXetfl4hwms3t6ICv3N+6iKvdAi3W2OBIUA/GKBs7WIOHOFfIvkX+Ft XKDgzNJzwjw/Rah0COrH528fGIjxdWJbqWUx6E4v2AvlL8aItOHMpLUqfIqduLx1iJONeke1Pxzo HFMj4m8cQncL9/Kq5uR5UddNJR/l9QCfIqI800ioNvNtEvetUR/0jYQytb7tM6o8bVmHBuhy21sB BIaTmrUKP7Vi6g4AWpdnpRO5gW0MBtzwF6k9eoeDfY/mh/1+pxYQd+QmAyE5lnhKgT1xZGjoZvjQ rF887qpYFGKCTYHF1QOdTjPovYfwOyd+h5sS5STM6gPe9VIAW060sl41s1KFKQEMNT2VTkwN6lJd 84y3w8Oq4y+5N7FH1eRb25LnG9+dYwume6lnotmeIuzCpnQhhQceynNrShLWq4QLtaKAH0FjNQ8+ ywvSEZ05Ow/c7As3WuH/NKmAMBPxsi2mjuvCd+ML9Ub1rdtp2TCiYtUUjf1+ULtoSB8tq3+oPPac 2qH4jIBCR/PR8xrx2p2X/Ch2mAlWMvViacmVdGZlGXGRrhB9OPUlaRXaO4owMlkmKcEOCH/p7C88 FlwAUxkRo4iDQIS+shTtqXOY6qEVR6wf605ouCv9tbwvCpfPQ5161z/JoXfz6tCaw48o9b4e73o8 WvgmQfOyOZaiasDAOSZ3OsqNxUlDOIv2aJRvVjOpUZxAW4qnUvAkQ4goVenYHlRnSupO6o/hv/De DWfyjXLr7MCZBSyt/zBuaLMM4y6tRFae6nKaoCeZj9CD94ypUTegSwuyCVM/pv8QwfVyEaPxyvoJ sv1cdFIq5w9DSpDZQU9yph7bu8thrGEh93HVgHZiRffdBBvqVubbbrn7BoY9M5JsMPu5YOw3vbKL 42vKCE3WpAChQhsJNbkVCJUjfbWo8e+JUfx0HCJhitk0q5+NDJAYOMIQJRV9ewLBZSzF94NwcJtA 4XGSEJwG2FI7gZGHqIJvdfxbsmdQxRCPr3ORLSBqct5IZv3zS7vAZd9xdWNCAjD2KfUKCOJiHE7R K2CUOx/FPNx10cjdqMUPFUnxDLj+jngPFyaUMl5pZSne4Rff9r+A4GUFzI0G4QwVyuSwHqPga9MK H7dpcJKe1eiARArYUrG29+P7Ild2m3onADBQvoHjZ8zi+NeybnPO4/DhsfdqL9NrISOW0yj/ZXVF Eoi2mKio71Um5I3FYa79M7JmCN2/rPlnFmoLt0Gf6OCNI7DpgNcAhMxKVCqhPcQmKHSz8bFTbI7R 7GmSzul70izDvhJi18tFrTuYMVK9Irz8a6ZDGG/Z6dAnYhPfj7r/eaGHWN4Efpdwz0MK+B5CCpUn gmL9bRPcVJwpYX31bfpiAWwEjKwk2r5+dY8OJpLCZ4SB5rIVT2b7Kriykt0pxVMS+2+Z/xlu/spj nMsULtyzxE767QB23j8np3shVh2yDmV07lYCyIG1IR5gu0Rg3m2t2T4qumQmd3hSnZDCEr0WZ7Me nPEpWVE9vwv+QL30nwMffthluW/4nQlsASdS8DmTRzRHeeMGL+3VMCyZ5VkWU/zezDulh01T34SD 9Y+kV6q02pNpqCPZRViQ84fQHd/dPZLPEWAmTGm/+dgDl6tYmaXipD3/KFqe2D9ZNvX0BX2u0QDf ZqjjGo4jsjc5yjJ2h0kGc/mI7eylgDqkf3xRtzg2pkZEje/krpl69fRNRQ/nNAx8EO2bEmL5IWVo u0SUhwAD2wAi5xhGYmOWqYdSvC/WVHwo1w4e9SedR6z7hwxLeJTmSyjC/Llv2to8UxEOeZwy02sX ZUxsIinJ4Rd8mi4AlrcIeJonBzj28BDEdt3b985k4g/WkWVw/vtwwWGgBZkkUpke4z8DL9q0IIal M1/mYw2bsXN2vNAtUnl0d2spn6cnMMjnnT8j9ku/OKQVRMpJfyOuHi9Kuw625EHa21RXDZZPIqom ayh9KABie8q/9eYI+krs6Tc/35Dc03DIJRk1uz0QKHwTcIehrjI4MEJokY6Oup22TalnixUcpPrr +/XiMQVqLmRcR++ZdWy9d376MBOWdpSvkuHg8m8YQElxdLoNYOhICyCWXBk7MZdpRMGkVm5YFqdG XZvYg/6U0+paasHtOEmRJpmUOmntfvxNiXo17u4ZedAhFpQP2kNp3TMgoQl0OKRATS2VuQYi8zOA 3nTcZ3Yxh+NjUgB70JmoIAyh9fI1WAtMVdpzsJpI1tfVsq10F9UJpHYUg1fm6IWf5L4rUMjz5l3c TEQT5azucIfYO5iiqYhZByp9pNxZ03DvUH+sSsMBUDtYoynatSWqpnOK6a85GVR0nZo1DYvRUVvc ReFwQsk3TMcQhsCaxGTK0baIy5f5dmjh1+CpSzZEJnR8PZaYdRZG8E38htEjWSAZYlYLursYN3zf 1E0afrfIL9ivLlbtlRBvWHgtENgCiaMqbrK2vdKxxCZq1j5ABczCcEt9BZistZHUdFSIt55Uj9eZ xKbT77o4Mut9zoxvOUxNWmaqdnOphmPnAoI662xYD6rznEXXhWqdO59/jtuuVtPBUhiPR2syJZWp lz1eucnEUmK92nz7eUBKmT4uZe+Q6WpD8Yciz5+sKw7zV5y/s31ytUAAmRP4L7r5+SHvgN3Q2Hqa LAWTJW+Ubc5vYEi1JB8z6ts7McxGfNMmJ4Wz7xkSqyxZmNBWVhmy+oFvX2RL+ehDXfXkPsOqTefG r5Z/0H885xRZaGyd7ASbMQaou6jKyR2L9+48wGWWxTTu4AYF7shFz5pvf3PJcwjkunxf7J1XDLg4 3BcmpnW8IRHPXf+uTTSiLsFZxWCD5GIHetnVi1+JEjVWIJ+nkZYfj6Qix+K3ebB4uOaQoVefLoXw LzEqTXmCaKD52fAM7isQ+wHgKr+4n5XtdDZvVxJUz2i3oXNOer2BcgKlH6wtJ5JL9U7PjyxtbB83 tqK58n0fxQM+vemEmeQGQGAwyrgeG4T6RkbCjilO4HW7/Fp86rTw149ECh8IForoRc6bSuAYky4p u5DTiy6i4B79vJch1IR/mBTbaVlTdG9zt3suqu7yllpplfTeZPxcnXYr0WFuZNtiQ49BhrVEdlQ4 kGwQpcPiUYPaSTSWEVvg1EpyD2+DwvEnqTZFxdmNkZdKtv4Y3vMAieCSiSgJiq+Kas6PgBXtjWpB gLy53eUg6R3B7kPGI5r5cxcvIrZysDnc2SB56L7EBrzOjCBPI3Ohn/6lkeDxcqUZNGTsBOGSBKkj oPMMhppoigBYkGXeJ4DCZQhSV+8S8xKQRX613woU1fnq8zGe+6Iu9SP7vS0n3E3GwQdt6UPOwsdG mynSGuSPHctxQhmQWNxU4+l6EFiHbT2MTzPGWRbY/ffR48/RY6ZQ7b/wPcE0hmYzJlSVXbY4/Ixg IQsLgZILy5WYtJE2MX6l7SBih747wlVDEDx0RjGcAoownZ02UOKz9aiDXKr/P9S3Hrg2emdT7cfr OuJsJAaBDkMpWoR8lkrSfSaPQtYi7owY/Qpakk3jX4B0tsuXxn8PsGxjyDb85yuF1SIK62N4i69q hooLOePEhYwIUfDhXlTZYKMiYEZC5aWaq81D5M4SEi7P+iioAokqUfCs/7VmhDpufbCcmEy1P6Ss L8wFHirf3whxgVAFUNblTVGrnJJMBsUUInAtF1V6eSY6k3zmxoddJIefD1MWbWua/BFKK4E/rhQL P7T3b53hYjo2eqFR31ys12MIwzQZuI0SJLaj60kDVsegH9obtwsBoSBspydD32JwvRKwKrwyR9Qr 3awpU1rK50J7IgPKkjyEUnOlwoOxtEHNTqUzH3ZAaSEohviEY+Jh4+zn9wn027eBipHTe/zwaHe6 ctmdHedIP9S67GTMvWY5EUiQHLgQ21WrvfMgNKRGy5SgkPeZ6JQ/GlAdwQGNLEVbHRKaiMmVmFJd Vkh60jDCmWeR5NFWNXWE19kS+40Tw5RxRmhGcjSuz3O7eBlH0XFfXHdQY41Q8wzfqvSYYGKBnfZ7 BtrIb54QmQ9Fik4Ro8O66tZpHrMhZDnwb6vXVhxnfbYpjjslphScSdvMXhhWjGpAEKH4zS2wtvAy 5q53CnjCoQHGXrBv2BTb4XNK+w/mifCzj/EApzwmuCKW4ZvUBUqBGxnHhCt3pHQKl2ESQVJ12pAa eDbpQplMGCWCLexFNq2MknqviYfziVWApjgcGjw+G1fRmtgrVPc2GIM4LQdK3FvEeAdH4dwb++RN PMGE5xzw3IX91sGjuYKnNYRgqupnq4cfUfozoGXQO7RrHSjA5aHAt853JA2wUE6npoFEj56tIQOu 4CdoRgBC0hgivGqCxVzFNi+GCEF7rrf/7IO6ijS/D3gDz7yWs7hmexkY9PV2UjFvAvJYf+5jqrtw 2gyyXJ/69+GYYjkPtQFOa2tERCFFb1PHYqW5j+Y+q9Nb1zyuFLVEotQXYZshh2N1mkXNaxxUldnV PNu7ZRobv1WF0MvItej0B3LlF+wObuELzj31h4CrO+RG3LE8XqRftQvpQ8EfHLi7dyAM5Z78Tgjs zrdclYrC07KHpncp3ZVMVfDmjx5VojldILWXGHkkRD8Ihyv3qSEKBviu36XUGewy3KZPof+qQg4i QvN6QE5iTF/aLrdp/RLjgDIeDlGmhXj+t6cZfM2l9wCWmVPek+0rWQeP0DGOCUUwqnhGzTh8wjWE E7W6//x6Prkg2H3hjI/JICvOE0TrpXi6Ojs4USqvXtpZD7aSGiOjQHiVoaiY6X3XP1fGKmkSlluG VrjLlBJ+xa19s4k13iDUYYTaA2AezYxxdhm4a01fk4Wr2ofT1GLbB9de2jipUPujfvfPx1u2D9W/ v8Jcoc++UTnSK2vWO9oNk5RRfmPqGDkVqU30RQ8dC8ZGQtBSYmgDHNMuLv09+WRI5L3IaGH+hFmn wTPWhkCrRFu6nsSnVb5lBdiDbLX1haErqYVQr9fbUgo3IzhhCkmqGTF2QE+ltuo+JfPWmt1+5oNq +xAKJfV62hbIVzhZqYdrXPNLdp+JuTwSXjrJCorJ6ULqwp09DnCH+WFcggDXBuOQPOTHDLXo8LPi OkmP9FRlMmEIMcOL5sQWaqy8n2K8TB6nTl/mezpbJZvXNZA6MrJqZ6GOG67eWmTsCHGHUxIGhC5Q ipTDdxWOmuZu3BHp4jpk8z/X6k9R8WRlHPFwpv46JdxUotswN3dfoZDfI3uv9/Wy1xQkt3dRpCP7 mxUzI5Xb3Zuo9LfBCmCB4OyntRy76QLV+mJF/0ap2p4TxIbmPmbmhlHXwlhs54WB6jTfRI5qf25O Aklx2Kphzcuv+0XAUoIxfnRkgmh0vMzKULCjNMO9FASbr+xBEAxV81QKoclP9n62OEzJrWkmDF6T S8LRNxfjz79bfn9aTnHGjFTs8CT6gvX3xkAGwW9elSWGhyDo7Q9CG+QHoSCKJXrmlqd/ybZJzmCZ iSEGC6gnzUKlDFXUsiUvAI6AHM8gdMpYegZeTfI9QGGztXw3tYQiqJgJnYJ55VQ7LGRiSU6ZAss/ MQYAAYeS3G1vduNgeUKYH2tNKo+MPsEmM7/RkrmS3cxNQM00C/gJbbFyWbodneri7wLE+rP43Q1w jFB8l01UpqCT+cJbLVxz10xhos8rEDd5bM13ncjQMP4yXuwXuXAkZQaJlG1/f1UreC1O7I7u7/wo UWH/ipU0fVFN7bSDzBfHaUpFWRX8pjK2XT/i5Lo/dPXmHmTCzMD1CBXmjNlTNytpH26JXYFEkVhl sZXGArKlpr+QDp1YUvjgfgU2lqjUe/zS3vHoFdIjmaVvwVTRAyE5YG5fDbRz9xuPR4PsaRCmeoM3 SSrcP0nOJ6/kxDssJ71UwXEItnmDyux1oMSCfrvx45DLS+vsjr3oKJZZG/47jPqoWcTnNwspUhuV rYJySNJEPgYkSLXFf8leolnZuUhUfKyr2Z9JCOOLPXQw7gJGR2jD746iG8gDQSlkX3RkEvIc4m1C noTlwXxT/jctTjidDY1LaYn8RkpZ7Xrkdwa2hMMzaFQ5z4sQCfoaIcpZhu+pO3LeLSHfL5kqKJjH L6RcHfb3e2lc8eH31CUnBHkkqyQS4bX8LgLkedxBEGS/+d/9nDzqDuVZmLiLJbSkjp7IfY35I7lx ohU47ejay+7lNzjWJUi0VBVC4UYOB2AbVs47gbm3EG6Trho52mpxUXEcQIK0LFEqgQJJ1L7oDN0U jgIK2ve816ZL9gwWAuAJva8OyeHZLqWbaOR5uR8rtyKdH3v3wPOjL9UFffQfuA2kaCCAF7JY8zRc 3IxdRjQksC0jyeO0kc90exEs05KJu0R5schgCO9Fkc2729oUjkB0A19HpPZ9Wt3/FNBtwyYEZHo1 RnTcFkl/luqpbSF42rbh+4VTXqlUgsQJsYq2uNn1cbdKuh1nQHwWlWFzW2xtnABeKxjy+jB1YU5B ZE+bN3qy/0O95lhJ+xdgQkG/yIPF9oo7tUc0YXwikMbpudonJ+zODC9Qsnggj7QfkBUESVIl/qtz oRzBceV+KTPt3mSrsD3N+BTP0ZFv9ynyQaN0d8IqWgjoTf9trL5Oduz3rM8PjEL6GosnND6GO/Ur qbUpXkIfZUL2xZqlApZKlSAijgoedCNgHy4jqNvmTbtxtoZKrJ7Wxa6onJzY7htytebiSNJg4w02 JAhs+IBzOiUj1pv7+hxxGTsmC8xXRhDKlPhCxltnghsu5eVHwYJlrCUlPxkrlHeBBCA00DGG08Ga 9oPJWFAD0+C62FMNZbtKW9WDkOj/YDMX5x9LcbTE1XwfnuBTxWvMJnbC8kqpwBkF8Mn0gDKAil7p VkFw2zPC8V0JdBJguo0HV1ZZ5L97LtYeNLaUrX5QznJKZdom3rhVSj62lKogRR2eSG7YOJLxAcSg V7NVihWSYqOuu4+DxvYXLNOoQHHOi3vGtFy33Z+ivsSeZOuOMj/H/jlYFdjkPSyWJbK/1dYzzFAW 2sWsQM3u6XrVs6tTi/3HhYb2CnyVnswiTEqYoeBfoBEiz4lkfphSTUiQnuWiF2AvIG35+PcTYmg/ OZOx7+LQsTIHIlIEQkEq13guCUyOv5G4b62ua21lzNu2qPj63p+NMQILwuEi19hhOLEYkigvjmCr QJHtYT4S8BG8Quh5MksaEsNxB5tXcuuSGY39B3IH3GRKIFTlxt4ggs0o9bLWP91R/lUTdK/p4h+I WeNSt6LCCdkbhz5FT/OXB2PyQ1/moNyDUiYAP3kdv4LXMqrPEjNpWn0cN/yPTwstWIh3BVVxj4WD hLb0VpX1eyNvhA1OZx/1J6m+yPFJFZrPqega6XuEoRzbuSDdp+Nbz0lVwkPwI58s9WxOs6Pf6OIg 87Uwj0XHceWrhabItdYpPKAODflc4o7nqtZdnqI+1lNfLkGOPsWCFPk8sZ5G1v4N53cF37/Wlgja VYVU4jF7a58+da41GDKF4v3Nv8iHQu0taswK34ReRXjRAPdmkcrqcKXkEdzwnimC17po4Vep4e/r OuF7p5l2jXNvTBlFDN7ao7xDCCE2TbXs/4Y8NkVYwrjmmLgwbOPIvmbYUrxh43F5P9w+TZsOiKE8 zUdz6ZjHM8CwnUOnSkCdVYppgHG9Tn8zgt6Gk/wqxckTLL8zDeVBg4HLsyFv1Oz9rV+CRACq7XXt DNOZZt7JifH/yHSjShh2XRySujLMUHXL6FLqv+Bn5LBySjL8fAAPCevopZldqpLeUHoSIohy/1d1 l6ZVxJK2jL66yoXq4GWqQy1fnCS5dyHed5my1riK2BD7EzQITxvYrLAaI8ScU4k4O3ecN2WRbVUz +Oz82ErU4I0ZQ+tPCrSOQQG5kQ6L4MNWx3OVcMNBCVqWoX1gr8dAF1GacZ2fAMzaJ2cOWSpS/wnM knLgBn0AdoiB7uSJt9IcWx58gjgyMio7KDqZNbxwve8s8ae2++ACPimp5eqkIILvzksHorKsT4FP 94oq66QadNtdSxnyIC8ts26eTX30fhOV1GQD+U2fUPp2WZzhbv3D51lIYsAT3T0EVZOKQAH/s2iv jH/H46FrmNI1nKgvBzuX/tdhvTUaq4dI+r6F5WHKCCUl6SOu7VBxgQFM2iRkXEBeCkLT849tPU+4 S8ZDLhZQylbed4BVvAr0Mgg5RbIrK/Bvq5uWVad+YpKnGijBjkhEcGR6T6mEHvwSnxsf9Vk9yGeJ xkwoKGidk1YfDAYggW6hgSMbxCEFtebdTK93rAktPwKDQl/BAznn/D2vxwjGqDu8+c0JaVsnxqpg VxTMmhYm4guDJFSrWnUTRsgrdQHLsor04x1RLmr8DKDIqVVIBgp7qCuBEAaW8bVzUFvHRTEmZ/+O zy9OnCq5KCq+h00UV2x8Mp/QYPasj6trbpRyiYKLIvbq+RP2qSNj5RN3oP71969XYT/QLS7CQnI6 e69OisfFbsTDbAobAsR+80Mct8CX35giDX0UK5JNOwoDt76Rycn54cgW33qHuWKESud8nQkbDPq/ 531VBXg1YcZ6Gr46SkU9LdN9QBbk70jFXZXwYjcPnioah4gG8r11AyMskAZYhifhMU9Gb+1I0Fsn 39nnnv2DKrrTSku6ZocMwxuffhCOUPLA8vaGYGkVdJZo43RMe//gmV44FEMewYxSABncZDRGdBh+ 2eUw/vzyDrNf5kuxYxcKfqBQHdl0M/+1/14+MGbarW6gAhO4AyZJJ+L451hMgF+UrCXEhO3fMeXX g7mUVXbGVlMR8sRNbQrF30EyOZko8knML+o0TlhinW1vvJomdktXztYnRB4SpPMMLCKl5Fhjac7N sJzxpll7HgL7+Y0sVtqd5b3EXtwxs7SR2zoM5ZTcJhzzaDqZhWitRkCkLj4gWUwlCFe6QG130Dhs yJIQ4Di6Ep/4e4eRQwm0QBoEMvh3PtS8Fn5mHwL06LyVv0VYa0CEoQ5P/NtYcwwgVMfg6Gsz2Tll kmhp1iKrbCI9Jy52xysRGMiTwe2kI+JhjfD7nE60Vhdv7Vj73HQOgUla6812RhTkG8/l8C0G/6r7 GUVmVpq1+oIC9ln2nS05g1iRffWg2m2i69sP8ujlVvuPl1Apm+PK2/2Xx71gNS7UsT/euWz/f4Jv 76HjOv2QvWQrojN4+vZ+JhxARsr/7WFUH3Pm++R1E7ShbmR5s7GI7UXAC0KXl64zkGVMHCFl+Y0M DSevxSer6TOvIlFPWF6d5ASMmI8pqgLR7N3D80llmkYbKJYf51XW/QdR+yv4lYYYB84Ke+RuOeUZ 8tD6w7+xnisP9ewec8S82TfaBSbJdHqhy67O76msMq4pzWJq3L/3boPbZtFVdMBpq//CGvFjRKLT ZyofSkdkazVMRhheYPXyjrog8CRf0S/JN+3g8xtd+P4A2KyI5ZNOM43XS7+oFQEve8TqSafa8lu/ Tz1z0arNOOhQW+Chs9c5IhyliyPb0kQ7D7bE8smSZ+A8IQm+u2ZY/eJ1kvGAaBZkjo/p+oDZJDXt x1v/XfPz1IjvvGCONOrGcV5kzJacysIQyz3J9zu848ZpWY+8SPHTOGp+RX3FvdrolOjq0Y4WVCX1 ouLvoPbzXI6M46DBBGADKo8yG//IR+wks29no0zJwJ4mR50qwIF/GQjJmDXjT3WW3J1ka48fQwPY OgwN7d/PEusKyyVnieI4MIuP0yGw92oIRIwmw9hARPSaZT/5Zcj3qevXyxjJBMKn56shhxZ/Z7uC AifCkoImHUfHVl3Q1W60iTg5W/nJ9XYWPo12Q8jjZaZeEY0dEXXqy95ezo04XURVziEkQ+QbEMw2 7mQSY6I3ThL9wdJv5JMeVBV3j4DXUUIJ+ZnzKH2MVTodPiWMILniHI67RwgxqWP7P41xS9lnF9WL WTd0f8uk61K3i179baYjlscrycgz4ngRUtMzVhl5WtKOT3RTOlIa4DDjMAWTQcj/s6A1WHava2wr Ygz4Cv/v+lbwy/ljAkyFyi3/GEJO/laNMmN26q0ihxSyr4m8IRneASkDxcrS9LitVAjCBlzrS1+9 bGzMMwbX6FqDxtmfq+S3AvH6adAxwOOzNFwC7QKZBl5DMTf0RNtvsCTC6bEcS99ybzkXcFEXwsrv VlbzprCbDcBft4LzlJSUtbMTSLsf+k63cqQZXloGMCvF4pcLVRcVFsC4PfnSLFrtD8oZPc5KAt75 LoIBoaeg2ywzvv+KNlnHlsh1YDVE86+loXDwq4SIBdFHrd3VB8iWyjmLbAvssxGFQJCp4u4tWkKE IQyxeXrcKMdVXTG2kdK2n856ExIAdxeuWhALBR8pNT3hUIo1WmAPLInShNPL2cAvkFVt/gFbz7KV zrTaPJz0S7lvPLNP3pqNeZzW8KWtFKdiYAQTHJ8qsAuWeTk5HJ/iDSyngoBrAuy5JDtNdsjMOopK bW3P+mhADxC7ONir2DWh6HsPn8H2Te/0vEPMSnhK1IcmjI1bOaMO1yNEAsFTLrbqAThxSatd4oUT s2PV3MGsLZOJzNNNWy1tStvlDnQ4MuAx9vzBRFP/93PAK3fVr95kazuVvoWjwlBndwJUXbStGObM ZpR5kx6cI8Ca2bJCCxTOxUkHXRPSDniGaOoab4rrlRBUvLPs2OFfzyRVIQJcRREcpioQklmC+8JT 88nGVsPZAJT4g6LDQHUti6caTDwWAt1Vz3p1za5RRJn3UTqWDTgEJztNHnJU6LGN96StWFH7QCbm 1f2K7Z2yHTCQnBPg026pQhe89DyvtLiX3uZLu29mimJChq4NHj4qLkatUlyjlicSR1aYGQgaS/oZ QmKexlJZWm5w08WravtJgHnY4zISsfuRXh3w4rMYEv3T/K2TB8EjhzAJFAtQlzYKOSGY3C85Xqea GQbNgjJ1DLGrNVa1fZuCnN2tD19s36tZViHscruofTg3OQcnXSSlHQhJLNF6LKXpsAxSB9VjFVKC H8clEi2BFHtS85rIEKM7Ki6BJfG5PIYjni22AMZdPrwcW3r8xjYxXdvPVukkKl5iQHG2H3SKwMAD 9zHhIg6b2VNiOr30U5GLyGmtrWP6THNLGmmfHysrWB2UuYVvrH/LZYg6dpRk3lajLPRKjfu+3JUB l7+SEB0rLEFNMRIdh7n84Vez3ZzAHnO9Qm0voe9Bn2V09zPj/jluiQHGDOjyIkEHLjbtNpRxhtrw sIux6qZgZZklgbEUcXO9F/kv6y7voSzxbCbitr7ohnH+z2/U76xJiz6jEjuHsZGuoltNePQioUfj rxez60GACTl4CPl2SmPwK9Prb3xCRxNGPQ107XNpyy1RmCLL6+AZeOKbQCIh18mlFi2Jebusj7Uf 1QFzJRiTIUPz5WzvcUNXwcPHeI0NLpB3auOqE4SnMkbP8nShfNfLWmvxyf3LQIj+VCeGDO/dEarQ Q5vOIk6MKjH7ZxM/J+G8/qk9F9kksGhPwPGYYlZPmtTNIOWy49uK5mJpBasfQRudOVCSTYLZA0FL 7Q2/rIsdPmBPmnNdKgW6ZG60SJh+Vz0D2tJZJDABcomnnFjSg3aOfMzGiZ4d2C3UU65qvH+TlxgN gZppyXRN6yojHCyCpcRVlNztHzK7nOwAd2cfu5+nTFUlfyPP5dZTN6C1R86of6Ot5vKI0K6nkok9 30/hyZUZDmVNhQpF4BCD3NYN6/UcatNn+PBpte4kHHSp2bEb261oAq7Q4uQu6T114jOds7RTrSrw TeFq/eEF/0a9NO0NLnVPpgNZMD8dqSaSzTbZMAUS2XZXhXeYTrn0FAlZpozOCWLFnQAlJ6peF/FF wNErqpCkMDfUZqqj+uYFlUFa+lUp2T1/i4vICEzJvyjJqcuoLoq2XOh24rCM1GBVgc/vmZstd2T/ czo21SKKTv803eui64r0/kylDPVbko/eDBr0SkLAyTyLyqFBlC/o3lJyRsYPMEU4y8eLXN8DQ0z2 KCFu8IsqhFBs3aeeM9VjGlKE4qeYD+PIx7GDl49zHStW5N3JYUypY83RUAxG9yTDTZCXY93BNZFu n9J6eyMtn12EKraeYPFpKbmSXMVugLWs5vYQ1QgDCq1MmC42ONhyAlnYd1o6VIyxfZ9KcIlI57qE jG3tdEImF1t4QpYeOpe3nSI1Ga9ZnKnlYO7Y8X5PalKiijRjvnKNeq9RYwyK52YuLKPUdzSvhzLc W7+2KdoNdjHtJjGZSgxqVACT1fkaFEZsvr5vnQb2IHceAEAChhimpPK051RWw+BjCbetWIiFaYm3 oGZasa2goPlMy7GCdJMVaPXis1zNgG0rJXlP54c0YbA/Y1bWr1D39/GF65P0UFZ8Ty4iSzx0+vcS fdf4Rnv3h2NjoNT0NEVH4v7yM455EbGN4lCuIOof+TrGYMd1/3pyJDjSDGwSKJnXMgd8Mnq/xFZ9 VA6Zaf8jinV4rOG6M77ZcZW6eJvJKV9PhRoTenPBerHStdf9W3vGxwbp9+jH4R0OQuYKDECm99MF xj9YLJhbeFElTZWMZDJ6LMyjP4xglCHr4hSmTuE5plJtbCrMNLweKSxC/1ieYM2Qb7WXhnVcL5pf mjLgUVqH4azaq0WVwVzxAKFBBVQMQYgt0EoUhUFFQ8sGbCnHxPzc2p7Lkq3M1D3pS/6ZbSAJ+w97 vsvQ5fovQ+VGCjCpWjRPlNosePxLwn7oxVVSolYgaTE+hW1PJJhfelaShSX/3vvHrsaaJvZsfNp0 A+8MXJaJVZeF0+Rb5/9w6MkuU/i3r0zN3y6nkbcVdA8q6LbFra1TsS2wUiHO6weH5I0HCfstxrOz KIKTrow6iBoUa1e7r7UrsMyRNj5A1IfX8d9wVJ9ib8OnL4rNmuEib7I5c5tcSewqGlqtm27V7F4O uUZRMulSXoGSTRtgLqk9v0hsjblWsKGr9PRNDkPkRVnDsma+XOSdJ/g5soOwoNyVz2GzugnrfWHn CqKOnYPsGUQggQzcDZiDJUOVnBmURgd8dSMdpgW0T2Dm4UJFCZ/1JfnMEbkqQLIyn36Sf/EqEDTS 3PIEJNrreiSkkq4iQh3N9Vd1w033Aw9Zun5UusViW/AhKh0tDU9t0QII5czec8H6UKLopjyhERse +rwcO1U9R53heOprpbBm1UKclGy4Q8TP8ENSok9H5N9tx8Bf2sDjT7C5YKOt07MZQBXNV6qIwnkC EHKUn+xXVFQq9Nhb2MOcYm7Eo6oGgfFqimUsl9qJCkyKgYtpcsl9CxxA1MWvLp9MTs0b/BIcfpiA VAJl9Jcz6qgLPOXF2W0AF3jOguDCC3RYZ5wZ5HxRGwNReId2yVbMOA+kDXLOkql9X686V6VRqxFf tujngOJnBEmCgUtgG7edmx/lNXqnzR7GDxtDXtEs3CO9O9JksTf0w9p0xqt53+HkvNsLATXY23WU 4f5Iqs97BFF2JfTgjAb44T5nLTVwWE2EcQvcnYG8ZDnlM5M6DmIFW0+sxwn+ZW7mb5iDanXYdQHe bhD9z2y9IQVkTqVAzso+g+PikPQs8kPfixLTmYalcIk770wDUhyWAubKvAh0NcN0LN+46NfgFTSL SxbMrNJpjfm3P/pWtc/XPHRQsMIr94n0nVEU5xkIJYS6FJ8aS07Zz13fzwju5BNfCoyIdaHofBtW ZXcBxS2EF4qtGIbaz2IgbIpEXnTC9d6YueUOniLnS/dYafrl9ySn/IZ7zrN8bdDdgVDvnSC7ce+I g5sYleWQ6zURxkJMkncY4smCD7TYZf781G8KUtISrzvmVcBAR7+isGsukPnSxhDIKJHeMRq5T4lY AAEBlE2ao5WCf7nHnu96bNEUnXy7/vAYRiSb2ZVlS8f5DlzUfBvMMKF/pOqroRXB6P++DrWsRZFX prUs6ubemAOZESF48HnqXlFyJK/By+hd5p+zeTYRydPZqffeXHTiUPAiEhSoyjRuahor75fc7dAy N46JcjblSCPRHg9UuR2JrIvgIF9yYDbc2CFaK1Bg8WoRI2aAIjoQqF1085IeBwRernrDXoTlDp1Z SQ62JYbiEvVcgxBofrDxiUdFcDsh6veiUefbhHXWBz78BANXo1CMDYllEAjUp4APdUEYTA3jCVnM 55fd/jGH2i2EW69C0nFWjWJNZBo/92q7sziu7IYWaZblDVEHSDqi6r4vrVSwLyEbXh3ap3SY2OO6 s3zkppUluDZYwPLDuqqodRxc/nzwA5gwjMSsfk56rGN4N9ie4H6hmCNEwKmghNQpdD0ZFP+yA9k3 rZ13AW4ougcuiMvBlTmQ0a/4PPm4Dob81swXxPouKNVW/03RfGK2aqhKEWdEBHbPYXFLmHuZEhQi gT3nDjxz7LnIpKTuwjUYpEy/IgEM3giXtaPtHhN9UMIrrDdDUXZe1lbC+t9qyRqvEd2/inT8mw0o KjCrppNNFXWPiuRf0wb/NwenSbDX4CmqKC2wibLMYk+G7t1ZRCJvwWhjYzvj5AROVdssf/J2M9Ul jHnoZ5k7wvBobH31l/pdFP0sQOQQj5FxYkSMV5uLZg8u+ZdY/TSzSNwy9U4zLsYRxBl6DwX879WD DUhGTpnm7x/nZqL74RDVu+qAqFiOJzslyfVVtetgusRn8aZYUB8TgUgWBwNwCUduyGyety+on+dW t7BCnQYMwMJPGKrOm0fOEk4dhJpdgsnVEQ84FrrOZX4vyZB9H8LwJc2VlHVoXluAG+N4x5mrpxaR NPJguVAVbsUYGz+Evz2UHUjMAzbcp1AmWC/ud+voPbPv9+fwHLMVGzWeuie4Bs2VvqpOgYwWxscR wCSKoWx9LpfeUXd+s2XEiZkB/0oSUlMU0n0HViZwAQRPBCUEiw0j5tM4TcwyU7eXLTKSJWhDxJz0 GFJRif9uLkLUB1AbzipG4lLslziNxdaJyhd5X3gtCGKKvvwlmXwT2/BD9/ttqwnU3k4WA+0kIz2h sRvS2Vu6hcijl2H9gCiVhli3bW/hvMK7H26uYJym0pOosSZkWVO7xvjCSlHNChw6LVUFXeUyHqQl sJx2BUQf3Akbb09ObN2XNFUpUBMzlr7AyM/3EqioY7KB4PaVPXBHmuWCeaTkxwqCNBfbp0Y8ukn0 L3d3Ch51lTRpDqr5DDsHr4Qwu2mbX5OwdN7TFFCmMz+WmOLEUpT4ixEjq1HcDHBK4mngfpHD2kr5 Gtw/5G/XIy085RqyxLsYzq8WNoj1bANSIy30tV+Ejfs5OwbM9p5SexLQDH/OMWZlQuP7pJY3TM9W o048FkHqLjrqCDbXR6KNwHNsEq9ZpN9T3TMp7JgaeEx96QDhjVVmIUMYY4iv9vcTd2yvZkPsLs6e WFStD4p933HRNPPQweVrZjddSFVZk8uzphJMdnX1Rb/NWERMz1nOTcNIIUsxf/+kLivdsjUOO2Dm +j4CpHELnUYz+4UP8mv9DW9LXoNlXTAW0K3PhQgTlkP3rT5seSu/Etyu0bxoIBdM+W7R5p5JNk9J +IWOsD9k+3UbH5YfzvuPtiKdEdAoe4clcPy6nfrErPzUuq0zOsgf/jBqeprDmMZ2naj4jRSLUA+V sYX2wHDXSpxvnZ4+hcW05u8hhwzLCbObwWHZMB9upqu+2oVIGcV7MGztZac/VhL4qc6uI1gpScuZ nGjs0xxbV+1L8MwAEw97jgCeRrj7nrU+ZfF12aeDMysSfABy6v31D/nSmib2XMU/UvYJCYIvKyXN mYlOg2Akk1upePJkYq83Mjk8XIHWBPFD0MLAtXLWE5JAuDrybivl6Qvz8waoHMlCmc2vJ4rQkKQ+ UyBYODEyfE9u5pOegtvTvwcOeK4iIShSWBV8o7YRTDVSTKo3XP7VEDoHm/UJJhRqD33wzNb0HcS7 ZHSVh4MtBzGm8QL3t+5YrTHLww5gtee/E56HtcXd9QjIYxLrcMfPQ9xwKH0EOZDqINN4Eudp8Nxo dlc84zeFHLK/yCtpIF2WWEKanUtTO3gg8lcIIEKcf1pm+MX+and0f6MTvsZVuYdyTn3P2EGSMdtq jl9ze9jGhOWdNUvF2ao0Ly88dlGF3cFHOcOniE4uWdHzksKgxiXF8evCQ+MJ651xFq2h4osfSLzJ XeH5lX95+rUd2rZtXq7YpJE0l7RoMxpHIGpEVLwMbj6gFl9vH/I951299m5u0sNqSvd9hSNNZ1XN F6QuY59crtgw89Iw35xXRBJK5R0YWUzRmcukMpVB+nA2b4LrGQu+a+Tdw2WjQWHkcT9qUdug2urp xqTWt3d3d90drpKAWv2Tl6HkUihsHE8V8s5nfxorsXLbWBmFPyP0C3JWqXjMAvia4ttmGzhZkbJS 8wAKniDJQjqMkPNixCjwzXMZTumfZszEM3+ACGRHWgwjfVDUfVuqa1oRt2gLmnevN4Xy6P1gdFsk YGom64rH5saf7t+r3NnSypCfLi7pv7BNJxbSUiL1FAwvXC14M2OdJne814tEhdYvcNxr4IzqpL40 G55IwLEko/cUUVS1wmT0IoGeWwQRNY1y1WJEoBKUsY/93Ejefs4wfXQANWJ8Bc3yNiJ4aQX2iJGg U1dQVoRwkRZH+eksTFE6UOGF3E9PrbnyVNMZLRoHlhgJgpPjIyXjVqV9MnE2AzGD93cE03dCZrCY qxwYuqmHkkuRviYknic++Y8+ks9JiTChgeoQpI1zCOI3QfzPqV3jyJDxmnaufBxO58B8kWcvn7Bh ais2ctHsnX0IN9H9fMzR8IScIlp4Ks8VKW4wwGEsEYPiqQSHOcP7UEcNfz+LGei0B6gXpefDk0Id HU8rb2I8r0Cl+bfUqsPDVamRc+5eiHtpCVi26S1yRIOHSHGmjlI9XT3eax4WwPFVBR6Bkc3nDuhx atzTKEaYNiSwZZJXLUFbvUZ95YFPQ5Dn6DhxPFtMY4E6hjYOPRy8lX4rVCRhTkx2RiXTZiv/k+ar 0HL+JkgkooAU3hO/14vA/6rUnoJELfg7GPK7FlibIPNgfWyTZzLoPxhlC13SiO6buN2RxeWdr+5B mD5ncbwnSuBQGw/wEuXbjOYitNVRZHBho1FK3n693dLHLG15yZFRpD3V+Y7UZTwL2D1JdPh9rPkg zz0Bd6eUDcD+nYHbYxZ1csrQz5BBjut4B3/FfLSY08H6XA4/ueP8eATVhMzOitSr/AUVeeNGFJix rKbKRlM1tK8ypjgToa6UsmLYWmEpCex9P8cs7tzgu8t1ZR5eQFdbFUS+3S+sVH3o+tr8xpLtNN0r cEGS9AhIXEywbzYKlwt/JCuyjEs/RjBeHNb+utAIDTX7/EUYDelkYIVZ/EKSy36rK2rfVz3y8BLO KCKjtMqxvDvUWMMoVfM/RoOrUNChfCXT904LVDJsWDS2gFXb+ZFQSJs9QDsb2PE4zah0cKh3IQJG 1z7bTJJvxZeZj9ViacwmLpdtSt+yejHwAQ1jymMr+olAxTQcUpUJnIEAffIsJYThaRN1hHoO0m/2 bFd1Hx0rcBBCA/60upDlEM+LY1pIVqZVDvHaViDYdCQVAHeCakkkF9fd/+SDL/I3/pbIp24G+nFX 0rZP8wXe+8Z9hSJbqHyDII2O1us2JXHdwt9hElG3+CnBOZWV9sCZ4w8bdXh/cUZsi8j1WFKXIrli 4Oof1QU0bxVFzhVWFgBkN7v951HQ28xO3gZmCxuDTmTzq3c/dapSdKiim2hPEjTRpZUv6QYuGeNr ltre0MCZlKOWfsOEJ8dFYSd0u5JnAObCBoMk9REVDJTXcFKoIBnK0zMtLhVZs9P5viY63O9Trjpq hUonN2ZU5Q+YtoFUnl+S1h4s/lZkE8P5jjaY7j8kfunPTIkyqgH/bIbtCEGi7Ad+pqU78NR04ROC shaCI0IGfd/9R8fm+lNkLQYrAphKSwR5SCYTooydrN9IHakHpuPhrnKeLbDjXmuctRPa2z+yY2LU a+g4J9XFbO3bPv1POR5UesRxpsRsFMN8Wkx5HDcMl/dsZba04dwdR+e5IatcexfA0gO6x0bh4Z+a mHk3e5XFpXqoBR1d0usvsj+9VHltCWiaWsVTYJEMGaCCA/fhI2iFrmdzgpxFq7p47JhQ5DRrYrIg 5WPEe8QXPIig9JCbFZCGkfzlb2Jw6VIwDLtmUEcelwxWzynn6wHmiP5VQKFYPxqcaIHucL2vlblG XVz4kVNfx8UvzGIWPOr+INf02GiKuyQY0xVcQihTcUEdw/JQoXKcgTHQjRVNKfdMGbACjeUvn3sn v56tIHLJEseNPAm4LOLyKQEZz/JjXXgt0hFknbyL7Li1d7JeKpZ0NSZO2GZOrxlPOqhNPvuHCrZZ rackYki3/9A7Py7bd5oQ8A31Xh91KMyF+j2AlvehLlguhbvH+WvZ7fRU3eGbEws7TElrjLLdUY64 JnjXhYc1L8b4bBCnjYTedyufsVJR6W9mAyfg/9q1e4EGOicV7om5GLhRyEWJK9FeTdOPPmf+/82m pGpMJInWDpqGsTBdx4P1PKFmPuPOuH+Y8YYE/03P7UFCxzsF8Tfz215F30JT4CJT27hlu9IQ6/B0 ZRtq/t/U/k+r2QtWVQfGiSpcRWe4nt8f+vR+MpVEOUKmlbcuB4ZUqYWLu+VmVs41aje3VKEmMJnp s1jihtmNIH0vz9qUPeXsfd7MgpGtEKHEU5Q4OtzI8DVxID2eEa2bJMpyN6UWdylYjl12RR7ht0HZ DHVcBQziSExCJe2dZ/fNKqeyRQi/Frz/DVk/5Bzkcse9ThMJ8V5ZnDCyGM8IF/nSCb3G7t4tgPmy xe5o0hFKZuOggB26aIQmplGoYtuVzn0W0n+GL6kYRNyBT0HHwdU+ADd5ibNn2+J7sQObVMUzr+oh lkHg8HZy/M1Qg8nlN8D1hMC8duYmvk1tx0YK5314iZAES3R8YZ+rSbFyqiNa5LEdRUI5l8uggcoZ 2T4UBQxmew6zvAFBQgm2HxO99wtnzTZwQouoJ0druZNgYWu7iWcpHaFfgRYi1IX+Y90H3aZrlDS3 Ll2hUsUUkLbVUbhWdBDlOw3LuRrA1pRX9dNewGpzy6xL95UVkzx+CTc2Ga9UeaHDZs6rojGrj7dy yUa1TBvV5whfSUuQgGPLiaBur8ulZ3bguVzPbRSqOhGiLQkU8nC8j7FxBURiNFlR1Q3FJm8oQBJx urob2VxO2R63AQadaGj8eFr1XxmXegz+mFsHlpd20SLWufuZhq9ghMo+2RkLXVMCihEbPEVbJh69 NOV3cIFHWLp6w3MjFj/eEI4Tf04p0qhzHMbqI84QhHep03ORy9ZrkyC4VtcXDDLnBtTSOgRn0vwm zK2322pKVbbYlQDwX5bouW3MqLsCqRlOycJNmBoRSqWwuLB8dC3B5jiRU8VK+FDOXxMSMewMKswt 3Srty6SnP5i1owFUmxYvkLbioA92Xf3kWaURb7H1jl5lemImNdHZ8Kdn600ZMkAm+Z4ES0joeUGc BMh1rK1YkRKEmvOzbTVjRuaMWoM7kmeFBfuKg//Zmw4yIYpR04iAxskoS4+7BuzBmf7xP5fy9PPs k+S1w79afewtONRUKPYJEuF8cOoR2fUkyNTafUzaZouaofEzFPIubTzKuzB3GyMbciXunIgSaWAA MrpOBsQm1tryl08Y+bh020ATLMeh+7w/SnKRsZqNCRviIbPuoGgr2gtlUsjNL4tt8Sc3XWGNLmxg PpR7XHPq3R47nyjPzKzBpOt/+rCi24ZdDJPTJWxbqJtcQJPGmN9KlTOfyW2em4WW59/Av4tgJkyt Op32oiqBVQyc6rN6LWot1LFGc4xSHtlCeQVYIclm8eNOwYE3nyiJCdca/BniBhE+MpE6weeHFUuH jTSL8dkK4cRy3WjwdLPUCYxXQ0iI5O9F1Wct8u7wetMACoTsesgrpsf6CsOEHE+NzWGdPxlVhRra osmABityiJW76iEw+KRzlG9WvLpwZamF6zLm8wztF0DWs1l5LTZmk6m1YmcZsxkDrirB6YAVU1vh 1FeMpKEGfTJ+XQIlLEZGz9Ths7YmytBAzAHwwqY/KE1wRN3aHg+5PTfV6R5+4Cqe+d5zNk0LwdcF JNppmwAfadz48ggOrOByEs6COqPPrYy9mvxJnONY9a8PG5KxEy345VTFdXB/ExSqnDVMjV1LLbH5 91yGK8JI2nAbAEtz6uQEVhQ0TK9EgKuDawUpt32r8Qg51u6DKcM7bjF8teHmgmSFhb5KqCEq7vTV kFz7JRQmffeRUg0nrp+q6WanHPYf+tUuVIP0+Is9aipyjJuHvAoMWKBJVIbQjJa7Kt31v7ID3bjS Bnf430X3ikKGZ0BrwGeWfXmIPoojbs/XB2ABDuNqrDNEAiZ9guEkjhDkn7G5m+IMr2RawhwFyElM jdEdusAMd13yQpJvBBEAnQka0y5nL1gA4+jK6FcM29tQYVGOp8gJ/0iDJ4xdaC3O3/4OEaySW4Iv NQVSu9mhmLq5XhTxmDfmgP2WSwXej7UW7OUMeqSY4jzxbwwwMrcdQGBQpB1FNO1sM9Ng/EkxT76+ 9ZNMId8zL9/mDYxnifbOKSn7twsC5wdtd5smCwmbKrbJPLyyBuXXDIHafacWIT/Of3FZP5Tvh8wX elNNX5lR4JI2dCKUZUKTVmRXUjhK+2/B6b5bdCjsFcSHlYJooesWVnaa5ho9u6y7fTcppXQFHCVX kPzVAaOoJkkUnzER3SH2OsOmGd8FNSi9utbIjR6uy1B4uEZDeAFSCafY6UypGyH02jShY6cXHEZO yG7/dW6IjTjyouW337DXyP5dXyfMAge/sIwy0tc8EGfTbzN5vrK03lBVMddCFcjrm3J6vHCNXan7 vQAI6EM+/BEopYObaTSpAkIZmjKwmaapR78webmEhkygn7iy9lZqW1CyCxLRNa47dRboUvUajDv1 2/bPNyiYvUQOPWGwfPC9vSsQXK1Sm8IGfyZW2VIW+oRqUK/iqBK1s3vUAzSij3GLokNEkbmS7XiW 2pRX8tduALocfrwG/wSr66jnZ2VffJgutvROODBhko4qbAc69kCeFsYSIVrvjgckLlgUm3Dh6KlI 8VMm5WhGAWuI5m3IONgPmZV1bNKFrqT1onOQYzxjlljSOUsBSYDFEoBt9cYpSu2XmsuWVeii3Kla nJ9M+AMG27kr5LVbWTm17FjjFdFlLYmYAuvQgFwUaxld5TWXoB5InN3MjEJ1aF+sPsWBDsRsK8Ee 2hlHFMJqV5t3fa7sSLepWyWwhn/ONEZ73DambN1v0F8g8KFnU0czbFCQNXgs89J8/0EuaSPF2yOm k5aEvPQC6gXkE2gpjtufXsb6uVbbTXPncoZKMvgxpHr4PgO1NZTLl5pnIvbyGrK9QYRuhWCSG7H2 ciH+9VDdS5lOE8BI60kCL5UdBV1bTe1IhR7ehR830h7+Ent9US2ArVTpuWUm1GmAdX/xe44it58k rMunPXOKtGXc/1YtrHPA59x0NwMoGna6r9PL6Q1If4MDivHZ7hdXEXVN79c4Uslvt+koSp7flOoC /xWU+9+8iW/rkbBc/7Zmd/Ws6hPM0muqfyPkv2ge78xJz8mBCGqN8vuqeIZJSCH71ChRDLlnfaRF Rrgq/enaJ/QuJqECbwRu7PqcrISp/pFJJom+VcTsT1iwe2DbZDooMNggKsasYynvkCbjOUH9kHhU hjmr2HRWEPMBBtWiBtX1E49javAcalnDMs4QIP3P9fuqUkYkMi7oZhm4wjlZqYVMc9ZzU6W+l0Vv WuGeTfnaogQP4ihQwsFS9UVAgIlf9iQ2LzVoOXeMGSX5NYzQEtXp0J4xn4hWIHHSAhkjP8Cc/cIt e4gGOy3s83o7cP0h/Hxk1ldqMXICaRrZeH6Hb/h/NuR6UJB75Z4l1eRoCyPwBnVX4cJ93+YNwW9w E6+fCXbev18qOchK8ULZqZlv7SQ7hGRpCHLr/JXdAjpVvpzxEJmxE0Aj5LJzIyIatTabtHe6lMWU SBWk9SJXlsHDy8bPmPDuSmjmcDJHEGu0Lzz671WJKGQdbXlGqWgXIsW+seTMluHV2n6P6HmIdmZo 4KmKselF0sUULgABQ2FnJdIjF9Scq3fZl4Z586wotdhTS8hgU4rc9AOBZpgPou20PO/R8+NQGskD FZLpmliBsn/cQvyNS2MXAr0ry1y0GIH9Ekl0K4N3TdSjoF6dLR3U3koDJNF7iHJ5jUo7PaCcaZol nvTRwInJ68tEwlWCRYqgy3DZVwZ/GaF5z/e25dHHaTtufxTRWVfXS5YPe2LiljzvRCSmKu7HBLDB I5cOcGl3h8qEQSalelyzLkahd8GLlS5MjMWh4YJ7FIQoawalEQ61PrVPxL4A6OPlj97mBGKDK/5R utydqTYBOklUJ8IbpZU27g6NNK3LvdKESr3UXGuTATPflwcgx73bgyBx5LFO1unbFKmZHn0krCdY 8Y5NP3wbsMsWvQev3ZcAud2TuSXBkzJobTlfY1/4oQm+iHumDgPRVlQn/PxOtVOuAPb6OLZfhkga xJTjBC4/JRpY8L2B1C3pC+eRWxVE1qJi4yiFoaZ4RBslMfwQtzke1T0NN/ze/wsbij2yjTq1e3VW votEa0DoInFIqBadcMZ586dXopprQLU7KjPoVgRp99D52xbLzOGZEZ2EX9iWOQWQsQ3NfcCUiy9y hEZLFYBaLFtH3SQXt1vUc8psxOkdV1MTH4VS0DW7Kc/s0IAXMMTW2P/YitpgAZUxYrp+ZW0ROVUQ gNAWqsIH2ocAMFO5vRCQQA1XzuBs2WW2cdHoEycvBdUCAtaCyMFSAvIxcpIsmZyaS+XxjjuUdq4D PTsux3LkWjH8N5CZMdsN2/BFrLGUZMXs/RX4d5R9TRoT9HhBrpM8EKjm+TIIcjVXS1YYqwy/M2CX J+nezRv8dXuHV6ZfZCC0Vil1zPKK0tx+qy0HNyw2pZEoVtRgkBo7StOlJadlASsfa5gaOyAlyNJx W5CRI78TASBRVx+RFWmof1WsXzJOG1XROIZa8k5fFlmUpgYWj2WzKmKZrCikfGv3BqwO/H7SOyVW ccebkAkM+CCJRSxA7gZ1zlGHEcLwMcEIA6+zJ/rP7DW4x3GnZOq4D3JULfyyJAWQUfaB5TIl/kg2 r/u4p/yn7moarDKoZ5F17XrtU79GQmpIDbP/TeN8ltvkabqaqyHD/FlCrHGvoj6GtmQXOmQsLgQY 8zLFP1eHP44vfXDtt7xQMLsfyGOd0iRlbxP0OUqrQmGLTjc5yMDG2f6lmQ+BxMLBw2jQvR4hIHN8 Awh4VtxiniTWJaZz7MIDLNAYghK4rZ3LiONGQaN7uEyGPZPgk3V94w+nCjp230oP3k0HLZFe/WWY vdGD4VMXxSJRjVdx+jz/ToE3nyTjvNeMeHVPIM53YAn+bplCFXTlRYkT1ff986+YAraABn0QI7L6 U5cWVA1wLjnziYw/Z52+XsuaISCPo1TQM1vcnHK7KZeVIhYwM2XJI8WE6MOakxPFAyc/tZlx8fAh unUCqu+7inBpusOpR/2tKS1dpfss8bzknJy1ti/P5Sr+pIbrmXZqe/joo0hatyft3XuhdOgLmYSV +t/1+TN3kcJpa6iGkXQH92BgUwNhbV0XG24ZL6VwaU/RWXmjkjCoNMiCZQx+kR9BEzDF0SxAJiVa Hl9GaYVDLzBPO0802GPmySKcvRn0ctiDfTUIfRf8Rlyr7cU82NBwjnAnKiB/ArmhT2q15xJ3Vg55 IWOMRXadz+JLgW6B3OwfW/NLMALSzEoEmHYBGnn0OrWccyqcM1RN49y/lpiyjweT2tGqJ1DPw3P5 G9warLYwPcKAHFlgdDcd9RNsaR0nPCdLwOK1OjdaBdSxAU5QYsbyMwv8tAMT1YdnAPPEa/7pYlI/ qNCEyViEi65pfsOXtxbDYPe9SskROoM/qvZ7XlG5YoFL4wtoe6bUWGt3Xjk9fcK8bt+MWRPPZNr0 nUQbhDvgVkQOqxZ2Au6zOMBOp1w7q4eDyJLicokx4i8Ngp6sxfRWP9BRyEtciE1ukUMfeDKkV2GH /g6iqhOoBsEXLI/+yU+x9/P6Qi7akrhN35J5zreRgqPZeXBtBbqYVy+vg8pClX05OT7xHHscDNjF X7St8z7uf6iqP4elsY5MJZjiETtK/2RrfUp71boYabzI3/jZ9pJxbcupYso1Z4nCCmlr+AEpBFRI 8bHVpeASgcEP503yTz8+zsYKYhLAmGEc0qg+y1L4d+dGIQZRGjdNP5nUB2aqDHyBdifpx3Ucx8su qEaFiGQhdx61HNTjzjRuKjDjMHs6L7xOEonCkZkqfkO9TZFDFjz+qUb4z481b3lxajoGRJuSephp pYcZ8Lmw7XF88MalGG8El1D0JNiwBfqjLqkmv97FG2+looU4fPazFJ0kS5uWdpTVIdhLikZ1qGC/ MA7Grhoyli8e67WZ2lx4e/l2M7K+/T9C2Whd0L1CBauJr5eyK3U4WGA4B2704UTEs1jytOE0I+i+ AcqCnXRf/jjqwC7LAIoUSalDxtJFsxbB7/GWXE5cKr/r2/xxDDVC7+dfH7ewUjT/DMLv4GhPgjGr 9Swz6Wy2QCaEJAsz22eM0mzh7mdHT847lSG9ZBemN56UtM+SSyKTQXL9SykbIlVMVldHAhXNd5LR 3mqTa9Z4tMYVuVFfs1L3GF//LzgCqrWHRN5Q0lIkQFec+lXwj9P3gl36MZbx9x4b7bFbnYjhevhJ EMRHaIQJOID7Ky9r2Dhxo9S7x/k9aooRVRu8JWLGd/5pfPas65bXiR6cEX1F746RAj8cEbiDsidr LE55l4iF/wsyTdGPzpUKHrw9cj0TMA4GHSv0URJUQIRlqBbjOiS3iwbcCdf3qCrS0HSFb7TlRWYA 3trzR61bEdo+zCeKLAB86OXOtJjgkUwPB9a4i3JWdQK6wUGcTo7EyrNnjD+FtoSlXHb6fo1/OA6R zG7EXwOtD20wbti/Fa+upOkY/fHnSG8Rk5tYAevLmT9Uk4EmQbpi4N+DR47GljdYuscY/yyEdBIf +X3aFK7bCcQzcDIIcvxnL5CVsxW8lpc76Gm2hTN2E3BK0lW8WVZtrP70sR8ZVE8AUwsf3QA015Hs NRecMiNqVx3J7p5lz5WW9wJDeuIjgu1tBciMYVYN4ZxsGRcRiMERrTlFuHYdLPkUeW5KZCxt/5Qg T2QVPWImcuABlySPjSIGG7sBqXsKtS8VAlPPsaUCWOE3xH67GYh1fyK3z6XM6osfzf0GMMEYQRJ/ S/HJzK0CPNoNxjaCn4eW7Og2YzZtUtPm3N/I2A8kjNzySG90ChEPvYbERohf9l00FSHp5em2s35o aWR8F0hJ53TC5oGgmoUazX1aPHNuePXkggK30a18myE7yvKXiyMePVifjDieOntmaZcQJyydhpbg NLBAklCipXgVPxULGJ64aUxQsvLQ6GnEW8/14uruVv7p8IhI/8hdBdzcF6ZQdevJq/Gquu6j8dWm nujp+iDM0xzkVQn5eBSnDdzp0dnoa1bg/gzK2FhU5mHmzGQpYuHiadNgj8zIlsol+lztrBnKBM8r 6kv0LHgYaAa2YkkybsrT5Vpci0eWnvh00cu8q/4O6jMJlp4gHaYqAxWAeUVSO0pI44sw30NGvQ+7 dAP9DRWs+zg3arwLnHxlAA5GXAAsI0GLhTCd+zn8Q+P1MxfezgLNrlD1w14eIptn0gPJg1LTG/Ko e8kK1n4vCr8jKxkSmtjwvZDnr5apnSfV0SZ/nRcPdTOYS0uwQ3udjnHc96uoUkk628S7456ewZ2c 3BnDcDmLebd+GVFJyazGBbIfWzZ43AYGzgTTKJzTFB6SeKt77h5fgZx/DEnYCodCMCxnNxTMTQ0b z2Q6+JIshI1IbcNgRp3/q/A3OllFfpG2lhzUQvWrrd3Ez0HGJMbxefVU/BfMYxuIciALr1pVMuGs gPQOuaC+BNhuTiXyP1RE42HnxhkDOn/BwKSjuvgeyVgiuY9aGmb5CheS0s1vo7CSvLrNpp6noATS jVXSijMENaI4FIfSk/30WkFjY3b5bllbmyOij/VP7YewswTAJT6lsPdfkR03s2obWTtQf/MwIzbA NRg822rpwR6mx7QX7mifZNzZfmi03gtL5BpUzVIJwn0Sfu0LxMZV40KFK+5oT8g2DG4QOX82ZdWJ GqIDewkeln1ZtSTwh95ZPVQOuOrSsd76y9mzsKID9iuPA+Tfmuh3TiICgFfbU1MNpr1qxEDc0uX7 eYhZ7OqbtMC5eMusMxKCv4drQqixjCzWpBw6y/XSRly1GjKZu5h99SsViOkwt6EeS8ez+aeTiReh XKsaBq/lNfNJ0404pdfdR8gK29K58Jsae/j215fDkwbikHo+k7x5GVDexnPZ4Jq1rvTCyy96LW5n f9TSjJHL/DkIQQqzWON1XxmAVhSCmCvd0bHqQf4Fpgs6qbdPbnUbSo2lQAPFLytNFVUz1k/7hKh2 I9y52M7qDC/0Uy9oXLQLZ2mjCwydyvdHT/N1hGVvKYmSBQvMY9klvz2L0BmZgVPYW6JZVTgjHrhw Pu5YhT2Q7TrfQ9pxtdV85Fg031XI6adckbO/oc66U9yY8VcN4vAANsg2F/C0eeoOR68WDkQMhkUh rZbWtDN33GUSzp7OoN7MlY/rmdb0PtlCea4XVi3gz2aM8+d44EVpWG41FxSuSy++xRInV3liQ10D f+jzNTFtkBM+B3xSByVRpsESiOlniuBL6gcmMDSrclHHhKoGWQwLHcxVo/RjWv5NnIBaxFiFIOlx aXjB1cYwB6E4tAh7ln2VJcUAvDCkLme7akLZuUI21Fx8ZPPLEoku3peMDUlFnqmoRUiA+rZhzJcZ eagCO3lruTSR11iYoj5RXY+VjH1XAsLAtqGqZx2kqEjNmcGv1aNgnxcan9wj5pAMbvwXdhHtGTKk 1vk2JGgrNpBBArz2k66aXFiYvwp6cjEldr9POV61QkdZaOiHN5igDP8cNbvijHLXHcI1THVI7SA5 ItZnrxKBtSEU27ceAzLE/ARED1y2ngoBup5TNDriDRPO4ZTcC8rRrgj+8STFFAc4ttf9JdCdZyXW K1z4cdkSugyfQPqdpZSzOwE0KafJupIDjq1HOJcQ6OYsJ0sXK8Pp1Bswf58Dxeeaz5pkPj5B/1I9 CCIewLtJ6UrsreEqShyJDpR+Sxiq914CdmCueB0Z2ikYy8eMaq7VbZi6MvN0FjTeQ7I61cQcYWt1 xvQbPtJr4Ih/0AVRPbKFBi4ZQ7SwGKITZ3op848VxB7fX1YuQ3vM6UXsKifAi39EYPhkfhL8Tcp7 aFctPtoUUdAHEaeJZdFGP4yjo+Eon9NKu1m6lXvy++XkcHkGLDzUtPDArAejyjGW1XwlKAwD1ZRf KyTURo3s3wdUlMlZcIgywH+Izwjy2HyBOMClt5Hkt3DajyVOsGv1zPVrecVpCG6IACAwi0ZBPxeB Ozwjol5Axx9YdQTa+pKH/2eFOo0tF9Y1JWhkmWD5Zhk2MZFQfNnxFOQ7iKtKVd/9qpz/rCcc31pq S3/VsuPvrgEM5whs1ASaMbe3JhD1Ky56oEN32OV0tkxHOt4er79ByGSG0P4e9XAMCVDXGae/e1bi vtJbPNfajmOnNeMXQb06ZgOaV+NSOAPjsMJjWBn2brPB24xc9oViaWHnN2/vz7DUesghvs4Dr/q9 SOOYDLL/SVgDp5W4AhRL8POwgnlJAzWnAYowwXHzQn6CjNAaCugFHABdoQo+V+9dfhCe0nQZzzea 9dI1wp9ufPCV0wFYRvne20hMgBG5++1pOUWvx9r1LyN3vWaJG+P9jB74moU0IgDPlmYf7Ss7h7Z0 EZBm5kXT9X4uAS9q/wOLIXMvvWe8AFQ9Hx7qCBJH0EBmp0byiwYIrl6+RiHyDb8EM88h7f6iIeA9 G0//TLjsnNpT5FlYX93d5VnSfzXoMj0MO5y7gDCOzCd/4APc1JJopJpYzBDn5dyze1keOkGCTDuZ 9PUjRai+qeLSARNVH5Q1gyt4pOTNS3EUJT6hSQDZoz0SSOSY5zzoA6F7ybCInG2qBMn/J26wYtE2 ORalHYtEARqnA4xQblIq7iUS4GgBFcu4V1ZELblGLkb2FlhGjOFQVHizDZfzZDC1CSCx37vpQRvA YuDOIKZx6aJMlt0/7I7ITjiwRzmA+WG5nPAp7CZ0yZHtETZsRxF29PkQEmF/f5Hd83xJXP4CnBcq 7oP5TrQTSNQuDQ8z/3WetkXopZXUYmBqTI7BzosBadeRNDjKrNv+8O0FY11epOBI90Rin9cA9pm+ iZ51FtTdrSrUmya0ogsW2pEBcVapwqf4onHJsr3YQxIU/Sy5gBwCUel8h/6PufCwjDnsewSASPKO DnKYVwP7hdyWc1fYXMejChKzuFretUguaGMoMA3RQN0EsAxL/GmGYd2MgDNVTiYtXJxwB1kG49mt hZAMk3ecZCItqon8HEFSqW9C4B7WqBH/VhSVdOTcyCgkVJrcfGjdnY4xGlXcUaAKwWfW+lK6RLx7 O2bm2JdCS2R9+4Zec+ktPMEWe9vM0VQyzosTyuq2eYwWXhKrtqa0+1q95EIo1i4H0iHo1oC+OvsP aDBovlWyDmo8RiHNpc6U7T7vXlwOKa8TrQ1THLFitfrw8B3nToHDu8twjVjMDFd9pp2h/w48b8PQ zVGG3Z0UlkrpGulPHy5tGe8i2/tWyBXasvSo4XqrxVRuVef9FG4XFb/YBr4cijChJQ1aAiTRw16y C9twTAI+Biiqt9FcVv/AQQSbHswHeX5scAaJ/f8fZbKK2MADEkF/5vA8al1R+PjMoirIDZaWF/DJ IkZU54ltmM+AI7HMB9WacA5bNdVCO/n2AzqX8qbvNIb2iFF40b9BZ0MLM3TvsYg9b98wBS3hPUbx EzgW2GyL9Gavm3TFZnleynNJIwFSq/ZXdXz0XzK04sT0Re2h1/6LYgUvIH9YY94ktbTVRsZYm+Y5 OkLcalQFzKyQVlEcHFkC/VhT41q+47GkFiB0sJC14esqsMdpJ59HPnatVUR/YeC/DXW9m2jvOgOr MdSiHlLZKRDDMrnoaJru0g8HCD3sB1k1136qtezhnWUsMZkP/rxJ1msssr9HzM/ik5hM/VV0iGlD r31rKjpAL0bTMhQlbI1zKOtMCCtlu632uvu4QioEIcj6zWqdeupFhA9Bb1Pnz5EEvjN/IKCnW9Iw x5GEDkKztlM1nhPI+MeVS5JM75k/KrCcKdSJcPm5BXQxO+UUA4ECnrVKQelvW+vmaGC/MjYKNT4K Wh1derM4y2VkYgsKUZZ43O9/zUG+L/4A3GsfDCDO5YglLirAfZnygLR3qmsDUlNTzib+he41CeBn RUUKosZO8kFXyPA07PvsSqOjMWaGM8K6FG3Sd9tI7FKr3ppec1pl5r1W1+vklG6+ve8Oz7H3JU4o VvcOx5RtOiq8gzmhzeij6SsBm5GXCmAN4FFDtRRSjwUU+emk+pJveq+T2uyGPiCM3zIrsgmfpv68 CxoYxkVcBXKmlqQ+Qf6mMSXe+PjVb5TAMBwBXf1o3Nzkgu8m0xBs/RnrnWF1iEqO9D9FToaI8gxU nyKSCCeMnmWdTD6NH813Y+/DM+rU4f167FjxTHQIMXm4txgT0/dYQbmTgRaA91feN3mWXcugKcKS TH3naiFYpNiUkWLnlbmVNHc/NLgXP4tSluuhKwNPN5II6K38tWtOVeVmG+FV+BBr4jAalHxyD8mF 70QQTVy0pHTLPo9WD07pSBuaCkxE1dQXOzZUGCcNm8SnKYy7XzyzvzkeFJYSgmpcxcekQ6Ui6O9E nMXTZVi26nkzR5p2FPPZsaoCD91sEw8dzr6O615+D5Kyi85SPtOKtQjSWnWecmN9U/vVmjZe6q2H 9Z6VuPVyHtY9/9E31Zd0dzFqtVxikUU7KFgDWcwmoHUmcL/Yp44W4cibM25ieJCDqDQu6tPo9wZj kC2RMFtGPE72CD280qn0JjpQQ7+lhgG3v93QT2iz2ApFgdrVTtzVyYNnz1aTdTZHhDyZgWK4bNnp dH5IjOn7IZIEUhNDNGIgyG74DQchKZGau1TysNRQsBaCbZGq9znVaZAVzXjkm16hYJbcsrlui47U HBVEVfLWsnJoUQbg95/AwKzhDt3S6mZ0geaGgpvIIxGtNrAVsvzEuqChy1JHLLE/jOcjZCDr1a9q L7cv8oUZD0UOq4UHG1wazbVFmMz8muO1p76ibvVLnhTYG6eFA9+kBwXAyTdxtuQ2G7ZAJOkGqxyQ tklQdTFzfO5lldFIizYBpYHdsEOKIa9UW3cfxnQxYhllZ2emdXgtGpSah+UD3GY8avKRmePcIghs z/xfXrndb6FcjRtechUZ7lJZbuKRWbN6vcHveUrQZUya3p3BxASUsGGE+1pTm25c0L7XOszjll/j j2Dyy1JB4PgqMeVb9ieVtWv9bavjpU2U2m0rqt7lBXzqgcrLze8bDs0XHAlEgNG/b0POmMbag084 7HTPlO9x0SvIEPzWXqWldV/frwM/ZjojCyIJvixYmYlM3cgaflgDmMdUDF21sDz9hDx/UNCgYFZX UUf7ouIjYJtSE1UAAFpwylSFZBIo+HxhYvIJ5RImvEIRtXAaPviUtNvdnuw3R9DTDdV7m9BXS2i4 iPLwfMXXGNrSATLy5zGiDkJ+nmLNDvvtwcigPiAwbNzl/lvRkYNH/MrBS5oKOqm4c6ChMa+jTjjc /tZX2kXhNJlZ2oA/IdPW9ZY2W6zplZTha/TsSMXC3x9GOQGBfUoWX3HMnB0JBa+itn6Om+RyBpMK aMYe0GEK9pj/YPHw4Q1++J1+syqRJ8v45XrxP/5j1gaLFPo7d/YQoVzbg8UHsOE1yF+B+NYv7/Yn xKeic8l9u87UuWDSrAyzjh/4EYCQX76kuFBpKDehLR+tlxQhOPPIuQhxbB7b6FlD+VYxeJAqx17e udImdkOqU+1HaLcxiSHb1s/5S/AmeHFhH7xw6v3BOGWuiM7rKWi+gfFpq3eqzfZKVFSGlzXmBOUR Gfmx/RapoUKNxU/0aEZc0Hw78jwqx8eKiVszA6iNmxLT4eKBPTViCFGY+ReuEgs+XReWyTFR2a8D KkA9568YwLwGiFeLhlHnOmPJMDGdwK95LiGjhvZwUIR9h2riJDIrixt6H64CKefBvALQGL8C3oa8 FUAvlkEgvsBPyWBirWb74NUXanNvLqraI1j0HWlgkQY6miK+nVt8e7Cs5/uOLhGDS3lOyih6GN8V 15I/H95t34ziuVELMtP52Sk/fz9AKm2CuPKEX4PDjr/RT3NnEHpcjz2oBWRqjHFtRsaHAojEcOyQ MunGrMwn1DTdI1mvjw370Bl82fAwTlLzwuOv/mSLhb9eNhTN+yWIuPb7jiBjpUwQEBO/7V3jwFSu KhrGLrFh6nOPJYEUoYDOjf//u2JkQXIbbtI8K9W7czJfrgh7rbhCVBwfy+717sKHjVo2mhhlVt43 90mopb5v1UGt1JrujMHyETaQJcFzy5QC8aBYm04e5K8YIqCRK7c53D5izQPFGq2mWzS4MY8B6oNI BjyBK4cv97fRzq29cfS7vsi+Ft3j+N5pRWGQ/sp2+DnthpSVLbEuWA48UtBX+nK+fXOJU/k6EqbB KsL/Y9NZroEcFUQAjjSj1eS/TA7XBsKT53I9zBO5aJDqGiltewSog0gTj6YNTUB4QihOHZcMm3L9 W/+CrEndm3COFA/vZmKw7RKGFHxge+vKDyc4Hhg5W9cm0WpUqiJnZwFU7C9cBLUTTO3Zq67aErqQ QLiCp4ldTXCFL+XCPZ2QEBck4SmPy9XZ/MxkhRNmKQpzss1omybWVEmLmAxeHfos9JMphbeCnXDm JB5sC1t8qP8vqzbGW0ansuxvvZcDwnsTfO+63H2NW1SDWOqfol4AOPm/THBuJaF+KShP2FYVlzur 29i8aLTZ26O1bElvDd+e28e8nZam8j1JiqHt5RUq6P+m4wV4BAzBGDOIGOHqQL8DvTQZCc9WVISt EfckZhyY/cdmSFKQJ0CV8rvUEvxzMbPfRqNNJNcolCnT7SC6+mKZyYLHDms44BjTbF45qYfnp8rv Pwus27C/Ou5chpfwLz7mOMLJPkt26tWjsLEuIhoF5EwGgklGoPvQxAKayxJL0gKjHadCnm9m8ib7 IGewLAIEGJ7k+NTGUcZeAMBshiGh0glHI4xMhqmzGIsAZ6HBufb09dMJ4A+0lwKdjcacdKy9Dz8T cygLvuziGrrig1fSx+C+uV+Tp+i5ebFvsi9fjp8uInqF1D+u1YY/vWm/H/5TcBdq5Goe8E64M2Kq gSYRlv13dz513+6VqOImUSf3tyQtyBxJJ1G+GfDS2tlrFs2ziBzJVl3V/iw0J04JD0DZKeoHa0So JGI6F0sPO2u4+ZcY/7OkhzOuduc2kfZpc0vqMK/UEEBXxE7Qn1zc3qmsuZrGAxe1kG67Fd5HYkum AalAQErg/hv9HNUSSPU9VtNR9FMdmq1AxQ3G+8+Rbr5d82mIgcnFjWBFIPgYT38ubyv9jW/6qQRV v6B/lJkCoTriwK0qOkIofPgR2cYjA2fjnIuPrYYBxYjorJK6G7YOezw4DmCZeqxdVhhXJD0eYMBn 2N7Wd5N0wYPCJoj5/Yp2XAZGDQplbmRzdHJlYW0NZW5kb2JqDTM5MyAwIG9iajw8L0xlbmd0aCAz MDQ1NS9GaWx0ZXIvRmxhdGVEZWNvZGUvTGVuZ3RoMSA1MzkyND4+c3RyZWFtDQo8RIKOhPQqfs2E 47xDgeia7oSTeacFk9o2nmSPk+lWoDeUycmwI0cwELdGucAXdOH5NGy/7oFArDk5b7BxRo6eG/JH RH5Y+O2SVeX4M05VipYxAtfiCbVZeGj7VCm0L7/D+VeI7K9PZ7/EAOsgSYYuhHgi7K3Hr/DcwT70 gW7TN8vDB46PtnA/+EsY7m+qqdWydIdjk05iKWkJpLF53NorFBxnu9tZPngkqt9a+YJ349Wy04oY moRR7G0uf9vGA3x/Uk3e6HLDNULy0uFb8rQPCo+vo9i+GNwJ4VjWLZ6A2ySUWtjk2J1nvPVKArFx GH+x8FBzx1lvIXETZEBxLOhQVnduDvKPVjb4GmjN9DNG2KXnRbHZvrDt2lsjCIzmaxqoqNaKnNCk EdYTThR1yb+/HKismMCzejrwcrmmzjcELZl/D/kGRl+bAwQUoyccnyzKua0dZPNmYT44headBq09 ipCbh/swu6kGZfyoIMcDD/DXVwsoPt9vd+sP7gQ1OUElJ/zn1qb8dpfdId0ZTGl07yzuR1GnvDAC 2KpDTHDszDauBbWHORZ+7OLjZUVTzRkiaFtoTm//OSvdDk3iVJPNATPcnMu1ERO48G8d6cTe/jgT PVFMHVtaq0LPle894lr0zs8lJd6zAiJjkXxaAf905cC67YM9hiEEzZkzpIJSk/3irELPtxE5oXte 7iLuEgFAjvima5vKgq0j+57ivYaZ2V4JohPjw0jZBEgIqiYLj8Z8r7xWBIWaKOGn4IrWvpQj1Rco 9rXY7codPpU/dwL/WVe6/+EDVh9yEsZw8ozknWpST2LyVTlnzsSMN2vSTvvJ+dLx1GxlYYm/O6b7 wiRRMVQMu7vO17+O2O5Q6QbJUnUsUjRA09pmAeoY275pNYqQtOZcQjVNPpJXTFNiRvVQQ+tSCHWI RzF5cpyPU5S17SCVjl6zRiMOjxmzv3sk3VyVIVT4XdQbIa3GiqrP3eTw7xK3Y5gsc6L3OvjBcT8E jBustvUTgzrMVcC+5I4uXSvHel7BGSu1JNWH4wE+VC7/MFF7DsysZl2oAycT7wQuJpC/L0rmrLBU AgVg/Qk9jucqR4hvMnAItSe7IiV+rVOL865+0za6vtoXUIMEvYF3XdARx2n5e/mU9s95jLFdBFYA 4oorQPGQMrN7zDxzWzsxl59cnM7HIXh5+uuTAqEmMuHT3IX6xCpGXtrNDGX82u2C5wfPG+7+WhF0 XNcTL5qZmNt4tt5MDT+gdGrKe8iR683+HKqAMHHqe6fCWhpKpMRV6PfXenKdEOEdshNTZddKiYQT /zz7aYDWcQ2ylowh8Ym8f9lHhixcsouvXiqn9jjKH9IUBmZnh/t/zLNud/qm8tbeo8h3MOGvEbSY 6MSszckFL8fHnH3QfkZmRsp8+FiMiVURH4RGaWG8QdAKJdSryPclG20K6Jxx/bnbk6EX/tkzgwPt y7FuUXJTZXEHDQ8hOmdYIC2DxFgPHGY4+wr6by4U55zsXCyhEGNnj3uBkHh3JroR25tDsuKfrtrd vX1sFX/0bu3+q1A6xjN4hNyZMQ/Eeq5Bskwz6FfQ3mjDut1eTtRyigbAUv4WX5XuAz62r3GGMyQT wKVLS1URwwdTACqfigDLs51c/KXGuiIUFNqMkhhB3dL2poG/6h910fIQzLre7d6Y+SgrmZTBRs39 PZw3aQZJ4sL3ptAdv2rGRWD7Mt0XuDgYyXLsZtPeIzNgF1dN5CG0VNHvQAI1SN2eVSZLpMaKaktn sbxXvk9jQOGb72yy66siiYd6VIBt2gd9HS0a19NcX38taZcaaA+/3ZlAm48SMmcnIZkHfjBuekWH XlijQIz9v4znmmfd+hiyoEI+YSr0vTVbAXfs6Sd+oa52uTXOKu97/BE0HSdNkQRRj+TwKrvKvLa/ d3Ly8HveZ1gwZ/qKhPi4bKOnHOs0pEDAxlP7ClR4OYVZNc2wRB45QsH/0m8ufzZOp21oeRf34xaB oqIGfUOjQNS1LsGEme4J0WSktaHLf0bMqrAZT40PYE1xh2MS2ptQNEbqkVAauYGyFEFYxGbqUMDn 7MEipKSAhCp0un8b+7PzZNOAgVNN8Xa/oS2fn+3e1mb3h2QEe62CbDSdBHYDemTMZFVktweGX3bs +7sOkFREb42g8xDz8i/U9WOG/WlwWfsCG9bYM9O8A1hEY8WE9hlbbqsTZ3yf9+K99P21b8UhA5RD jRCvL/DWGW/ZznlOeJTLTr/2xricg++i0L8TJx8VWzcHWzJn7Y9pOnYRPrUuUueAKKbBy7RPZzEG SEAizRgGLVJ8KCy3F3/DX33wtEj930aZtC2iBpvnQ5SDULmFD6fv0vIhbHjeOJJbFL2p1ZOBurlj ssPrwrNz3L5WgWHKO3iZcwdEaTIexorhx2RZboQyFXHwL1WFXE4a1uOzxvjCusZWNzvAdnwCfubO fMwm673/XdMgNz3MUeha/E36W0RAtYln63ruumdFZMrxnDA/ypjfpPRuozmEiDGhpprdWkXZ2OKW huzal0ReUUYjQP1g4c39HXlAPplZaX1+k8Ht1yukwe1Rco9IO1P07DGl6So9D+3C7e7qdwk0VcT0 8SfQgczOzGLjwHi5s+E18foi6sG2K4IlYkYMPdfZhpCNqJFMNLJf5LWTmeJNIWLCcaO45haxuvKL JCuyh+5iAWiESFcjMH141oCyG505cn7N117h+jG/zuc2V5PnIpS4eUE6gwVNfDhqh2uzTUHz/kDM 01yzGlTKQK/OmXLbEYMdR3/w8Wx4BiUnJhKfb7N9g7esgbAGJQgbtHuI3zxIzRik+agyMQGi6h4h cSVgWGX+WIWKWphhoOjhhv43WmO/d5HARj/sCkMsw9YLxCIiXE+7OwWDhBMmoq/aKX4KF3cY4CiG uyzLje6TqrE5c4sQNfyqEXmWiq064aeYl2kuIl7X5+rQVsPkVKXunxM3wuMqjzY7Auz4mVwtQUuO v8AZZGXLLUoxr+CSGyWf+lvN8Lmc3DaG02ZaX6WbEr8t566YeKhcn+c/pJj/nhttsBmeE3N6tJD/ 2kusqtKM6iavetGyXHSjHHPbIllwNlrgruDGV/Q7ApCXwDe53u0fH8qjJ5F3/0tqO9NnXv4puERM DebgNskJk1lKt+aHvqlGAcuux0tqjGsL/vkGx6Xv7KbrkMIvx5POXFVCIuvqEdHLEiBIBxpNIAXA 3jU+v4BiltNTCd7EjKZt70npVM++Q1Cn3xyxO7N1qQ3AaotuszBmGkJnamhH22YwxXE5Nn8v5o6B S40OeB4/CSRB1vuT3kf1VvyL1jVhHV+d0+SQRX71qTst3dMxAs8yTA7SzMFtjNCQCxVPb5IpNbpa EzuQg8+gNIHJhM+ClXo4n9IFtOJOQsngh7udPg6LWmgZFgIaV22SnH+Jhw9vy7taeGQDfhaHnF9o X7S7MZ2tVNe8wCErVZKvs4p/1ic0jdcT4ceTpjDba3mCvntlGi4XHTne5nwAjkZRIMHXJQNjoSBc 025JY274MH0L8pMRHGnNgtScSZiGRC+Z3Wbq4mK7j+Na8Okz975Kbagw/vmBs6d3n9ozw4be2U7B nbJt3f2EqexQGCTFLzUtfO9JySMCDQ0LWZgv/5FAm6sMyp5s+8jCXwYfHgHsT6K9KFOekMBRDTE0 2BDxLjVmcDOGNcVkHZ4JlP8U4NnDLohGz8EuLpAukttiJ/EmwdIB6JdJl7JpLlhapdqMIFX6M3Hw 5KkWE9wXdzIyAmAKsBi+jFnxwyPieDF5vrZnFzWSM7ccG7eLK+AFm6jC6XD3xgeVAWcEmeEu8zL1 fAHLezKOid2tgKVDbB5UXMhHsKYqv5MLrx0I+0UpmXwuajAUN3M2F4FlOSYhvP+uB5ZRvaLwu/kT Z+f8Lqn3i8aTTvwJDMDRIP0s9dkWEx+zDjhM/ClBNNxWD2MY44LdL5knxIn97ZIh+pZDg5uiDZLE G0VsfkIMZ/0DROahUKVN6WLYvf+PGyH4AF5g1+G9SlC+wneAH3hUZaNYiwJSdO2VoBSxumGDxXgg xltmOG3r/hNrQbJtZ5xhDLOpH72fHPfuHS663VEy0Sa88AQWqpRmpqivhWoWL7pfMfvbMJq2wx+h uICSqrhUd2FV2hvb5nMimQcAFjKQghOx2ELC0WEerMiTvBX+Xr7BuRZLqObgsP16A6Hxfyy/oGxt 1mPLSqDmhfmEERX1UIv0LqdCIwtp7HsDqVTWunXVhssY1u5czMAswbZ7GNT/Lj7o/oikjSDSjsNt Ryo+K4tEiWaP3yXzlVdLlJMLjEIms1iTQpkSdRIRdxe4OZzAqy8Afka8D5IhAXS0ZaCZHKARSI4I vWa7ktm28JT4Ey1cZteRc78dD8yEGxD1swc05f4Je3SCO8v5d18ExxySxw0byHv9x8xBjsiJFDCA aw+Z7BPj/5aqnRaBrK7u4UsJHPmlsYT9xcpx+7Dp9CRaXaNkwRj8hCr1X9frYuQnZAKjEAXBAkX2 GtFrhNPg5Yen/VzW6xiq3JSBblEuJm4cXBJTAA8iaQW1vWQaK9E5AeH5EqZfj3ASKxAfsF2Xs4sA LXJRbFfPfBxYqGRCjV+bE/01fmK0qFGIhE1u7Lv/6lkNMTTTmiyuYtKPcPeMwQeNlXtJETkrM0Qs qv2BirS3crydIf/Mq9skfh0uwGPa/Ez9YWBSRlmnZ3E42uCMRT5OVw8mC6FKQK+e6qocxV/1/AWQ FbFcNHMzi4SXkWUZqSLFgqwZBTn/gkTTSorvblWcBt6D7t4CsZ/sYTWMePfaSMGpCHBnJ3s02gd7 XkAYs2tuZfSXZQr0UM74l9Iq16YP8oY4rySGos3jbgrwClFY7S92zsSGomkJBPKCcvH4T85SfPr7 /S0NERhuigAPNU4Bln7fo35ePOWhIhMdqhPNFJ/Y0yQjSt8pJQx2zRM6D8qHCHax8wxvdo0FXKlR qq4u90pHBAJrQFbwF3Q2sbaNYf07yAx0VVXS2xO5FAcyDBeMnDklpWZ0O5SXPaRVoA8N7pZCjiYO +aqfq5CzrhOGK2tyrjhQvUWB+sOirvpk4cmeuOvLaDrfJfdz1+YiKdWbOukN1+CGNrdV995RAdrz XDt6jD3Moaq2BCDA/HgZ4aftNZQ3e8lYm11Zs6QwFoZR/KBOjB+ZTCbs8BXW7BuLNA4UipYOnESt dhjdhbMkoyS8fsx/B7jdYYR19N5wxmlTdncAO0WGqHiUUTt/clkTT492DvNbsHYJhFDeLvcqIBxI MpLEBIhlDhCIdq2ozN5WBMrM6YnhsA4Q3aP8NZp5pBVsTK0eJRFxb2BQ/7COoiNK2QFj+lDww4h9 zndzrFQIfyBChDGIJFf6Ttht8y+eHnd4cQVb10rUaSfRTtuyHyJ7uepU+crXiA40/iT9XO28GmEu Lb/fX2/lQn9cAza4e2Gc64l2GLQE7LtzAKEHo04SLtx831GSF3iNsvEvExU5nhjyFhd1Gpftct7q bPuH0YRrgiq+h3SMmZ2pmpKBbLkFvArPPZwZMY0ekt75AyGKEeORDg0NERVQhl/d17EIxXwtQKWh 5Q38uJ9Hp5gEPsbQ0i+HVjhIhWe2pi+jwPw7gMoRdxaoHpNkcC4zmR5qPtVvyR/zkMzYx/2FFuGu YfoWzKR8fxYyKbemciztyoLwN6u2qhUWLWuHeZDH+qYnJ68ldNtBisGr1yOIxt9vecgmRJZx1OcO AH6N77REmzAoXbgW9YrFUtsPJxCU4kEK3U+KMwrBz6475zSqLQZmPImqexBl9xGCeru3/NgUt/Xa 6sjVLIbKf/+tvJ7PUmWGPmgbAx0AVihU5ClVYVWuckEG4IHPP3rMt0ILwW1W//raB2oXsEMfPb9J O6I2ZURG5yro1O9FVFE5ln9dx2tukgYUaXuCt6vPbqN/w2uvcGWlqn0kM0gmizPBSlxrYh7nVodV EfdUidRQyxHmxjcuViWP+T+vDRHfdl3LD7y3JO0zecCq67ETgcCkID9tHUK4RUVfAhdY3Xe0g3te AYHyv3zVa2a/afrO4I8HIwUFizg9Dj0eqyqrIqOIf3RRq+wVknI254MQ04L6Ff1bRr9cAY3icsFU wAtyUGuPqJodKJc9KHGt8izKWU8mB+1ugmmeKL19v4jSGgeWQgInidxFg3mGJgI6y96VXzWSk28J hNlJjeUGAp4ihdISJOentIZq6C3j4cb5b3gRhW99nUM/zab6olE6cpURWM+16ZLgfniFYkWkrJ9Y jaW6KmtvduWNg4PRPlUWW7ykwf8w7ev45bzro5Ca30e67r6Y/bSuL7e6+gg9xP+qRGuzph2Jec0V aTcs4Wa5/5UyHlT5dqNKssOcLdYhgWNjXskWX+zAf8lIUKu8CGtgekkt3G5z+Pb23aVnyHpJzkke XeaEs6m39R6K0JO7/dw3vQJmLSSL1l8FRV65PUmbe+Dt4FS5RpHygJKGpZvvunTtcCQiX2O2tasd SdbYC5n8sIikCRIfqVS0rCsJ7IkHjzX5OsVXr+yu/HOnlgCyeA2ST3hbIhZ6cvh9vuaU6bbwQUWB /K/xN6cOgrMXrKtZMHG62jo8cmdPeJn1JWHATMypss89ukNhrsXjbQpOtZe5HEKDxy0hQlNIUN4H Q2CJbLLzXMuLOJc0mrZEYpUvO5hNryWaFFdlrAlAKH6LY1lN7R7EKzbCHslLqgCqut/IS1fuke/N DaENEE5zh4DJzR9jS7/A7MpSlbA5ywDzw0ysJFVH0yNZFZFLOeEbwMuVLdaiZi1bkewJ0+ykEA6M j5GZ0+SVNB3frCSfqhBB6bOQMY0ns7r9SLv5lv+8yN1wzh2DQOtvZy9PqPrS2kXUrBzXydMWhly4 aDyN5Tp3GI4m265lv2cPA/coS1RczMgF/BKl/6DeUu0Hl95IZD+UMs1QCR1KdLaU2m3KPBTM1eaf ATJc94PzGputCpzgwHpSoTbFoTXiWTyxscU8quVkrEQzrXptzrjnpw7Xto15raHj+Hx7Z12wTlVv XQw6HAR30bGVOk+t3+fSHAAUMuVqqSHxqslpvhfmEq/lOPDJrlYm9//ydHdiOsdxeqA8ZrGtMAZZ XnTNbBvN9wogvzatyRoNNnG0YGYZ9VXQnMWiFY4s+dEu0tyZ+7sJrd8QB0yxaixZKLiHwGB8eseR II07F30S63/WtcHDWDNBFxf+lnRCuFwCReF5cfGcWRmNMTjc0g2/V7khImmqWK5EDxf0B8Wew4pB Bhdkykbf1sMOPROFB1Hu17n89Ms8R2tVQE4rP50dHGcs9fnqlj1aAu+WDtPQ1DYmWkcCidbpp5pA OgArzOXWyf+UHd6l7G5Lx9GQ1rQ2pDYpNpx19ycIsQXYSi9GhINkqpsue27lFhrBsJPAf0qxmtzg lTCJkpYPJgDdb55m8CUpY8ylr3F6BzZY3+aJeh+lgkHvoPQ1aUuaAO0SS/eQ+CvCiKo+rwQ2EsCk +wdWM+UjTq70UtHlUXLfx03NyxFH70bHkQDoj1k5QEA28hji+ZiqP1LOhScZrBwjTogWqPdS1atP SE8ACrnYyCW12gY+A0xnHKEdIt3Hk75sV0Y2wjzO+xvscOi6CbQAPrbGHDG/WOEjbn/xLAsxWbQq Q+5FPVcWzs1GkL9BWbWo7CUkNQX57dfsLj1Mxh+1dL+/BBC+3issf/6/N3FlWlJhfqOaRI54+/8a ncYNeEtsPCehJ7xzud34Ijpwr4yUR9+qEqqLVXXW0Ban0r3VTCfpk659aKj4pTe+9VXk75e0xmeD +fzkSLD0PliMGLYzoUOvYpTMCSjTKGg/umb/AAYHuDEXeqHUhOxtemjO8sQKWnqqCbz6b8B2bUfh sztC0Vqxep67XDcLc3rKU3dBnGu7yjX50L6FJZMu5QKNAnkcefNu6phKujUsnyfTk4Hd5iZZydlx kENtMIl4qocaDT4Csd5dIN1kWnC0gM2Oh7FPdSjDZCCl4s8DapNebW+jWCobe8x43Dt693rC+ZYP SiENsnpvm4IIeQ04LvMjnD0l1N7i8RveRHpa10UzUTeLqQqq+rGnkJWUPHsv6IeuKvcqjzeKUD8K B0qsXmmNzCg1enig/RNScnRBUfIkGwp/c/q5jl5byzCyQmHqFxycoWJ32ocuBiLPL3zshPJYiROq I2LFbHESHYuAbrmPjlcLhXn7G4CDsh9rTyQXQ9W27QGCbVeOjTiR3jVllg3bBO5CVoCwJLsV174T 051xZ1ex26bymoCxFnL24Jb7YRVtOIehoay0oXr2IjLwfYUChM8KGt/1lhQ6ViAmggjrOOZdpyIK QS9M/ZclDqFgMhzFZAwiIrioGL7c673cbNqC94+qQEH6c+myY+mzDIf4ns3Ej3eXkgo5epXCkVeI hWMos17oI3+RgYDPtL8OfClF7BLoxTVxekEAOxpJ02GPnT+KV45QkFnaCKyi/OU0E9t7CdvL3gad 4RcI535W0CzxBurfeH9yCrJWsI3+u/fBD0NnBUexmiK6IwU8d3h0B571hWHL6b7y+Ocl1pA5gHPk +AbdDw0Z++XupM90RD6lrkPqq9fUgWfsth8gB7cdfscxL31PT7RP334J/tnhrjQnguVq6N0CA2tx s1bSxjgUS6irEFTZQqa6izMmA56efrdCU9PdvDguXs98bZ6VPqvKxl3uwBgalTdVn7RKNL0yUxaH fM6u98p1fz1mV7W88IkYzcmRZodjA1IltJl9Z6cluvltg5/7z2r/Gw6MhyetkzPZ88rS97Q6c/b2 RBB1UhHXO4CgTZCN5ExlxVZ4d4+EdnMZs1L791LoCDy0EPiBeV8386F1UO7Cvx3lZGQMe4j1yTfv umtnqFkR6zjtKzXVfYMjaj8gHwkyjjVEdE9rR7Xrvh8azm0WmWGC6hQlgw5UCNmJBF8OdkpZitfj e+i0JZAjZULf+OA1m9Yqe8UyAHDIeQa3yll+PH8SDUUHrVoQZvb6bWV5WIqzOcttcIg1wh2Om/2e b5n6ho4LL4uV6MnG9J8lYvsAs9RmGufN/Hpre09PRAL4bmjobEPSTTdYuV3YHNlgeYqIxccSyYGY ttqbhi0SfXiKPnGGA+pPkCwugwEu5CK2p6OMS/2VL9iZFaSYV/L9sLarcXgpfFgKf3pc0zqHtEge ZB0HJ61hB5TnplbRe0ypHDFmxZLTMlz2/m0avVGpgpBBhpgDwZW/3K7lBy6YElCvYVImZ1nVfb7/ SoqaqVSaD5h3c7bP3BQNEJssyj6bHps22HwoKkJpNzjOzXRnq4qb0v7yK4XZsj93Zh20OipDrYIB Syv5ESDVfIO9xShyReW6Z2T1UQrVCv5mvNIDaCT9AjoKF9pEbesKjBacuj0I32lSxEzs/GLShLO3 81tXEplW7RYGXHQkRKYZ8e4BtpbQ52u6KAcozQ+ldj8RhdvGE1zoZcJACVsptMcU5oF6ocIb6BxI /JOLeG+t5IQYGQs40LM1yVRuQj6aJtZGPQ2PCUT1LirWvNHOaHaDXHcxLe3MKFDxHF1uLY4CaFl6 BG27Ri7mC32RhYDcpIfDSdEVF+qjnVvBechn5bdcuPyIrYwX3p6V18vD36mTT95zN7eknyRf/1XX XXiZvBFcPF0JGlLt/lbYF+OyRNlcyrf3FK0R8LxCUJn2Vm0DyMytCe9Gsw82M9FcGYcru7fUBuzg 8NxfJ8BPjp09+e+wRqfbg8KW8DTADAtVj3YjC+RAy5Jy5b81e9xi2vL72vlpMsUOR9Kbn80ISf3a kHJEt+iUGyhNaJ6lsiKRzKxntHz8wRNMWkZRu+Cc3XW14FZzxei8lGRb5WHydleaLeri+cJPTZEW Y9gSbBpMaVN72AoZd7eimQu9sD6xWEMo8sfUukfLksL5sSKDmJImLMYonrl0b5fqa2oQ4Q1es8f/ Al8z+5C5U2FrRCBsXREYD3RCpp9I5MV2y4s9ImcLPHXf7Ot7E49xBX2JeL9OfPXVebg96ef721MF 1WDQX7U0Yl31n875F9c4wRaIjx9403eFfHk8crKjt+ELGq8jMiROP7LEffUUTX0bqTij5QboVN5n 617TP36iW9F0eCANL1CyM+TI0cdWWTMLwuiwWulCvRRHhyVQWkqe+PPUuJq+IAWNfVfdSb12pR6F 3T99ttN7Az9c8h1HulTV8xY6+I5TrPAk5HXsCctQTdbVlLTiGXT24UTCdrRWSLMLJP9msfQvpI95 V1/HEIX4fkJlzMvawNkrE/KgXK35ARtYpYBtxyMHPFjaZRQ9EMoy9qmXok7ber4rX5CjCawQlX8H PDm/tcpOIpv2ilC6ZTlDNoBRfbl8kWJwRAyE4hfOeDr5BLYd/VISyA7zmshiTMjeR1ZuFu90Fu8h fUBkrlINkkqgOdHeUzRCbde3br7eidYOTQUrs0ph6oeC23ky3mk08iK4pnp4XyGa2arNsrgEE9ic EiRhfNXnQKGexncBk4NNfCCY4FnzM8qZFK+fN5Gjffm+w5/lyqAGZyH8Q0kqe1H287KztWIE/a0Y 2PjuV5NNhapfvSdTJefySPeJ7PMv7k2uThRIALBqkuPwcVNEJcr4RFrkgTLEhLvmpjvGDLbQthj/ j+ssEqKSlyI9ab8qduZibnYmLeTfJHNiiZgOp/cM4KriyCLeeE+a1BXKsoHyOV+KR7u5B7pFXmk1 YiIBsY5ejdC9GrYKgB6I8rQJaHrFGqdl6uY9Q/RCT3OzcGcsnSbmpVLY+4LrwVXBfPQiCturBzBW 1nrTqk3OWQeidbH63tpc3j9Zr8OF9qAldEn1Mg4rcabSXddru3z+MqUKJNfuVFQ/sxVet9z2DBpC ntNRUDZlf0D3JQt/lUdAN5etdJdNZFHJxSQss6oYiY9tCkjKClp6PMdUn2mnxOuvXx0YSkKzCg3V 2HLbk/przuSG9BSrKxNCr8njSmgWxgBnf/EDZ68+KK/kDJAo4Pvq1bGlJDIEIRBxjhMbrnWyLp9H s9vc/46AoWCZ2DtUxRczuWINQYUvRb4e1RnX3hbBKKSU62Juju2lf5vRc4UVYrW37LJgh+IRuw0O w7jvLTMuPCtZOiPjjHwSPgTusamxmLH3wMKcZ5rErAHDktTmBJwH2tsefJ1Evsto1omc70bOQGze umMR0pmgH0yfoNzK96WirDH/9f0qQe6M5qzlsceYl0ybzoBpRroR9hbQpbDc49xTn03WkIZgujIy CdD4S2J24KEjiUiPlcZgY6V+5UpiRpUgYF2pOZ9WtyTOK9U+H0P56ai0OvXaSFViwIhnWPm+gJnQ W2a3y++FuXN+7em9aV5c/Zbg/vMBApJNtZglhGsydM2MnSygD+fsBPpskWrlggYzmPsdELqNH6Mf czO3p3OaxIBdZkDgpQEgGatr6AO4BFaJNyR4XlWA5lbsiSZLvN0TMAq8dW79qOXKP0Wn/gYPbo8b TxkOPFPCKlo1qSalrMI41XeZkYxk9w/0uOeKVgclrcx7RGpMSFxSFjQekKCgeU4hkD3j6MtueGQU 9Ky15UG4yipj0nBmimdRCqVUb9AcQ0fN1gCswqwQiz5+NZ20i9qJLSpelbZv+YPQeuP28+NGZ4kF qqdnjgQ86+dUIOZFvLGubbmgXtjF2A/ZfhLddokWkqOADplLvGGoqPnh9rXd8LiceBtkmp/x6y0I SwD+0kbG+GhzL10EM09jDkSrTLVIOtMaeIQsqoJDqsk3/1hGbCodtUQc6NsCQfkouzt2G5IYlqKL Ooq/rtPXmJKmieZoCCUBeVPx1asY4P9I4LFDuuHnLwIckztMZlKthAXl8zof9xwHlg2rRZhNySFd gESQaAGh8d7g+ERlmSeu0bJzbsK6OPdDBZLMT2HMPLYzXzcaA0hhVTJ99DdKkLrTkTw6pEQTJesO MAazVm6pkQUJFvHEKSNHEA4RoWyYM2tayv9Go2DzcMdvOXiUUdjfoRXjny+8TB7Sn3YSNv7rdlxo 1kgy8LMl7ANgXBS7m7v7GrQtZLLRQ82jmUTYLhWBorwMGbe1ZXj1V8SxlbYztIwxny2qN2QfapvH Cy2FYMNkyih3XwUwlrWH7oMw9Cqw+H3o+zzXLO/Lu68jgn3q11KjbBb6BpZsEJ95BgQZMv018Ga1 tLLOeOwQSDiraZYr8f4PYncxjfbRM7lS4C4Zs4GL0uwz1+7YzCYPFlXPAA/BivxbaBAUJdxatWPp M43yxZ3TzGA+M4r1YRaeYqZGXdjacLee/4hURjfwtYcxsg4g0ujyGhUWHow8SMlwB93+v85TPpDi fFIWpcD815yYk5OOSBEli4SUzzDtuGGfE+gKvAIL3vS0vRqveAqCnm9K/xmXosoVxUx+GQORlK2E qCQd94xxvBWpgecajq+zK2H2zSUDC9sucsaPy4Ovh8YvpWRT3I4DprJRlLFm3jhluN1QSxbay3je 7V7ukHjulXdhEmN1y/gSgi217jRRXGrLB/1w0W3Ry6afcq9qCxiNlx9aOHVZxU1R1WN4DOIF6kn9 5GOxmyE6s240KzDQhXsQbELBAi/nAj+YdnFBHqRUY9O8knGN43GdsgAQZtwrN+7hK/Lk2QDos8O9 C4CQ2XJuznNbmO4p+e0/CGz4uw6QSxj06olze3MHOKDLYyRe3Qv2gKW6o6mC5XjN3yVbOCOmwLqi msQYHf4A2LRF5QuJax8tJ0USnvGzWP2bdhf6pda8bgeHoifKQd0mCwaprptG0/X1z/XB39FtKuyH caRD0v+IN1RFKYAeO/kBuBMBzwaeKAJjkHpAnHAP12jafiQUAta4ilZrH4g2FZeiof3fNmcPTtje NVtbbzEMOOt1NilB8PNAnmWL70dV83UCR8neY8THiYKC2XGotkM9DfGHSt/9RP8MBdnRKQshsRdc Tz3dQ5TKNUeActOv5bfBWPjwuPxapdoAuO1NdcJDCqPaD5wfIDxOSHqkez152KaEGn/LIMN2UBYG EEQx5OiAxtU4UN3NDMOWFg+wG8lvbd5dBILH6NQJemTWZmJuuN4FQ2PKSFye/ApUwVzw8OROyjzL PcQ6HfhhLT0+7sx3o/1K8U7xpe0MqcRoUtkXr3rqHQTJTYVjWc86TVCF4WWmL9i/IAdRtxCEaQce 6TnSCxnNHqKO0keOs1gNUEnqOi4PRzsO7Xm5rVAvMWfcSgfTe65jYuEwTXdmpyCEbH7f8cB13xGQ nZG5N4aDcfibKmJ5+ElkNkxIcj3ElfHYeF4GWbhXO94G8ZlM/jmZTczD5GomnqkAT1RgEJk4ZNpr WvC6MB+myGgOYoWhe0IWzOvieUhA114rJBprPreseXIJ/iilqqYrJaz5lX2M/t99k7cit6dr5C+x Ms3NxCVM2PVKnciBDb3IlEZtWRiRPhSc0xywgGW4psInCfcltv/C2ctMEfE8Jci2Q2+S3R9KTh4H 6t3dpibQCmJ3Gz198HifJofyg8pbJaAOW5zIr3R2cN4MCUNHPJvn7SgmFELkmHdxKKIZoy/tIUJM pKkvdAWY7cyVqX1Mpo6W1vJm7xCz70RgYpLILcqTkwj3tRemSjBbxp4IDcKh8mXy6r89+J4Ke0ei emG110wK02gQa11TdoSKKI286Wl02b+kzT8Qsl2bWEpHzLyq7Djp6fAuMGICdbHxYpE0b4FvfukG O1MV1NB3RZNaHkVjj/OUKNpFFWSgi8oNaoNM7c4Tf/7tL3DzxIzas46wCzYION85dzI0R98BP44c QuKe3BZeIEWB8ZN+TCIwWnlQjzAd3oom9Qi0Et59AMp3x419dzOTaC9175Kh/kwWq6tGtTC7BKPt Fl4NsQyIi3jFFDNbgtHfCUbgXGeYCgunvxuwbnycubX3lK3rEZqTKQYQwt8LRJRkhM03ltpruyIM 6jclIhzHceOtN5fDFItnkaWkz8ofmJvsMIR6QwMPEg8706Z8BSQMnOWipDZcWdpUi/6qiKjt5Lez g7er4Fk7B91YMi9mbsEDuxfY2T71/quErSBgKDjye80STcvxwXk5TFa09u/qERcmx4OvqxkjygF6 gDFZ9Klll6gNYherSPMHgwxSLEDx/QmDRPRX5YTXSqgg6eJgQDWGUUnJF2glXUanDE1gBco/1g81 XkysA+eTqAAcIx6MYjDY8TjkzGjJrfywgDk1ckUqAmsVXfRyC6Ndmldf/DUpEo8VXfmXT842MPoV dR9qurR/fPjQUNDdP29pCDWOiGYwnEva4nkoq6De/IVNCYxdtL2GPM/ABPB2QOTvLs/Q/1pwuK5B rknNmPH0g63JmkO35mtdOsorQn0HkH6Hzyj+dyEqCMrnE2nZOW93TGsnRavp3ACAirpUDLlAJl3k sZvGoAOtC717dorjXeADrniVwXIRbN4uBDcwesUqf/Pc7oIm1m9rmMKvzCOuS1EFmnuqSHIMBrnr eACos3BIOIw4+gZxTXMgkY1fVA4ZqSpVEYdsHkGrfzn9wpOg/Kc3SZPVSU3XmWrbPGCd33l0sHcL jfxT1CGYIlO7DI1EnNb77F9XJEi0bGOYimEOOdblsPbDki1YortH3KlAIJMPpYhxtLh2FXqSwvZR LXsYZYCLQYApXG3oK1D5LFdQuDWhw1MehmhtyIaTbUyXKn4j8jU1Df7CBEf7pPbfgeG4Z8Phrdbt qXZujU8kXWEYcbeP1xqXFH3PopWReCfAaxqjhSOjasEymQMTZon2nEKrv7PkeawTVlocIiycajpL XkGwhDTPQEeQ3yOdh12nVF433JNA+0yA/eXJejKrVVw7H/NU0FU3RUr8YFE8FN1oDY1xRiCY1xSo 8fVefacBchLSiT3VJ4G9Tqo+Tt8aQmHGcEsbBHPclQWNQTKzYmb+UyHhkNv7UCuDWdnW9JXSPX8c gkyfcfNP0AqrevuaZyyt06LRDi7AGoO2rgOz+buwgBrY+i6noHkHdWjyY38SNVREnCGjYjIKwomJ Jaqi07KPOvoylivBqxMYgFJ8ZbWQn+lyLoUQdMU+ChcS6HLFfstfmpRaUi029GD+9jBm+bwJkL5a drqCCKhE1BY5EH96zZnTC9CMiroq/LL+bGx0Nt+hOWl8pBuW7kUHatKGrKWhB+4mhRW1Sn6xe7fK gXLc586wPufhv8Qea93zA+ws07WAt/7s4BBdIMtzYS4PDq8zhRxYfb0fs6tIhc7FRRtb2ZSrvoNx yAtdQt6vkebjRqnIvF0y6Ezrm+EGK+8Bk6R4J7VlmtLi499U6oQoiPbQeXfqDoa9QY+VRVp7ltoB B+DAcPQ6OOb7brL8uKgMUTrED5PXyz88hAUruBSsD1Oh8iBdiAzXF587yLNvE8BWEosW8Bd9DUpL s90b2P3ZLRQV+N/IoHmzXEOIjbmSiCtn771bLIdwVBftl8Uri0tdwtMSN9OxXo7QhRM/VFtLnJuc Imj6+yxYzsqdqilNdZWSK9wK6dGSZfY2ooCopgNhLqiJU8p3rE0VJ0QNTC5PzzSMEIuJiy8domOQ d+cuvW522Iorr5pGRsy28q0Iy4PlD4wl/N7FCPAbmu5ArfGe/ge8RJgrg8XOcapg2W3QMay8crjK QUVASqYcmjUQS8CLVUaePV200I8uuPD/3IX1f3IOJ1442SwyDfdHKOM2/hHH7sVsq79xVFiBKizd QQfdv1c0mnxpHVa4G+JaD1Cv+I/zgF2UssiGOfnY0pBOdqe3l0sNchcKopOhyIorRYRgoBfmHw3j dRc/AgW7zDILzoWrLatNAB3G5EOJzmkERsyLcl5N2vgDcl4LWcqWrg79+fVBoyOYWwjE8FLpWgE1 ShL0sWufoz2eyMb7VqQ6rJC2WJ23aFPDrGEEBlF6fd8wf/8Yzu/qsBu0nxHSR5BFAWy/aWCE4GYT um70uzjPe8yfVSwoZIw/zqDqydGI7EgGj4YBS+HpdsfMdl4+a4DGZDbxZNusZXxIgAAprPj8hLES mgXCFRfE9igDUGQnXz0NZAYv3CoEmBj2B0VgcCoMEmHJ5FB6sYuCWjNDcc5mxMbsiAeu+VJGh+Ce R0p/uDIOYE28LTcGbkjQMCaU55wH9TeoX6UthaQx1yxJDF9J1zMPx7ZtdgOXGn0HwWW36Ifja3r1 LN5MpR2XqiLVEAxwZ3KdjLm2SUw1a/WFyd4Ym5jISnYWesXhyYKnemEdSHpP/3rCg2RoqFgoqsKm kQUKGkR0Gqw32O04z/B159iJJigeLbD5GmcSv534CKH5i1TLI5ffz9jvBX0nahmW6u6PgCsI5Bga HQHebrKULB2I/HZYiMl8gspgzYXczTFe8CevagMe22fvH0LSBPg1bRn5tBDaPwrARjR7KW43b9X3 RDg+KLjVZFYOEGluh7sgMDw696RwNxRNiM7Jeus+CHAsmfd1S+A0hbGgv3Zn0yFMHUBAPkiISG8B H+osAOUNna1TxbqRzd4ZtGLJHYLYIjeHXFaVr2b5mWG5XpogH6YOjZesda2FrB9yWePqaxsOLrw+ ej79+xReAL/0llm8GC7HWgda/dKUENOKY6D91SAHPdCwVkKN+E+PPUkAI8g1FBu0FrCsD4npOb7P yf5d7bt8ODJmx6pHxB0eimgRsEGbKdY+2YAPeiK1jPs9cRJEoiFyIGgs2cDjK63G9w38Oq1nPEKf 5UZLAGCmtdIn32vLcNhuUC2VBzBRDgvMtwTN5xPwVOfAFR2RnRa63/wS7CdZyQKi07hyjkzkEnIU JNyHuFfefW/I91dqea9vCfh5O/oBIIRxJazgzW7y1CYDmZJ4hZ4C8OwmfTWVrh5MAgNqcR4xUirE WPuCPBYU5UXOH/bHhbmHOAlEgL6Muxlw1ixLdg/9BdqsKmaiJa9ncxpGkmxa4tjilaFn/UvMjrLU UPmGbaDgt0oyTRdutYl4+4+qLlxo04JY5AkkRSlanGoHvRZq3nK0K5SSeofsPEUSmnyusSd47TM+ 2TtC5vufcIZHqdDmHnEvIyeH0NxjqExr7ezMymqmAvd2D56bRfUvli8TjKVq5cft2ZYsWZrTjUbz P+aCQhxEoS38waF74nsNILm4+Vu8NsFdFDV1pFoT89cRX8uZsghBgx/1l5xhgzv3mmxf/selniva DSSY5BFJP4OqDouoVX0Vvi8u/51d2/d3WLxiFzBrLEQm/gqM+9MJAwdLdk9x8gByswpr1poUCQqH LTEGMj2CYnprVCTccaIez4QO1B8rcjPzTFtONzqH+htnKbNZuHFDFRxS9VdU1tnBCqMGsmRfhsYJ iIqawBx1phTb+2heyzvv+jo7kJR43D+MH92L64vI6yNLCNhYUwwDnGciDyWdHhwd3GZcfzKgGtbx XAsqjw1bxu73QsSmzpwqk9nGLt9PU3ZkeL5sEqkdO329WbMQSQqgIjy5gwRx8oQk798zZNyrj8R8 /BHPE3J45kqCb0Ah+Dnw2kjx+dsRBgxj9P4ikDKcTioqlx0rx16MhCx9jLPx/w2NoXxMy8yX7yMR MfkyXwVRX6T0hZkwAvEAVey96XY3PDV5S/4q5SUsqAxblIuIdUdm1s50bCol1vrb7DQq5+zJA2sU wg3QZ2rhVcNDsHj5AbTa6JVFN8LaW+mL20hEh3TluKMIcRme1HuwQhd7J5YBRW87wp/39f/DZCYu AwFxIr7NDNOkJ2VAyb8zBUXqfjrlQszJGc0tVWjPAImiK0CsDjApICeGbp0i8d3TltzRh5tfbfKX aEOBtKv/VCYouiyKxoyf2cQAThf8umw8WaQszXSoNjcje2HU0C4pyIJC1gNjlRya4I992mHLsRwB KM61Lo4c4en4Ay7eozl41p4MnmIsUChBBIYH3DbTn2dtkw8nsfvuW+GKLOsFOmCs5/OF2LfkbrtM x/w3lLrLF/sxgKBJNZTiIO8MaugbX/ABQIL7/oip9RcLZDLOAVVnFJ6a14kUk7GLJWrQng1Cdwlm GZDLyApUIRGcDjaTy08DXdz5WiKxq7wzyPOfOKXjAHjntGFsBKNiDmqsWYnum2LWIAQXpGsg6Rfa poNvwfYDU9Cqwmus5pSBshKnor9ieCgU2fgeb8k1Q7J6zf8AJatangjirPeH54plmghOVvICtDjO Sonfy8j7pNqpLuMNzKWIxXBntyaZLZnrKViPP9UuPSjiv02kjoANNSXcRw/YitW4MH2l9+iHNm8l EiqaqNeV0GrjPZSLvqgqBmcSMvsd0IwtrPz9DmNitPq/TKjmT9aL/mPtMXovEDJyRMyOjAYxEN6Q grWrEoCEKQdAzn2p6ey9mx0zfy4gRu4YCUXdz6S2xkl0K2dR5xFtWe6p6GbEpVfiUkBlGDJcb13s ci3T9IUGxB5QxPCC53pwZcf6Rr++ThEpBwQldeduAdJE01z613pJvozntOd0PNreodCZTgC21+H7 MpqmXmhYQhdIWb85CJvEqttp9jzMCYim4D83BUAJPIJHwc/6XTOkkefy8msmjvBVpAIOhu+CqpAN tiT3wwQ8W1jDUXKcT7OV5QszAo+TejZClU24HAUukfZ/2WBijRFhWV5x/yODqDZYWez0UXgfvrRk VqileUYIIBrR602Qgeohd6rJurTXMsQS+KiMJaSaA4rjFgqCIBy6RKIB9ZvxF8Jhpg9KDChCTyUG SE70MAKqQp1saqhFynOwqWKI0tzsOMh4C0kzMYG3IqfPwMtGoRhWTwAcnQXh8fNQOtVpRcr/TIhH EA3aGhlrQABy6j0Blm6ZdB2trdFRwqidfOQ3L1H6AVvjuxkLYmv0516vBFv/9YbI0ufos44qVZbf OkN7YET+BfwY8h60RiybIstCGMXrjrHRlHrfR0mavJw/GcVEdFCbaF6HXC7juRP1+wuorDWV2nYF mB33+wbYNF8kWr1IHIn9TgFHpxDogNIDLf5I6kitYmk5EYWHc+p3JbpXvACiofNS5s1Z2vDCodeE aZ0F2M/ZviIg3tvstKgV7kOo+kEpOaoEC4ugLOYruv67XETNObm1/g1Y849WWP02W7U465TtDRU2 kMirOEG2/CoAEiUStUUynLk+cBeK0M14SLTBra2trxp285hGCohts+i/+bPY9Z4xCcM5ZCvUEl0E 3fykrlGpDiSVUpGSYMjsJx8NBHiOOXNCguXHDaVDZEvv5EoZStPuUpegY7NSwqAxC4FLb075uJ/i jJpFU2V0queJ89VnPQaC9jpdVAT2flgD0zlEUvC2kWdlGQN+juvaV43RwCGqWNYp4ueuAwkBBhgg zYf7aMF0FMAt0wihjB6LDY9nthKRng5XQxA11hmMNPSSUN6HaEQJU9rjaYxaUqSfEnZ+fY2xczHE U6/RWQiGt4t4jvOfR4YQ6TDtCvVjysJXAnYYBLE7mKoqY9lLE5f/+8C7hIy5gFdW7nWVlnGs89uc /ltJ/MyeS6+n/CmlRd6aO4nbhRKfjVhyyEMyeW4dEgiFB1FFdr9vFjxqZsQktFqqA/Fufqo7uB4K fTgopaaRMuKKOmt7wVfohnOqnQFBHxLFEIM/pn+ek39JLBvPaaOiRnZZ1dFICE6pSIpskvvnvOEy K7aiL+91OqqTs97FiFQTHfxtl2gSTg9bRqIzSAxDMarGzQVQVN1OgOC5B8RPxrbQETzmyzeSfIJP O9U0hfYEw/6bnymmqnJ5MbytRTwbQPkSkCqDAwcVQ6WmDxF44h2EWs3QoxAcCur3SECRqmAD37nK MjnwfpnTNetcb1dyxBTufGKcJIQg0DvzaBlmY7VF43p2wIymSbVpEvCK1uA2E9aPvpo9IBnCGjwb X2Y/4A8cdbdqz/sknRsdwN+oI60pGePoAfxQEbrtXu/pp+oEmMUVjVt/iXnENWAHXw8gWTbo3g21 k46RWzfbzHRJYZcJL37ttHqw6+aoHn4HDhykbxXWUYFjYn6MBo4Zuvc2ebh5LOcwhfPW7iuLpfRE 9qiORbdO6T0fdIQgkM0j6WsOEJZXN0pCfM5ldPyql20ouzrQ9Bm8kBLfavZiLKFLcevXAC9hfhfL MmRIE64JvkeU3mudJJaXHnVlgH2WBRukWGNs+XjSf39hlWs78r4eayF60+JKDoZcspX56S7VlO0d CgjyTf4sGzilmsMNr1sRtQVU8eQ4Vrdl42BVpbJByf2vEOsxbNL64D4Dg+6ZDMSdgoj/yMVBcCBq Mx16xcseKvUWjNF0IvsIJL6nCpoezXR6y1L/gbx/APi3goPJawLkQeZYoW0asAWIjF2zZTzt2eRB 9MmriqxUXcBpvZMQ+9js/EnzJVRJOVkV5C43yIEESgVIy1af+kKTx61JvYUym3NWbMegQgibxQ5K 3L5ODacZtvKgf7FI83gwxLTQ7a9XXr7q7Gkm9mziRrVeMjXsAPTr8xscBOI2ZNkikvIkHCN7GF8M FCFo9doqltyE1QXziE6C34pRtAoUwIpgUEHN5g3Zgmg7AdufEwxxCFe+cnY5mQj1BZHuDVArPA5S OfBkAS91H9RnpXRn6eFFNDmIMmT573+Fr3opnzz3JHN7BX6Dqgd/Dov7m24VG7M+8+sNqXrMpcz9 jLwI/dvLQgKagRnUuhDRXH4bMqOHVfblID2FhU1fHu0WkAtgNiXPEDPx7ho1NgmrFJ5DO66JMrrH Z7+lPnr298g96IPt2C3z3hn5LCI8T8nfcXBGGzY3HF+QDjEeBpeBGgwRWkJbd/GNDlqzXx7SfY3S dHfQvYs/WvYsxKp8+TDGmr0UdRISHDazi0cheir74z6YsT2eNWvfDa9PKWARF4LYvCUSTzg3WWDf 3PwkOgrjyPv6wiX3JLpcVZNHWh6roL/JjhV4Hi8+QxKmylUG4qOatPYuUDAujIEci6tvj4cKl6Zf ncGkEFZIKfc6N/33KTdltA929kKspxCPxQLEXxUZlW+mDOeVGOQgGx8rM9OeQios9Zvf9UUDqmTg Z3kIR7hetkRQVsaUbFO2P5bxUTNEq4jCokdVkC9Qj/m3bMUTuYtFnVNR0Akt/p/OgvA2Ez6FoyU0 4Q1N7YTkw5z0dwmgZX/3nh4/bnOCEpFaSrbI/FMGroOvoK675cr3rPCW4pvoypnVDN+pGh8/XFgn IQ1PdnVJwMSncOJeR+wepH1JH/hvxPLDkqV5pzIgeTbiHv5fJlA50PEnJI2qtbZ+i5nsCJJBH4JV W3tB/q11hCyrI23zRRsgKbX37+dB3ybRf7uTFlP7T47+dtfodyA+86T7ZebIQVT5M7qERfiKGoiG xGi4rdCef8zM/QmAHQMdG/zk0OeP2dYQoCGRgEFdi/jLO0N6TnncPQa2uU/D94dsJ60qJJvv5nOP /4+WHQXbGnhPHbZ/VYml99hA08ZXbzKU4Y/n9fp7krxGEJVwiWxx839RETO8L5aCCIU7DYAbdYXP EjlsffoCmg7gG4T6/ylOw401ZuBhnY7hJXUsKP6980ZkaBr/hMhEAg9PpqSaQ8uAOOKDdgbOoamv /+CdS9H2MEG928GHMVSbXWoAQRrQiGN4UFHME93EtKOrzUyC4HmLB/rGw9WreWyynymhrEPD32Xr AQeG/Ytce24NAgBhsyhkPvwHMwAoaz2Q2UTeZEsvXkcma3Mw3gzL7FI8v6vJjNJyhB45HjWe7B4l F2s3iCy+6aUjCJPI4ZxEICZnMKJLFC1/NE/07qdQtSEldVm/jK1j12Rj0b4o7+6ADmCuPROy+wZN 7YYpLgFL6xydvzCbWLxrExqOvTjtfQ3KTJAbduE5m48/4RjQtAD1UgjswCNAbxOmThGXOx1Jyz/D OdnOo4K8IrLNz6dmUJGvfcfDZXJrqX+lx/As1esp0vvXR7HKJRCOGVZsqeb5s3Psn86QiXp00iLX yMTXEM4Lh4lfBMQ8ZnyULXUZuVwYrEcP25Cv7+6RUzPZ0f9UkfRoMmYbpS06cPmXLUg1P8iCtdNp jaKTtD9Cl6thKSZtDsNY9/3HGeuEXIeA9bNmynZ4+sYanonevScjD58Gc4oGvbPd5Ku+3BO3rE75 h9Eek+MHXIj8gXfVGMjNtHhTKNjQa0FBRbpSry+1t+L+N0ESBeKotC0zl75+aIo9RwzHN1ziScvz ob58KaTfQJSSicFToGWkioNtb8F62CXIy84NFaze89DzaD0nDY3KKcP3rZ9gMl8V5eeuCOqyuu5z u8U0508d74wyRDFRH8ObnjIU/ZVNergr18c4NLIywbKvqODZ9wwl7PCZBfM2//BOneEVdUU+CdnZ 0bQixGckAcHVR2v7Y20YEquwhP2muRZsUZPbCh7TcxgEfTEbholCexlYeJ/LPFJcew1OPJGBzeCe NCrXEUI5TRHnr33uf3dWAuX3Op5UICX1gOfMhmo4xIwDZQMZNfV0q2eVvbIEuvtftgIN5KuwZjUf pfxMYCszmN0Ziuw6Xxt7ROi6fCR/Bs8A3+DQHRumP4HOekZNR8nLgZJ0OOg2Qk42u2Go1Uw80tgX iUP/NdITiR0XTOeuvtn5ganihHFas90DPDMC5ZT1b3LlkfADqOawJW8PMXfdNaUE1pdnBkJeM0pV rBDSxsWIvcPP/seQMrnLwijUt6+o0OkGGoVu3RECW5C4Regf/VgCxcatOk60QGdnNgklOfryB9xZ xlqm/hq34Q75/vvZlgfD0baT+lHNyWoWHqhSZO/snbfz3V77lA/DkKZaBvyJoYNrkojBV4xB1+ze scYYAjQSYIWnDcW7HKl1jAf9ClQme5d+OpK+zN9JOYZsLCbK4YfS4T9fE76T7dAR19Go0ZgXyvA0 dVVOOMy2vjNmzD5XrFjbcBaVD4rYg1YsdPYu0LANLrWzKmqO7QH3fZxrq993ilpNrLNuRt+ntRAQ C1MdtKj4sMFNNO016U8suJQeFfrAM0MNnJz/2PU+H6tMRmidAZf44j1pjEq8mr25oyeruVbk5A5a iqoo19r9V7PWMg5aTVUK2EwMGU0uRfvtIjsdT/1duLpWK6k2DZMli7nbZvToy+6vL8SYXPlPO4p1 KjT1KuuoUvP3L27xfkeVkgFNCTmFcuEUOUDGVfMchDG8qM/GF0eYOoaNB9IsVG06k8d8QesrJl82 5Rhciu4wxkIvPetLRVIjRsyYu0wc1wpMiaUV1djVbPXX4nEpBy/c1/M8D3gwReg+lKXPUzRZWS0m ti0tOFqhqOZ/6q6xetD6Q3nUJvKRiHjHz/PrZPpdg7RIb3XJbRvpBjTyXLy97tG/Ox9yfZh1c3E8 1CUaToXCD3yqK+ph5sToNqrqPlkizNc81xLXcQpAT4vpufm/G7A47M1CD2azTP50umxy5jgliI+U jGGAR+Jchrtc2rhPl0XXoVppDsqRS5hzWlvDE+79uQ5MV1QKZFpHSP/3sKnpJNpugcRiZzu9AeVc zQt4LBRwgjFdP7xPoDqbJINhbZ8t5gX+InBTvf6WQyKH8SJ8XD2GJQ7PX+TclPjPolo5eKd99Vc9 frNlrqMcHVrizEyDytiJ+PDjZwc4gotVgW2FYF/mie3CFm3vppbYb2IXDIcCHUAHCTWJNuYiJvzC EkLpltkfzdoyIDgr4I9yXK1WbM9r6y9wk6MwYvTFWyRJAyNxffsk6V7CegUzjgvZJXg5CfVh0S2z TMGB6EQdFyZEeRr8KoX6BK/z0hTXG78CfRkSK/DxiFDBSedm1dx/7BSyuiGo3KMQnTx98F9q8arB 53m2rSqm2+5djg1+DK/38u85hF9aEtMhB8bzpy4wRT9ucJPbYJclAUEwTlkDv2JVZ5k1O5iagasE PHm9BiyZSSd0oVXk/dHN0vvP1UgrP4H6xzYI+5l9xjXsByiJ+3Ix00KJ8iDTKdVT9yAMBzNO5Zg2 NgbS4VJLRfypQxMCOW5jnTMHyewlC6wwGyXkNmeDKGXcpAMQlrqjJKyCSHzHFZLRAWZZhqMRyP4F FGND02RskCocKV8HeOaIlWnliyvpeKKcrY2YCrdEhgAjXXO+ziYT8fSqiHGDo/43qv/ZKU5Yrlsd IExpmboWyKFo0bX/msA/0aGoYF08plhxPAlwZ3Hvx2tDx9isEZYzIXvAgVThEKTWesdjVya/w5Kw QeSzF2Sx3jm9ftyxqv8owsMiI6W9YmWgvQX0J4pYkbD6p20wqd0MdrK3MO5hcVG/fukeKTocUH2d tH+rhXfb9xUUC+WhiuDHg6BPlu0Fytf451AGZFNpmG5Lfqhq8UnFjKu0dH+Opq0ScdkXD6zTcqni KImmI3T8s/stK/DsKzX8h559dc/WQg6tl6+Xi+2JJZrIThxcySt5c/jFP2mw9ngGFzWOE+9+aQZ9 Q0LHekeLc0OIYnXc2cXO6+MJIA91IHJOGZw3ONGNSqPXxjH4LJpPUzXeORCrwsyOJPUCTpSNVhN+ 65KLCFO7qc0qtscqYfKKkFKFgbRKvq1KjpNiGA8vU7gnqBOB6m870AGMu67MwdU64Ru6ctiYmNJV sRpJ1TNwTOl4q2ldC7zoMkpjtj6ct6csXhOz+VXzdsms31yhggdg178objVFmaJ0LRgWjceoz8KN /2eMmgMNJsabTbKfiqoc3ayVvVzYGK4txcKB0LvMztX6W8KlShFS3JNrKbink5UxnpVP39eY0CQG xKcP8gr3/Lro5lXbetn69axiT9zB9aarFGXimBGxutsp8njt4jhMMGqzGwKNwNKzbo2HNJHGph3h zOVC6H/+I2AEiyVTLn2L3JQJs9o+gESxfVSOUMwS9Dla5S/DSKckmGUafce6e+ZsrGxRtNxrly+y 7rFRJDGQOtn1464Kb7ez8sxSUzA4Q6OjLQmWimpGA2rLqa3RTXzvE4ydwSpwgC57oZDlJprCwBto Lj83sioqxH+pqypgRk9D7uJhYmR+spcAZDgO2iByile//Pgok6ltdwQEH7IutFC7RObPCw8/D8NJ Qo4/tfBkqqVzUlS4Fah+Mf3VMZSDaeWCfSgTR0RFGY4ULTZCo/3FQNm+O3lf1/YtZwKanK6V0wIW WYYO9gs+RiZLCJ9upOi+rJ4rFvpHJiHwWCjuRg4NAE+hpY5yPeA1pQHF3bXmNRWPpkre8AVdFwXu mWex+rxgo2BKSvwrfTnvjoPfNTr0qxLnyUnNC3K7Ay5OzkGiSCKWWoymxOXjpnq2xM1NKaZ9zXJd vQd8diEo+Tt8+0YwGNoJGJiZs9lDOp5LREAPj5mvC3haGxkAhL2j8EKLBrdttNcJzyxrSeLP0nqE D2M+te7HnzUPrB1AQ3GIoO2R6Mm3Mcv52C+2BTeFoB49D79X1dsPlSk59y/xe4iaUabfzTY+c9rk cCAKCAxdHY8BU+V4uwL1TsE16uMhrhAY+R6ODa3zISJbpSJhQNOVktSc6g1T9X11r7eB47LNi3po vz1rzQNR524NRgGLnbUzXw2E+irA4KnmZIVZFJs07D8rJ0TQ+x3eyXq9oAWRs0X/BkbzZiabJEH1 tWqOjK5Rr7+d55eMAKmS5fv1gG6TArEfpndeQQXxyyaEpn+u3IdUltI7VzKTCqhzFozkqqT+7Glv TaEwpnBE3jbC5RGMIXX/1FHcO5yLyNxKo2ci1mH4wC1k8A9vHnzMy8doNlajRZGNu1dL8XoNB6ih zWKAbEEaWtl1oP98mkkRBEkh/W0fFRp69Qi57hoWJqoOTNycLzI/hRs5Zco7vsBVppLb9uss3Uu7 8fAsBEP4e6Fb3EqEaWud6KndGdujc6O6mlk2/HubYagpWKc5k0HRcTgEl0jYTP3n62yPoirHWMk1 KAx/X8FlLhsTcl7SqGA+gZzWTP08SMKZrGkvsMEV+j2Gx3dQKlNPiPpPvRv+BSig13//REZ2y9bG 2umGYn5OeooZOLUtxVj8uFQ/nE61wDiVRY4WSeygff4gi1nqc2BTSAQ3W7WsOYrxMv7/YddyxiBE 3HvlSyP0tT3bUATQ3rxUQLaVVM03K5uUqIEurgvmH9kbLY5NsGmWS05v3N/FTVXqQbtuukDdE+N1 nw0KI0fLW5tbliffHtpFaMYhsTvkwD0gAMmuDBXxc/nvjwX2afubKabjPWapqWflq6jrECS79N9X prD1sTYV2KMUenBSbQJK6+TGgJkZQ6QRGLIxm+tbEx3/lFlixMrT3qjmnJ6LQ7yd8iKHz4NZfn87 VfgLWbYidya2yHRoq2S1gGoHHkmuuujas1AuiUuzBeyI9vrUKcN6rYF4JB66BW7BEWMhCnS/rIhP vBA5t3chwSnRtb6+yar1+jV3TFvZok1lt7pmO+ve6Cbm+pW4yuhnKehwTNY4u2GJmWUBdjjCbyrO V3pY6/7VKzxOfvOawMcTxITrxrwQI0kTjOrXWmbq2z9XlR95+lG/T686f0rltvceWobmZmGv36hH I2jOtPXq6I9t0UOAslZELIEqsRpmK9rCe3fM0oiVPwSYFC4awzJB7Ey90nkZru0ekH2M+yyshAym sfeEcb2wT1Ea6a+QzW12vpgZmbB8Ec/dAudJ3Lp2d5aSrZ3LgVmp+CaMGrlw6TqNbWx5R3N4PhG0 qsiS4YDiquQaphWscKjmzyQ8t0IkibVVe2RmNpiBhnFl7R53xRR/puMZRZDBJcyJHkc+CLcrDx/P 7Vk2ZukvMuxtwistmB1THG1YFLLe8x97bMf8praVS6HBUV9v4PQ3KVMprPaoyY2ipLT+CiCstz80 S19Jb/x40a7Mh/2WaZBMqz1pOf+gAiiTbRQFh3wiTM23r+gzTgyj6y8T9Sk9mIsILcSunJV878SO SsqdN6SH1x8PfVmlWhOkOppkfEwm9OGcHgWOr/ELQ6d3PuH73ufcG1hSXdpK5LsSSmBAr3e9D7w9 P9Qj5mnY7s81ChpfxP7OlKXnEA5GDLNEQTYYR3khw7rYjnNuKqDCG+O/Gelk7I+m4pszOjl9LauW SU99EeluYFLewsiAi2HGwtUK/jMjJ6dgNnu11xiDS00f953bnq4WwyaFpqUy5dcqyVkeDWCYkeNz C2ezXfPSW49Qz5x7BglizxIoLq8ezN3l06XK25BNKYNEjQhPyaogIBukXysOdk+DWPKgQCVg4Ujn ifQZdOXExx0v2Lf1j4awNRAPLQiFTd/qvRsO3HmpsEgysT4AAN4YzH8ZMd2ODs/VoQaVWyZKuEbf X+FC2gxR2tGEUS5kxdzS/BNns50rCVeo+9AUjhhUkgdFcodawJxsKRvsp24WjccEEBEcc2XHScLT n1Gtn3VfOxRvjSEwF/tAxMffMhwBke7IspRnvchuZ7eN37moncqISGQlpBL+QKYqXW8ASIG0dQZV PGOCjYNLIL1cgUAZeza/yRpmjkpHIT3hH8y+eq8ilFrQcvdpM+uRE0DBdpSURrXYYDsJukgvYBo2 9gAneFEBhLxbxkew+49f0UDVh8yeZIRrYGjmxr5pemaX8SB2HLXeNdcWrRKvo6sUV5y4Zr+k+NuR uHd8niEUXP6xlPUN9rn1kPdlkArfRfHFU/ICXJJEjOVv43hw2yqzTGErEs62cw3K45AN/1Ec7IBD k1iescfOwWlOWyfrTPwEkIjT858VF2VR9SaiRUbvnBfSEIEUGONe1wabo3NRy4k2ml6CS5u/Hys6 wQylKvMlZ+OxzZR05H1tVwjUo1WGiPreV/rblHCLrtL86quKprcioHUmBbRmFe3gT7Dgry1gSmZ+ Sl92+c28iK6DZo6/W67aplWZpiloO+ZQVOCoBKHqnCjQtO7izVFtLcsi8gUmpgtmT0xA0QXXq3kf i3HX885Psh9mkqJ7okidCKdk0wksVjniGVvkKZw/TeLRSPlk5nAVxrJ7JvV7F+OGemU2jGyX9ddw 5aWRwXmpozN3UHQf3RgwDdzZ6y4HDmIEiAegPl1X4TaeALGZmqTlBn3IGmLjAs/GCG23ymcp8p1t Qhr4yPdJUa6z7Pze6o9lk2pin0h9q4FXXxhcm9xUK4RiEWZMwthk05fMKbeqJBwrKv92/f3YfdJB i0UrgiyUvYaNhaPIVyJj8rkTIrOm6+S98Qcc0U9ARFxCibgpKO0TBTP5XJ/GXqCPe9ovnPei7ZSF 05o2oYzSCtej/4fQrb071uuiHT60eIfXbxVfylcV8/9gAFz+4i8mkG5shTgGjymnai4MfM6yfF7B +yDTFafgcBbwy/RN/9+mEcj7ymYjNPF3fsHfT/evK9JRodWFz8625uZjLJHjLAh7vaPsQ6uS+oRd W1s2I0oii8yMa36MqnMyuehwwc74to0Gn6GTTyJ20j9PVG0H7Ly5G/eLHSYv7quTIn03pQXeUPkv bWI6qxYhwMIqH5wV1hsuCY6nn7va/Pma6satkHElS+mVx/gn3gsmg157l+Ny1DAOTeoUVVE8tYeT 37SZEBVvorxctej+4PgDgoFpBd/6vGaLZYH2ecGPvzQT5XvJJTG1hhd2E9gqs5AyH9U6h2QSttx3 JJP6WJinueas1EJDet84LCzom2wDiCUEYE83/SVGIl5qyIMCjs3PpN5fdJh0MOudx9dA0s7mXUvw S3QyybjLdO3XVMaV+61Fm5oIxFq+UqQVAyg4qzHI/DsJUR+vPKS7hX4D/JI8Gdzy2jJoFHctgXlo rLDLqF86yoa7crYO/zSk7GEfmQIyjehwUaFHGksLSWExeoaaRNviqhc9n4Go/iPZABUQcBCU7lbL yXV+ZI6Z1CRh9SIOvMQ7Sy3xaOg67/j8iSb3H6DwiFjCZl0W7Dc1MzlKgSZYUesv1QawsXT+oLM6 OTzsHVs/FpUIf5OvgNl+lm/BqXXVsPMMOte/BlCNq4OBvmkdiBTQSX9y6ioiksmVYBfp4t5qYKNI JfQy2mna8cjXiFGz+PZn/PclSPY0q+DbYiNtWn6f41ZN/ufD1snEodyp8csCmB3Th6p4gy4pYDlO lCQtwmAzZj5MN0viRIlen+5pcI0IMI+45ZxmgbA/o5QyD7I9hwdTX2O4VilSpvMPUAWaMu84wgMC COcIt8ZgeCS2RDRe8VdqkfNFW1oN6l9iAW+vJVRkSRqpm+/1qYoe6AAs0b6e6CBwrsiSeeI2J9q0 DiP4Oji4PBZwjyVVgFd/M3cBQTeORZbwL0cayw5C6OTCX4074hngJCC8+8U7e3rtB2b8y+texBUN wmBIkLkcLQu4M3dtlWIpXcTYwSIlZ6fDQR+6muV7vKszWX1NiGNU5gEuzIeFNkSNvtk1+nrreJ3E dPzCRBGo/oVKxpsPYKQg8vMP3LErYKbT1txhf2O67WVLC5CNsJ47slmvq8cLHKujCoOmHBhEL9ym MDbiZ7U1NJIpibnmi/KIbuMS5f39sHWUJq6XZ/aqwDSgYUcqKhZ3xuF+E4RfQmLGlEWe9l+Di9TV M/gWJp2/lV16u/ec8ohGH5dGd3G1nvMzY3Ss0xOVDnuv25nD1/obsytO9Aq5EA4uAD2g50317JlX cNdZE7vpuiBk8rtI2uayjF7yg7WT/aS6uiuUDQrfa3k8bb8YXW/GttEKFWwWtJ//aJg6Ls1D0A16 Tdd1jWsOr8QWbDtjlIeFvKLYJAJamQ1e0W0i0KOaFoGWiCvifwG0/kZUryIfscIB3EEyQvI0lQoB ikAcrRq3PV+TYhCzoUP5HYpOko6gpxXZ5N2lZtiRFpDEMBjlhslKyKANYAQ7BFz57J41CLoa2p1l dFggA7kiaMeGO81P0PB9mBsIHwF/pC668mrYWJBBX8dx3qJTdWNaoQKSOFt0H4fAe2CNiXfQDED0 PQTEwLWmq0MxyLrfxsOC6GAQIzuRMfWgPk1u6qcZfXCT/eH0fr/jYqCPVmF68iJr3xLsNue0Qt2s zGagCYpGh8xW65OJhdpqnoTiuF6MSkweeL5x6l5i9oCa2zkVZmfgQZdwdmxUXRxEwFB7DrR3bjCO g+c0oxkdUp0OclEoCnYloGn7YpvoCurHKBCbVGI/1YimOvMURiMLSzh6IRXettWpU4lhl9PnR6W9 kaiP82Tcc039NC/AMThitENsNswXG91CLA6ayrhu1iqrjZVEfzxl/g237gTYY/0KQ2H6TmH2j19l Lxx0l8+J8oFsq8cRMdd6KKyEwYyziwmvMFKsFaA2jemxayhEVl2MBubh4P643PS5mwPL/wUniM0r PN/Z/d9q1lOlhToAsAUiCdmp64Ga/gfIy2szKQdDH8EjHsAQKfKxkcO0rx9RMIEsS7YhOCJAircM H+4NtkYM9M8BSPioIpSkAtujBeBxzZpoi23PaQmWtMLo+/pa4bNxTCHhWgIS1TeWxUxxWtO+ZVIJ FBceSFOSOLNFYV+jzfMPuCFfenhWC+/ucTMhjP3E8Xb9BAa2YNlQjixnT6i0PU8lRxN0b0ZvAwXN KZfMr4G+k8e/c0P321mXvAKtc2RA8AncwMyE+gKHobXuZYuJ3AyNqiARHuJH7XROXypIa56s0ULA DwoqLqJgcdLA4cbSxkKDkU/BCo5mlZ16iNabm+9n/bmEJpeag/uCCQ3KQ4ss151UyuWHnu56tFlw eHhFBovDhY20GmnyuQemNUwdvcH0cD8USxEyr3SAtGTsH8QmCegxU2KUTOl31f952Rg06C5RK4Kf Th4GSTcZZLtaWUWvKqToyWlIAknf/5+1yabcZecfQvFayTGlxHOIw8sdEugcs1OM0aexQR373Ifa eFQVz5F3ZfT2lU1eNq2POIwQaGoqNMN9Qh6FNE/J5j3t8itfeYrcx/dzkD0BTJiNpq77akUQQ/8N u0NYY3Wufd6bhVg79umfhP29/KGbFP/k0Q13gonYwr0ppYwmUlWbR2SV4+anPnBfZcLi37y/AABg 7NhSAMUkMbIrGLXh+w+V57RQmirrMoXkdv3MP1VTJ5XmdAIm4GZVU8BbJwAVFOHFCM4u7ZLg7QuA I8x4yPIeRPv7r3Jzn1BS6MtkY6zC6jSClp+wCLCFlmMjJd3gRNtScRrpy/7x7HQrefJ55L5loyfy xFGTHykgQdZVx+bPhMH8lNxvdsNCqtmguQVM6/OxEDD7mYUwZ0cvi/78l4gLDDjA8FKmRVdy4ChA JDrxKEGKhvDq1FP436pKLz2IyIJqT7NQhfQQbQQeeWCfaJy3S8PHkMdFACPWxxoZPU/N3Kgk77Ad I2332ilLHxXb5lBl2Ezgjaf/B9KLePf4gQR5W+jwwM7JOGuTmy0mJGMrzhV4LwaMs5dlsMQAqW/0 xJ4FMsUTQR/AgI7txoSBdnRUJgswPFAPgUZjW/FO4pQd6a8Ild1rvF+Qrf1un0x9A3nktwmiBpTS d+xRZM/xRvTG0F/qRBtSlOmFXhaz76l1+EElOqG9r7pA8j4xSLHDBuRiM/wpNvgj2kL/4q7Ucg3a wPaDR7Mp84ufSn9VGarSiipMYu4enLEHDQkVAPoDyPfI9YxYzAGKh7iC5ptKQEWygk/n6WVWRAjB 7gdnCqcu+XQBmpyESH/qn+bv6iDHBzTvze70G8hR8qQ6o3oLlf1hqShQ0VrbiEAEL8OBhwT+CXow on3YkqAjZuq+MBNxEY0AiSlAI4VSzdM5vaTUgSfekOeJJCimq7m+yTGwdAWiCjhel8XRxkCBPpTp t0BiYEbqL87uwDxwcZ+5oaMdXLXcWSZSajvoJZ3RVCh9Ua5JfXPhf0rq/cNLVo5yK7fobp+VeX+M 2n+cmUdLx6i9MyqNz8tGprEqUrHDUrqVpJk975iuVRp0Fgv99gp73uR9ZP+dQIU7sccjs5ytaATA 4R0KN4EhvT8IeMOj650KQqa0e91Gu/D8SC5Z4+ZEDCw0iaXAuqcjk/cgrHGcUHZ9yjIit8/MJXWS p5xuTj2/nQZPzQ05NJpBsfQYxjZHIW3QG8/SzaRwKh4VhUjjJo/lR51FxHnKYt9YEygWhoeiGQIm t0vvan9MOakym4/41ZP4mhGzPdfoH9eEwZ8FQ3qjK2cAmZpwLNa852A7UGhZ4eouWUKJZqP4+vSD nFy4cmCm+C4Fg/DmlGI4u3fjUugFrqa5lj3HtZaheOyCqyrq+ZFxSPIzEdbrkbFdtD718ZDBJXhG S5DNUuUQdMM75P5PcNjSbcU0smObUsGdlqEFcARMDcBMtR8o5K980pVrnjPMUN/y2wiJgEkVHhlI Z8W5jvq/lu+S2ToR+Q+bpX2Nz60GXm8bZOSbx2oeNrLk7Q/vBVngi0zd0fdmTHUU+VwhdRe0hReI XMPjJcKXbOM+UmP/lKhRkphfgCS00grogZQWnVfGuAa0bGstCTgJAPXxKDxRaCMx7wYr67E41tqo 2DE8vythaXvGmwtuxYRru6RkArQ2fTwliuY5UQYOjDND2RL0nuwCcmd4TtHRLqhNX3ew8MVGhZqQ bAQWqkElkdMdTxNPHf5YkJ2leM9IUSC+PkjJQd1Af7Z1oEOvJUd1uyiIiXWAcCueho/qGi3fr1FK xnARKGS0Isy0qg0yaIIcV0UkkVrrCc60H4BDDoJIaV5HlGNtII/XkLhISwoYgcb1sMU/f1tmziAD +Yfbh7Qx4g56L8oysxw3cVRSePv2RXvVaXOBQT4yFdfscNkxzmGw58T9Kz0m6G+eTuZ7qCo7u9+w v12GPc51acM5srXYgY2dbX3/jKrnaqOlHAspA3+GA5gzYoi5mrPrNfnzqQ1l2CKdJ5SgDBO1MbRk YqS1zUXZ5I9L/XmIvGK4m6SbdD00ntHGxtP2HRCYVpqBDYWlwH1OudYj5IY3PxFjbHs3EiWXntRe SEaXRNLuj7Yxu1ie593GBOrPvoiSXcoSJB9n8KfA7EZ4iJ8wrI7HaIrEJDQ4qIx44w/B04Ow13Fz FtR2lCH8DqL2r//reH6XqGGVHT6Q8Z5eaxanedUY1WyJW7FFyniiFEhTBEJftPHoChDcyTiWXeVt oeoALXeZD9iiGMWFqSR5o7i3JQ1dl+gW5CI79q1Uuy/BfiAda8pT0DNAdor/G6aqAuOENaKpPOv3 1h+atiTkJJPRb5bx5ciN06BS6V/UaeLrVtwvupHnjoYJm7AVMMgQ7pwgebIr3yXHB/vOVNvt+XPO 2BgBoCxCNtcl8M0RyAEnRikvdETacb0Anr8O42dO3nxlL74UwL4/T9Ew7f2JnWQJ/MECDyZ3II9t ny9gLCpIqur3uvG+2C2tYh4pllmaNJJQAPvt95TFbY6BDIPZjmhlLs4yho9Y+3uEOVh4GvkSAf5T DkhYqXIrPIP9Uz4F6/+WNP+pWoY4tgdLrBe6rMKMrf0s4W4cW1JQ6XcUJiUnjzqvN106f8Bfa1M8 lSpyWYC/ni1MAi3OcOgTQpk9RfPfynYRy8hTT7gyQbeX02+Z3V7Vy71tKorfu6HiYG777FthCoKD NjMdYNhM/YTXTpm2u2JVC7frICKSHUVnbyFPFAp5Y7AmswTifthVj2XwSYAlZ+WqsAAJaqNbabSV vplHoEJXWtGd/XhZPrnP0cEAVlz04ncpmI8K9BgMfoNFuHc82K/VJVz6Z5ZLFkRF9gV5gUeMo6MT hBZUhU6xtRijkePR/Oaq0ZpzUJJMDfJJzh3AW/Lo5HJRBS+puKO/+feeCEcvw14pMAKbqJ/em0uV Omoz9NSyLG3xJkJb+U9PxZXkegsEXUobw3Y0oKz7oNV29gzWqp53bqioeCP9W7/TvuutI84mme+R q5vCjNkoXfXWM0jJKLTstjn43m2h7lulMNFRwDdTQbiH7MMv9oGyiGrJ6NTs2/51v5k9jw1T1KI9 kBXe3RFZj5DupPB6xCX7xElNypo8SUgcYawygAPYewPtq/jxD+jG6NAXl9pRezt9lN5eBlDL9z0/ wiC2kDQWAuMuCaMTf8Od0LUuO6YGwGFVSf/UmuTilV4sTc6fbsOQSHIPXaW7uuvuWhkJv2KVc85F Rjm628vHsof6SCmAI6ZFscojH3sqb2zk3afw7rvXgV639HmqOs8FJzwFRdA4kJm/i7Jex2+WdCuI 6hMYcpCGtU9I7r9erruWaRSXtHRLinBWZJ0wZbPuOcf6evxaCMrFCl+PXhIu6AI7/VvSXNRKjfAr Fmqh29Iblp7NxJWzC8wmUfp/xzx2colz+gIRa+gI8+WaisF1hihufoVm0RgfikV9kwS/10uM4mOL 1+mD0tzltIqtm69Jk9BV0na1T4WVKfjwuriYsoFoA40uBpKpNBdhMzLvRJ2pCLUnhm7OiU7XjMuU KJUhKs40Dpyw0NVa1TjIx0zMO57WhNRVez2cZsDXk/eokTdp4Xxb8CyYbaaOHM8BaMYPfgle+kA2 bNtXbiOvJkqQQjxoEPEFIrUE69Zl2tisLxJz3aRb3snLSIoqHzpCOht780cqWfQyM+PVoqrlSBaZ CnpamHxDxHrNEdVweT9l9gfFob6JsTM0SPqmfSgUU4fOqO/MpH3o8QkURMI7kW4IIM5qkq7L+yll 9F+ZR61pkCUK5pYo6+dCXJ07sqdWYbS4+5T8+feSziK0g8V3YEWmXb0yC36aeiIl4z5GCvPpi/qZ mxqayPS59L+5Dv8a3smVNCeOiLpjQOguO/cqGlzPFM8/EdELjFktIrT7514vcgtd95jNxiC70rVZ Jnk38xf6BfELOHXoj9lAJrGDfcf8qHLjxlrrYL83/V/l5HpE6wV1fG9bjivKqszlYtgcZ8DX3wbP cyzp8/wV6pQ5l8KmK1cpPEFT1595nZU/f/YuGrggrZdYCFg1jScx+vSMfBU6W8KjfQticf2zUw4e KXwridJbUVkGyP80Pm2Qd9XGbzI7d9HxXAnyDtgOJB95NerxmL/PtFhz4Zh8TocSFTHfLaIBYRP9 OC9khTyt34Nrp09d3EG57YtCgv6iEP+PLmc4f7zyXhmFM+lwRTvkYU5jbRXjAYiFptOfVhcHb0Zp 38zM6OckEXzs/w5lywGfvuR/EYTatdhVxTYAaqXo8GR5wxYsUkArDyua1sy4Rn9ZZa/dsV3lBxwt c00zgLFQSbsEtMT4jhxzJewUPkgn7GuWrnWz0jtbB38GBqtZzfoeBGxzQtmWEHzmfpjklO+9I4Vu Z+0KILBvF8SF4uDlZnZ+p/iACYz1LnRruFooQuKt0Ctid/32Hx2rEnWa3czQ2nDKYuAPj6eigpL1 1ycqlMKhIPz75HmACJkjv3KuKLo8NHMLEpwFbyGajAWXr6mx+FWL8z0BuBqezKgRy3JXalDi15BH 7BxNQTK3d4ekRD/PGK3srnlGntv5D8+UI31WJhkUPlZ02a6w8vHOoMl9ZGdszg00uOA8x+xRjdT0 4IIVVSACrhBRb1YYgRfMLapfTW3a9NQW6NDysfy7cqNYj3PT7KqZOF4RYjd0sb9+xfXe1CYUFKGx yXHnS+mhc/lW0uRtDrKwUIdz6VIdY+Mcl4tIxIFVZIBjskwqYUSGJGcx9e7aftIR/mf4bRTAwmCG uac8Kw8jICAPLaCOnyEFcNL0cvKBO4TA7I0SfPNkrOBlFoewexTuHUbK0z3p9FGiC9opucM6Y5Yz 9hKQokuFldGHzjtm1z8celhrDSW02NrxmdRhlvra/gCU8TS3R6GKr/EU13R89EynjPWWkYI46BWo +90G8S4KCl9GTSGsLYFY3sCWyjFGfbw/d1VyyaWKQOaTpVJJOJ1y0nCeTuzJ+wwzMqmKcM8gjaQu uHuPdK9sQfLXH5SKDuCCdwZ3yOyukdIKV/xdXdiPm3x4eFuIxLKO9xiHqnn3EyBCw4n4pVi2RVnj pdhTxLOar2wntlhjATXK3Pjk9dk0e6ngHKcbn84gr9qA6HdAllPBvFc1EVS1nQyyKPIqBi34Zvac Z9NWLkA/0IGyE1YYq+PZI+Hpll6d2ydSWbfN4bHfDfjSvgFDtsbYT2nNH156fCk8RWBr1mRHk5dg 3XGUHtQ2gyuhLqSrIC7bFiUIzigUkpw1VFhKhz0TY7X8ochs4tXbfxUU5SQCrYa3bBZ0UMJuufCY zY2rk6y7q96Ee74eT8MZB8HcdxM3Gz+HQL9U2MvcM2K/s4qe74V+9JCLa7/hLKWQfaJTv579P2o5 JnqqkcpZ88iyCxtwKRKp24r2O7L6Or0+dmiEJYXlO0WH2er2Lm9lZM50cdPH/8opsi7fAbb3zxdf oOI21oKwi91aAkAcSZZCe0s6lm79w3nH9wUHsCnejk14MjtR4idksWqXtcCr/jN/VvoJgxQWY5Wt n9b1+OQgeoC4dVFkjJnYH4fYxsooiczsE1LN+8MilySlduOx3vk2RK9qhvMET1j2VknHzrWoYfkh UnZqJInm5/8eV+70Lj/u8v4Rf603+1mjaBMbdL3xKpO+0xhmPvdmaxxhIKcrgGbiIX1gtd746v6X btGxbi5LP4/ajJVuZYoQwcRGLdmTDfSDIQymrge/4yPJ05YHsIv9oDcV4yxCH3mrv5Yufalapyrk ABKyRlczNbSZWIaSpmNhK9a8RZBO/Ffn5HifXy1ifrTcXaSMHidarCDrkMrVjUewcPCY8mmDAY84 8eCHBU9/NbeGXwUyIZGsCPCvBW0ZVUA9wqq41ZH2VKMsvJYwt9pBfSayKFybPL5CZDD1WkWzmrYK DKQlyjjcBFpZz4vivNpij89Xg//r5RwqnWEytOTBW2PxHe/WZ0fUhQbXicduLRuhn7IDh3irEfG3 ELQMCMf83oG7+bHjjKeAhv6qOcO0xuJHTYpQCtuL9+vx5LJKvtdyQzSzYRLbSIUd68i+ucFWdfco 6MUPeg892WoPZcTVjyHiNrhirJyBMkF8Dg5+hGORtqkPIkenjoLm0zgbsqSohYTcLAKbLdKFzDTD 2UM6SwRmxNxQgVlQgdEKrE42z/injCplrUMLh9BcPab7wfC5Ty+Ol/e5wJR6F/3qwAsXYaSl2XKn OAB9aEsRsCoti9hw4keXXT9fdUZ5/SyZow1rXPARIofOMzmISJHMx0zdQ4Dh16ZIpVCWdj3z4T32 tuKsKDiULQ7rcS4JqcTOfYVK/N5ls8MDKGLsbPz+Hr8jIIBuDBGG+HOoHlqqGalWlDoYi/dI/wQJ xrmQUdoPg14nQx8n2ve6Zij/axM/WGatt3vn4URn3LX4Q/iirKpJ6XYt58UEhfypvO/kJ3sJ3982 kjo1hjR3nLBT/IEsw/ln97FtjZYi/RCyMt8p/AfURTDyd1LdSVnpJsTblj1teUXLd/mKRouKdtTs mxaMT7ZrWk9Mup/L148sHhMAMe5k/87ivFORW/lXlUkNN7DpdyanlaXbIEPxkOeOsVBu9E53AEi7 rBVC60G/H2eEHFkil4LCxPDzQTjjoVWHagUQhIE3pmk4gJ/GWk9Gp3DE0jvySQUkzKkrJ0S/djfj 1W1hS5BZ2Oi1u0UN6RlJXtH3lDo4htavC+so2Ma93EOPFpm+fRSAT+W26e5HpJFa0n0ZlqRvAO8r 9E05919BDEbsisU2O/DvwrQ2ie1U5/WQARF8Geak2Oy21mGZEb+8AjkwLI2FWqcVXTAk9e/PH6j+ /p7j5y9Z68BtKPhA5H3LmzG22OCvcmAp1zC2x6ayM59HW/7UspZ4YlTi5jTAsHOuOzawKerqE54A kd4c69m7KkkEabkSbxKXoE2D0pL9rY7GI1hZLpRwx0W47nmDbMwyNcn9nPYjLPnUszaK/TXN8XdX sMvuCOq72q1ijrIzWcwtiMUF8DL7JrvLg3Of6KAPJAWEIdN9W4hd/bKGW5UqCSX99sGFk/zPJOjm iEjKHV7yyH9QA7haEpNfxQr7N7qpEi9htBP+2k4UmyksSao2uiO6aY8D8KtlAiRy5dQS4PnQEDU1 w7Ml+/QjCgVNoAy5wKdmfWZ15QmWP5GmN+E2AeU4LX+wbtmj9HzT6f5HHdHIrv3/wDaZ5TaJ5XCL Ao+E9e1n7swpMe3fLoEY2lTgE9J9GSLEjTiNHMLfOqQopd1zdCy2oIxT1s8WV6mrC2XMa+q0NeVJ 92FlegsAkK3QGdD4iyGMGCN7YFrf2BgDTGEZ7AuFU8bDE9eG6v/Gl4ko+UMnp6wO1bF+8r+AKlGt QIPfZ5wQE5qp+UMDs4M9OwqV+zYP/ihc+H8zVFW37Q0FTd67Wsp4rjCalVEVdA5n7fYhDAx9IwHj vDjQ+TGzeu0mh4kznsB8f6iHPx3GjZaY1eLQVkYP4UbIEHCHKFkdGrJlpBQ+kjWk5MWnhiZuUBgJ 2r45SRlEiUl0pT5B/a15gLbsAsM9yYUhl+LpzOyChSgnp/lyCjTF9pjVc2PWbvxDyVEMkrsLKKBn 9dj66cSB0ZjZsVMVCvebHVaGzS7i7SdT8soIKsu/Rvg7WHTNTAGh2b+R4zoMocBVYbRtNFCJYabu xXcu/XcL65DDt1icUgbyCFX8STlrX/TIIy6dD94QW8JWoftNz1Tvqh/c+YzhU/UUJIBGbSMTj3xJ fCP+/XJ3EorpoBU0pLp1x0M7tUNvCkeF3UBpNxKls7aTcKv022iKCsvCr7cvkhbmPHhZrgpL2685 ocEus5eKOZW8U9HU2Yy8mnicPXsxLBnac/cdp6cL6C7nWd4TZ7q2kOaZtblI8jtauUY7WPvaKE9q 0qAe7qhkI9BXrV1UyNLUn3HroOYZQzvBp7ALE0kLnox+b7z9Aa/3wfwCaKmPYZchY759cniTNl1j HnhIAuT/v3sK1huDrDuTYDmqjVQXwTkjmLFaMM0ECkrBM/r/Nqruv60groAeorL6ubzJiF/j6twH Gd469AHnjcC7b6+zNew0qqvUPXCyJlR1UBmdCEbYtB6+JAisW+OF4PsiMZgGJi3nazkmDIpUHSHg IepkwPuUjDIps108TO1VH01sKr/rv+M7PJbUzaRB6fn6/X9vre1APSqMc1TcRtAWkW4pZlXcrMyq E/Ri1jHo1lHN9AC18BPP+ydKdqSdcECjuLP705ETu2yiQpSFVXbCrxYKHCV+/39OlbnNaxj1afj4 fHgGRe4QFzDRjKMWE148V4iJQDO7Li5AHNNeY4QOp+NTPMu+7QrW0d+I5gVPmRFoyLEtoSJInXgw 9gVQHdVnKkjTt5eiJjIdkYLTqV0tQ2GjUswHBT3+wmVfyhZSKewnorbZ8O64hjCrgJGLPGVzSSME J36kS9Di2VSZ3GUyzZmTf79YUZvGA72ls1kK9JZ65q2aFdYAEf3TACpjpykW+32TUeJSeJlFBQOR DisiOFyStCLpVRKlAGPGINn1lIWmQdGNbnMxIMkk1h58oVOEFwm8cPKGlPGeiEFIrbXAScajZDB/ LYAtb9yL4RC6m0YBKoZQPCYoAkvHtbRLLjXPR1pE78O/AKLfuVu0UVyyfQRdsjuItvM33QTiu2+L QG2K77o9kwns1nSN20chyzAhDdb78As5glXDXuR3WnrQhRgEm7Oqx6dja+KGN/SHjv8Yl+DriL5S xyRmkAfnBYnvRjF0Nq1tQHAKFyKqNH8lyMI+ZAupQXZfKyscIFdki7y2fPHAyO5XEC5D4glILTWb r9oXhPfGoz1w0ap3csC/Hgz312JHsM3CuUEa9iBoylcdsKg4EGTET2Cz069r/MuqX1foeTDkQ/IZ nOkCTYfvDAzUUiDkFDpL4i83qWAurxpES0Mbszwh90Rebv1Xu+XbsW/Rzdj1Wbs68hV/G4aS7C7h 9nF4yk4vEogbARsQU1+avt42RmogdBlEK8nOyzaSukIBx0YSC7vnVsU/lzpMgnAiMPTrQ4AxwiyR z7+P9kHtkZ3jcUka7b33rxobcQnEy2vDJcl1kDBK62wfnnu5zkCvUBpWYMq/eHq0V03joYhmOe+a oAk2JIUXXtxt/Utz7Gu5wwARtUAiQ1TfGMoF7sYiI5EWvVUR2BNFTPwaBXceJ7oP3Hduw5q2Wtsx l3EGoxl2BRCQUT2R9OeOdiMsPYDHMnak2IEoszZOKvu57E8H+1bGwLIMnH8pkvWQQiZzhPBgXJvJ OoRmeiz4G3wTVwMMvVnKDTFbK9b0No2LvMKbHtLpCPiL8TiOlLYQDJZ1eGXRbqTvQwFE7NQjZGdO T3uJwWFU10EUggZaoi5um8in7FbOdnA0qduEzfO4CI4EZSZBU9Nt/VqoDjJX1mo9rLk4EkZ2uuQz uRYntza+A6TpLK1zOr9RGLpQN1q3Kmz8Q+ibdt4GpA5NGK5D1yS9Yttn1Abph2u+2t9urC55rv+v jPTskuyqeiLa3fidNkYGrRJWXMGh9ln4u+UgIvQNFWyjBsjQ/92QaIAFT8YSnYJhCIGbdz8QkEk3 RP5zFZZGWhPMR4TpspiOculQOAXuXEnQkW4eHJfPS5Ba3IuRIbUFfc6ma58YPyTNUpnaKjMMr/yl E4eQNi3hMqxV0olHWXPMLxOu2u/M+7FGLsjrnnscQQJRBBwcSnlFyhcnDbj8V5HfrRF90ykVGb+B 4+7noNnOJ8LQPBsxRpGEvC6/MjqWe4Gx+SlBp8E5mrGaEI+2dgo9unk6CNGFpimC4nGAP0XJCdEj Ics4vSotB0E0XfQpsThiUqjFvFEe3NN49/2i+9gEinEQlRmdQp7SdYqit/cvPAxCNlFrjM5a2gId FMM1zKMSkobZ7UE9swfi2RFxTVeh8/5nmNgwK6KoGyB88brKiyhiy1KwnRcwu8tfBUIA2R6bpe37 uR4S314JKikDf3HkmlR+HK12iWXAsJrIoXFJSsi4LEXnqLp8n24nmNh7ZNJDWlj2Iq7YKUdAgF48 YodqrrkUrshlW8zWw6+705pVW+VXK9qtsEOKkTAdqem7yjxz6r0Aw3N9xn/lWk5m7+EZ/psP1NIK GU9tZ4iVgjIpQ5Xj6IsbZF5T//zslMx2rbO2rsIBJwhq1HABq7u4AZqVDG4MIy6yzMT0DCeIFDfW 3rSlK3WK2g0KZW5kc3RyZWFtDWVuZG9iag0zOTQgMCBvYmo8PC9MZW5ndGggMzU1NDcvRmlsdGVy L0ZsYXRlRGVjb2RlL0xlbmd0aDEgNTk2Mjg+PnN0cmVhbQ0KrnC2DvxzquJ0zDNK6PJAJLWCjwa1 L8t1hfroyKtVwGppaHR4Ys1ZOEWwoBYQc4GRrRALZTe6UvQsRRkXKzCA87Rng88+tYbr8IfpVq2O cq6cDUQC+IcaF/VyXJLoEOlJFw7j6B3mESDvQ17SUI/YP3YSYWhNWaPjBO6jnTpji7/QUQPRQBew WZn36Cp+0l3UnE0j8zmHnVr3ONJZg1i4Rn1BSP7exx0JrUsP+GEpON+3A6DpGt0sswoqbeSHEkSa oU1fLNOS3WQjFNZnMZ/YAb3/xjorEDNwAY7zSv4rl/rOX8oNopeQrcuZYdz1OSzkfLUG2ymm+v+O SQU+kY7lKtuZo++Og8fvfj57trnZfPE/4ULIC1xwTavLmLuuSsE4AInJfBphM+D7mLCFLkxN5W13 vDE8BpfZ4yLRpP+ajTy7/XNwCT8monY3EZH2IfHmWmPbgm1O1fGQ/owf5rlXZub97c9ze3jcg7Yg nKS+zLMAnsPVIeA8WBSu2VnGY8Hnito5vGvUYIJq9WCJo/LrBMAJKs3ovNJg4qp72RyvkUTt3jxA lSBdyYs2an8AO5AHMU1OeC4Hu/M4hDvlhIrYzP0Bh2Jh5RmcKqhFCbTx3NvDG8H1rUYHYgmokZci eDSh9TRDSbe3QnC+krBwjm2KU9jb2H8B4V3HpdPC3h3iqzrbg4v/twRPN9B4zUquSbA0Pg83+lE4 cPUdjWyR4XER/oUYSGZqCTN550Mpog5eQR8ONXqrb/NxrWe4oFl6JMewaGYEHdGSW4vaxlZEz7H/ Sx1VVfqBGANYmhjlfbzyi41R1RxpKV/M8mMth9TOc9lrWcb5IZ8sqSfmqwJgasH/1j+GaaOCc52j nLkFAZyD8cRS7vLHHPMtzo3Ye7l3i/QUsnjo/GUdY85m7eXrRSNGVt8KCatHg22M/o/os8/r4aIx vjF7f4FRT1qJ2OLEDYxcqdMQmaJC24hHUgClhuxapwdDuDjmZdVvOHqWVRfvXlqIJA/++a+nUIM2 U1b1PMLagpR8snqIiT52kmSyKrBubzLHJaP64afEwqTCwkeYs+ObAe1ARB9JNG68E0JlOpJBVADQ eC0Jz6XutT7WugsTfkR2JhSQGV11fmMS2RFR/uki5HC7JquL1/xVO59k9QewqldV7BjTskgBnfaz f4xg3P/IO7tNRvJVFqoSkDk1sKFfzPQ7/VInqErWcwfLRwKoKRqrHJDzidv7TgEqvK5ujDl1SsW8 xZztw2Gmvz2jLFLIZDCgcaEzMEBQ5qDv2gHrgQYUU2z6Az9fQZ+NPlFRTMhWIrYkA8qVX3QO0Iyr H3Si08Tza6VwihkOvPgn7eIvIMAoqzdk7Rv0NDnIbpkuBFd+MTJiaY/FvehIC0Pk+wKRgvuuSlie /fPbB3/GE9Y1YyjlHHeiPiKtH/ruwLB1ku61cNLEfX5rSfpt/wHMK6y9cTyHU8UzGGSKcX60w/gk cFmzgtpiPS6qFovw8rVB7b+3SvRVBKQe5ors9a6bqZ8e2Vxa3gugYxmrCzlIeTzWbTeZ8vcMniVv f0uirS8chCNqM3W1AmGdiENNDu0VtuOTxvG+JSKO8jCl/0hHVE1bfAEz2mrnEpGNF5qSVdPzq6o/ nL1tlhUjZ5M0j6n9JwbDd/TFPGTJKlSj8hj9orSntFsLu9ULETX1yGTiMHMIa4CACAbZ7cw9392F 68t2Rkvg5KIXchb4hOpXsOwBYuBih5czDrf85AVkPCXxynHS/cz2C518TnodJ/iK7lNW6cq6AUyI OPGuX0/tXwVK8MU1dWK4LuHVxqq19YY0dqf5CClbylWu8Tn4tS6i9wrfl9pnl2ocpklgvhD27fye PYwAX0Rq4V8zWVRzJWmHo9H/UK/LNDYnpDlWWHiapDtHRlVk5m1idu5+MV1x067dJ80Mu5YTldHF 37C2tgQzyK5FJykgtal9U7Uv+zWiL9aqcPN4Q0XY90AlxXKyH6ssyNcEpU3tUoxJnQiJREL+JAI8 WttTAUGaKmYZLUMxTHakhj1MN2AqWvzYmIOqN+Nwwm8kLH/9dH4Jcq6snq5UAejh8rovbrwlZN+Z V/PVKxuL5pHTturtT6TJUxQzf14yeZbRlE+EHDXHHt2Go66xxMJyKWItgDDbo9mB6/1TQ8YXhh9j CVdwK4UHi44G/uQLchsAorA3oc67qb9nxQcuVsMjEf3RNYpWwF0FJZZt2A6dvo4XDgnnZOdJtpWP fFzIFkdDJrIK8xp0Hp/uPPoBC1vFLiOr0zgetMABATRBs0JsUjBmoY3VlnfC0jvEE2+03j6e+9xS XldMZL+OH2Aj67EcNDzPuIZU/m5MEuWnMYGfYCpw+KTaO0wLTgjzQ1ANzpViGMgzeu7bPA8iQxBo dCyAWUi3TxdZxfrxxUjuzSyCQQxn/2+G3cIdwArceRkRzHcXrb3mJU8M5ZNyFU+NWJcHLKFBqJ6U W9xnysqm3d/mYB73xQ3mZrkyedy+qni2T78OlUziydFxv1vjkjzQECUw9YGIJ/IYXDmt5X1Hyt7G 5Hu7ypGA0OmweimHRH6zSG/A8c/2oTIlLFXP+7rbpzWggE1IizP11+wPxEzkZIv14n+6ChpNos/X nbeocbDvYJMAaOE+TCq4f0dnZ75Ao/bwY0Mj5ZN4tWHZIf9FU3RIra0DQHEri8ZgO9w+tEKUvfyM RugrKyUu6yTRUttWL2px+Kor04JohLiCZ78FYxPkVTnEEpULdzzy7DW1gjV1Ki81WcW3O7V9BBch 2cpIHoPv6pe/v9m9ILJZiQ0ffPx09q9R0uJ24ORzWThaGMTCpswKfFyXiDfjouObfGEI7rwHkAMC YMu5tSbYGB4UNkvZAjKh7uE8dZX7fxi7PJLkNymXHa+EsQZrpHz2nyRzmOsSoNvzYA/voA2zFk9A YcJWtxLsgC6GL9Q+UUObrba2qBdwKl5+UV6ELpLUtz+P4uQxQGKw6av8LSEhnTxfS+YhU9hLjbnD iFlhTIo6WWn9aUqP8qs3ISd5vt/ETp8arsZZcgwyf9qQveAy/0E9Fz4f6jD4XEpq8/MINJ62co0f 6Cc188leM3o6TLWoiseFAY7Sc/AjLxpZUaOkJggZHvS6ub2HwquqNoFSpLdW1m7xfDRsf/2jPf4A IHstps3o6QkT+iBpPA15KL3tp+d4rMMnVlB6zMHKS5zn93VIhs8aA0QfKYaaa/wXChFXCPj4m6zL cB//4lMPq+WfwNJRuKvKcvDlYAjtYRL1wX7TxQG85O3FqO+dILFA71blGR7OSskMDtYYUzgQ8epu aletMEVPJKCfn5lTtwGXA99jCZAVoK4wONk/Uus+bxkKO/vHLKgL9w61d7epVSxgnJKdxBVTJoei TwO5NvRrXKR1HxYt5Y6kuiQgqHm4L2OuDn6xyPUXQ7dwaYz8ghxyDSdweVIPeyzaANfVF0P755Ux c6RyBcxeNC8SIBv7a0TE5SS4tc+YVs/pNT8pnQGsaSINzaqCp7uIDz2LbcV3akIRCsimOvEmAf7a g10seoSyTvtw2Fo0SJ/Mg4aDeyJbLGcBpta/Z8+Zh0OKZP8A085/fNQ/fseZbgBiehGVlaCX5AJj gTSlo7NFwvgAX+qUBgtAFBJgCC+gJz3fJq3SIyVlBp/Mtj6XReWDikmCClzaBUThfKMGrbhok+Hp rUvA/ni5Yylhb+bal3TJM74o7xfLOyztcULHotaQRx5z5LU2WFwP9QzmvNcWzytulba9Woam181u JeHGz2XBkGTTgys+p5c7oY8eRGtGMaimL6piLECqpSFprqXXzJMx2P51QMRrxk7mmw0+hM3z8YGm 3siRqGJKxY823IeAkr5l0fRVSvueHKeaQ9ALOCCoZEPRysFZsupZT6czCLBqLCMjVqUXrLaf10zg U2rKrr0vNzE2JEoM+NQN7ho6Q024m7if8T6EGxGvM/PdvxVWEvaVw4F89zgOFSPU4Aprz5JvSPv9 QiJjaZiJAgZWsiHYK1angUnyoi3/CeAWLbJr2gu5YkOrc3Gq2aigpUJqSJECnubn3mHtwrZh9jjv n9/3NcIu2uYbB4eefC4HKWAO8qdulz7xxYhuh/mWvITFCBYvk4sI+fSXcU/Lg8V85lAI0v7zOTJm OhU2u3lru46QoNNqTLpXJt4SPRoOwo17UxqXElaKTdd1rCRlHftTrMUUEeAieo4UgIRVxOYb7FVX bgceZ1ckrBfH6v/rqNmf9y3vaEJ/0gHiGdJEt5pcGv96RNOUWGzzBZQ2sDRg39/G+TWSZvIHNqAo xKlEYs9n4T5taN00ocIm/UJEYCyCBIpJCSsPRFwFd25D6QTeEQDyn/24m1jmlofWliiJyZBYI/qt X6zdJQPrR/A5AhPEBYMpIWbRdcbknMmws5BIkdTArmOPpV1EzQwagC1FI+lc7Xmu4KGvpCJWimVq jGQ2Z11ixkNXz6l/T+hsEJeBPaTHOWdJOC99Il/v/xFsZAlaVPXOIKi7WvEuL47dM3UJGlenFk5M 7ss5TrwcflkOBkTRJJoa9L6uyDWP9R6bzuaYvnu5ruWrW1hgv9JSlMpMdbbWKV3bQ6sWMRMYX4OT 55F7LNHLKmmLoMr2PRIJOKuSDDQqcNq+H5ACZPDjRBG/1tkt65nr8DiROAR//pL/c2bsGhBxU7Tk a8Xs4jmgQjz7rzrOS5vMrvhpiZirzTCI4bM5ylt2PMz5eA2bLHPoXIjfhjfh7EK6LzOczxqINeXM Jpjv+F2Z0m3geZlUXFOxHu+b4bIrUb3/cyXxUdDJL5RUg21welqu6RARxTQIjwFaTg+arF1H/uVL O2LoraNu8Il+3tHxo2Ksb7lj8FpdMVY8y7BzS/sWGTuKE+seJGJsBGhtOTGmNn4MUHIhUxgiSvD0 9N73FLXrPly9fYKeQuDpcOcKXPWOf2w8vdu+YF1rhmD2an+eRyxtSE4Gr9cNtSaEpkpXZL2WiT7o 0cgfbubFUEQwyLrHpZWKjpu14k4Cza3l6PTJO/odTO4oiKSV59d3BFFDAu456yP4WzXYHUeMuCAJ +dRUTGpLgUoP0JVj1HZ19LZFVBlTe9lqdPgTSTwTdAdnQjk9fGMCvlFWBUB6gt7z8rb3uxauytNn ZGclIMaIJ4319NPuuhOWwEL0hrckHwCEfsk61Sg06/Bn4ZozLvxXivzzBxrTGInQKIJho+gkcNXQ J199DiK9dQq+R8UCI0GOkqKP4PR1SBWX1ntuInFJ0tIEHA2okGd6NPomrFwq529Edmz2GEobEgrn SRqERcWdcYUXUeKUUZWHSjm5azTVfEfin5RqkL4LqezQKi//eXp+IUcTV+JYLqDYF6r7nI5ZxgLs FKGcRQscZgsGOPzNuVlijhkdw9Tui3DZxxQB4WsXnYflpTHZwK4mvnrLFrlLhA6FVZ1qRNn09nq0 wa4zhRjIdudI29IL/m/1et3JY6p9UGfcKdDR/DDhEp57/NnBz2qouB/lFJZsXbr6P1tBSif1+SEn w/7HkRdxN8AdFZhFmfp+IYjymnNzz1i3iNOt5k5cC0JM3HPPDRsmGRKD0L5NJYAEe9wWSQ+/NwgE VHg1z2sVhqjQqRwhDpn/dkxhi0intug7ajJiyNALnUSPFa8klTXqQHvRmBnDtQhCtZ414faDL4xC QQ//6WCYS9y+m+qjOfHw804yBbT9/Wcnis80NH2G9wM6NRA6hod1icMQDbYpxoqMg5LAbTx4VtQf M7XCCnPBot9QMmqIGgIcrT9Ccqy3KlLK/rewgLTfTm9Yq0sYpx/Bg7gD+0X1k4SD5Xj1Q1r2wLVu gf64lwBCLkf2baviJIsV/1iKjqdNB/dh2bswQ9zkWhXyTt6OuZlQUZpVJ0jSoVtx7a1pqMaFORXQ ZVVET+i+zosEvZZ5m97t/CpR/OABzVggGlmzM1eDUuOmoZP6jGHzYB21yD0rknwsLfVMejOXyhG1 w5sThg2Vf8P4cJULioDhtlpMWK9Y+gLkP+dQyPEGfQ2FwxMwjY61U3M7/dfvF2EBGiQYEhPx04Sy xGJ7iUG5/FGm4Fk2dsiEMvvEj+Wd57dd95PrAlKHQJ6XMLiE7l2iQLlabcH81lQN37BvCZOXisOW tdlHZ6KReF4TSFsopi/39Vcpi8s2FVpcyQ409vEY9C0X8O9QovrGdUDSGsJWc2ukF5OluD/feb3I /rVzdsfDgpJfzvALlz6qU6TRBtIGCysma3a4LRTBY6j/fQS65uEWIN0bNZRo3Etv4B2edzV131Jq 0sjmUx2XfXF4VOMMj6ayX+EuFaDRdLvkXPbD9o30/2GOYx8P3PxDgFt45ElEzYGeb0wftJ83lwGU 1TkzdogNfcXxllxv+F+5E7HK8DRv+w4rSVk6XDXSKBDjGBfwPfnSV0DZaXKERGxHyb1ZzhhE2xN9 4bWTsVqwaIzKq4CscrHAmswvPSDIvhdJpbohqo80NXUxb+9MiXeSakEpGQyTN5BIe+WcuCkC3ivN DQ2F3sjoSZAXA4L68BqhKXVpNHJzfMfFac1kO+QRU4iemH+IktEO8FXTgDRGO8K3v7GK2XKFLOGy fAF77vgCS6lRuMwfgJ2j+J6wKN0xRIzhpoV18BKkVpRwHpsQhado+DGpes/2zWY+OEoFrAMpxoOz HyseaYijB/TiEZudg6hpxkSGuIFl6tpq61cUCZQzNu//XDjdeE/sd3rf15oQbag3IPpig/QUUqfz zLVe8MZMIL1hhVjZ8NbkG4Kj55v8VGZTgYMfhmq5QC5kmcoT5MdUNXFKi5J9ycTVsFt1AyF1jL6y orDgAx9nRXlwPGMWFLJQqH2BycAYqUEoeYGhT/z8yIGBlZXoRPyubemtyZQvzvHFcjPKdkCT8Oki 8h9HZfl368xPyJTGf0oJiuOr2sscNLVOVTh0hWRaS613wX0LKRoDV3j+ymAL7A37vZ7W05FcAFVH jl7Nn4/tj6/MQrh1TXV6Z5qnehMo4NQvyer83JbiRnXkTcouyFm5FWksJqQvm+BvikO9mvJYGnFt P8BkGhQ8ZLyu2i/VFEgI7cBmCKvDmU6aYMuOjnApyoqjL7/Q7tckEsDezUaCUapiFaS9wwgdHq64 tMZTnSxPSAyMwe1SOoq2YedLgq9klVp2OGgx2i3S04W00IcVL1SWS2UJNjtGRvVX/crswhnWbhxv lG+CsGgGTGDVrst2GO3hC2cu2nca6sctNcVn5/TLzF5SizMr0cpL5PrA5L4lFPEuF3pZ14JTJR1w /gZ7uHMpozGHzeUSqM1o1zWzoMDh0su2LieeAvATzeAg6duSiWCM5Mel5jHwapdctDpRiuFRJq/O WF9J5cfksBA4i7iyu1n1aJzd84TWlf1JKUerGnydppex8yRAq0HTReRxpt1v+wRD7Yha+MfR6/GP nM4REaawLlp9sF3VV86bsaFGg8UGx/wwak+pnn3nHbK4XYJxdfMBd2vPaXaDkF1QburxcnDHT/8l 0PqzmG8DM2eodRZE7sbidSBiJeU7tODw/b3y0nOa5QgXrzH6gTB4cmJ2kGJAAQcj2m/uueA/YG5W iifSpxobJmpUpOdQ2qEyL1kTlnMSg3LLqg/stOKH5zhWAltO0OD9fMlVzHlFVzz+XIsxRhyMO2DS 2svfPNaHmst0vbnavTavssh/YObYmCTR6RtB9i0IydrMlE6M+n5zyTvtL+uw6jMQS46mdrm5KsLH 3sTysL6NGsT3twRmRtxNhGEWYG0dYRq7XD9f8h7Uvg2LezsZxMuWZOzLsxbls0tZNz+vHKjDKZL1 Hf7BF1NvHQYTdXx+mbsd4bFuErn19TVH40N6d1I3e6QliF1clPiiG1p1+CadchxTXfBMhJSW/h2V EHWBR7uhUECnw9wtIO9SVko1pKKF5RaOHwjTTmBjrvEePvlg3dEoNPRXoIXylvRyWMtpv+Zc2xcw k7ksdLFqXy5g/AZZwu5qNAJmSYnZz6JVOEyPe/0xC0y0hEu3IuseX2a90EzryVEHD8LK+Bq0xS1g XGD6KLVYkJZeeHtbv/63ID9k2sI+EX/v5os27fyfzzGWep4o0cMDercLjSmdokbfnklDIaAdPy0o Q65ezsD8MEP2fUXdzGoxzj1Q5+DNvNqdtycq+Fcth4U3a6MirXMVLbdgPwP3gkj/tf9z4DNrQya9 N/XDLQ1DKqWGKCIBz0WRIyNh7LOyTQUm4nBK9sn38lqzH2uzfk3AAu8K2IYM7mCojevaFfTIMvD5 MX348zj5MtVUNhIf+2u/x/6dS3lwTfvVO4HK3s6X3ZlL3lrymAcoZZdPCLggDd7doTShRO7HpNgR uNnA7Ad5TorCsl13Ce8Z5RiooDphR2ry51Kaf8ROcdTyEykMXbAOxRTur1itgXMtzHxlCHDzh2vu YtZeAmmDkzAKDBn5xAxhdaIuOzbQFr3CplbX95tkd5+uAMuAMITT2XJB6r8KWV1bsFPlaCMesUSV W55TnktgN84xYh5RFetLuvRiomz6hT3huWRGWIf/RcIn6Wwfd/FeNnV9GDQlLZq9VAts/S2sD6uu gRXkiVx4ij3pPHxmEvLW0wUCzWxCwvXoPq+wQoMGLxqYqlhsgwr8oPdN9OlXLglm/SYOPUVWM3ut j4PI5+e5k1z1x5wCM0QDYv7SEU45sWVJEwag1VVMLRwpsvxoUO/HVrYe1Vw0oNpX3bUG+TMsMLny KTmBVIwNFekKmWCDntp+Sab/0JERqOezqgOX12EvTihdMSXi9tQrhYIeKv3MJ1v2+y2UdpY8t23y KYzMXQQpfuUWEnbpaVfI6ijItMdrvsgazSzroA1MrPLAZV/giyqhRa8lMHV09KsQlxxQ2lv1b7cR VYSv6nouavEHDDFzV+geTI5FXNdiVee1rI9EyeIGiG5yu5mXz6MfiSJKeVFi7nxeYIf7GBvCUsa9 FwjmQXR9rhlGYEsn6QU+3h9XpGEPZkn38v0AO/h4u6PZqXpDyNGaeWP02JFvWKU0rTyqKwNEAvv8 lGv9RKfxoCc3f0vnhbK5LHnBb5+2B3BupH1Xbv2/SA9G5tUNq+/bTZZi/9PRVypX3Sk4fGs4gP2f K+2MEPi+6moMN+/VIRpmC39YRpFYe+nwynQHcwb1d2zyTW7k6SF1tkl6NbIHFy9x1XwhyGgKosrt eSEAPFVk7zdzqGjVXlVjcNFKswJX98TIGiY9ry4aedWz0eIwWhXNKFJCD12FvTFgGnHbDcNASTDh +hR8kK11Q5fB6NtLjkhX0Ge5LKZ8dkq8SAj2y245aRxV/c+zHzP3N1BKCASGKO17HqFMoBhf8JTV r5YgQvVR+NT9+veAJ9Zp+SIMRflgDQZ9L2sLzGQ8x5R1PoeVUpwADioR2CWENaQ0Tugb0EO/aTS1 4ZzgJF578e15WnMT5PRwBVaWJH2g2EmW9iQMhfSYMBaTNGm93FxV4cssu8E6daQWY1d5pmDlSx+p /SRBIqJ5phjQhH4kU9+ZXPAzMW3HTUiCKcKRrfPpWnpYsFNHzi8VNzCU0ISZlAcTCwClqBnmMABw 0NNXVmIKh8yhwE9E4tz/d3swOVLWHmttawEkxGNIM79ljRDGomn3sAJm0aoFEj4GpZlYtAEljqI8 LlukVmoRx+8hi5lKXBhsDhZ0N7FJ/pWSGEfxP9+CILTjW6GRXsPhotu3pUgT+43eUOeE4BVR9z2O pJEZ4PIaZme7tkjJehC081ScZDdNnOOoknnDhMMFVPish3ZkaMhq5rWQaQkWLHVIcSYdoHFtq+K6 N84ZLXIYUXwbQIW/eSHRGW+40/LQbzmYlInprGhe6vO4f0HSJ2X4W8R55q8rYZjJpxj//FlpzvFj hzLoDWo3719orIafAa9ijBfm49Smju2q6w1kf7jXcq3DbburobNQbCd14Q9FDccHq0mEba2ALwTH MjJDpa4cTBwtZFwh121+WyYwM0dwXEYN5pTVAtmGLXjse2BEfs+lrySYJ1ko0ey2i/dM0T5kEkBB o0SMayN6xbUNAl5NJpVVi7Ar21G64Hkbr9de68gE0E/zjwwUl3rFKTPTiEkWh6vLNdT37JibKvvF IXWWMxV0jnXFkvBYLQ72g5y7tH3PTKygg4TLXgysmZzC0RL5p+iWDcwWMRe4zzsOmQcs160dFsAY CcSdq8nHoYVnuAo02ssSznkF97rJlljqzkiOG6zoRZF0ii84HLUV3GiAmIcjCatuoi1w7tjs7Ged bMz6/mEuOlUu174AbrfEu6UFDBwirABKr4c0oxBOss+b0Y8sd0kUQMivr3tqxyBPDnKT3jj/wPew b0nsy0H/ZPK6iZdeqe8BvetPX7YButC59DlvU+eSritaIoy52HFL0j7LdbvoNIcPlUUWemfFmCiX UZhwabQ5VELz7/dG3pCSzH0H5oQPTX9F6COuCBPpIxbRfSSWABvAQYB+ddsJ/gFMRlFCnX7GMDT9 pmYeR7Qe9styixfkLrqboqmVeHPWWQIRc6khZVtauimIGzH6wttUX8WvVqLVLNv7fwredjO8LdQa 27YJ+7UpLjOsAfvGntJWEyix6gJLtBSEVWLtSUV9ZP2+20tsEVi6KPPSzrmq1yDr9KCqeYdyt/SH mBqU/5E2735wt8w+cKoSHVz5mcnOEDJAAd25Q0EgMq2kXIeXnl+jv7NMBq1v7yGzbE2r+72nxL9f gZTyjFG5SZEBRJ/hEjF1n3eg5h52vrAktoh7BHUYn/yT80wWIQD8SsOAOEugX+koTSWmTWPSX3fh ERWhDyWV+ApsEx6mKRC+uWK7NumrTtc6hLu6+AV+ZvmQXekq0E2Y538sKrQRBXmvR3JQ3Yugnuf1 tCgnTTc+7FrHpTv4PDOaoeCiGRrlhDbkhBC3omCq6MnJPjr+DMxH1/Jy3OCyXItbQ2HSjm20ODgO tyoreVt/TF2kIRwnzpic0M6Dq2/b+Sxj1VxYFm9KKynLD2tu1Ymvfq1bkINZaKO1Q3XXNJSn/b8k UgmTQubkmrpE0Fq3vDUrl2PEEMwMUj5sm29PG5/4q8PUfGFHSXBP2Gw7nkch5/kyuA3rs1BjnNAg jADjoktOpDE/3TlHPzIlKwO02VHaiSY6Ke7jcaBkkXhdXlYo/whxvFKY9sFe6A3FJQhmm08XbNH7 BIFJwvd4t04bk2iebWgbMhc5botdjsd/EVkwF9aY6saHJdiNo1LiZQkOSij5wicFFWoy6VP36128 rzZ+izX26v4jZLejy6nJyV+uZM/ypSKYFgluR2riglrba2F7ons0DlDTtuMk8rKFRBd27634+a2P DW9piGGn5N/U0t+YquYJ+nDBhD1+tqanPypI3c6zuP/QE9JYMPF1tWYjInbwZvrFRRCzrBsVySXM kwGbhxhEzZIuTaVF4XGiMabtqVbQl0+ZnjV1BnfaOO5md/82E/N2x28EWD028HTmb8KzARKmSDMV 0BeL9qGh12cCkMuL08EAM6h2xqLJYZPVbA6+OZdyR9/e3qimN484KVlDZ7wx6FgtWem/2ax8UARX Up7rpsP04h+hWNd4F0a+LA8kXPHiGjWGdBD0IN4EnjBIEx43/uwb70zHxaQOVKCWxC01VXicjE9i XfrEVd+nc574pG11M2BLzsM8v8DCFUpNPzQwLPVT0Z2op/bePhlcHJZ3SwRX4/7jsvr9xneJL6EN 6+8/psDeM8wh9vVGGw2LTb7gvN6R+pML/dk9op+ZiG867Wzvu0dXuwkOZaYb+POB7qpx5vxEm+LQ loH63xcU6ceR6o74oo4DaFpllBYOxnJfYsaTQTpEkXuPvJ1V6D+zFsrRK3EqSp/cMS7NUyAUyJX8 83gG7gvBjkmY9iMyC0aRgLaQX+dCy7ttAs3kjtNm/8XqWjrUQPFXxBbUgYPtIO2tpafL/9HKYqPO fzteL9ceGOptxtyyc+mh2iTlHfHR5kw7CmiEkx6U0+rWH9jmIceQhy+DhLr5iDoLlVIo9HWob0zP Wp5wx7Sa36xQ9oW9DIkZw0/iZe2cCya6BZHWrj5kFwpRcsWbitm6o3GSTw4DjFShFJ84HNb714LI cSz0VSh+0iXyAVmciJib12Gr26UtsBKOofDMV7XiwJvc5Xw96jAMZOB4L+6LsAkpUNWDT2UpFgnU lksDhyHwtOGsMUQfJ0J6R0Drr8rcGlS+UodZjorqv58a0x2mgc7n9C3qgSWsysQw+Jxkut3SS0VA XnDltbINh35lH8M5PO/+YqTo628YDz+rbL04D4yvxCHGooprrHvw/PmNJ5HVMXzufTzduJN9Gsia xpRWqDjzq9Wujc7H0ymHYkScjaXf4JfOLoMzDXCO0MYzNxzR8FsWXxi/0ui01yoB9aRLX1ve4ynf E2zqzL9p4o9HKpOw2bmXbKfOb3ud2xDzSc51A9b/Bz19ptl7CFmSaqwuLbAkxJJ+W4EaKgbAs2IT UVvESvhKFXfSaNzInbBXh9k8XgAhFXsatCu+YYtL19zSlwgbBbxIMovUXilQ4jvGvPMFV9lsD3UW cIcc4g2ojndykg+apP6qS/EtQxGEo9gqNF4jeELvCD4bwd19/4QJot1wRfAkA1J7B7WvF015tKjM sYuPClRJtbjF/smSTtDeprGo4FMxgUDE04DT0uZ0QdW47JrU30Poi7sTJFdmbAPfRbtIYCamve1Y bGWSJXUPiNhEkOolEFB17wqXUzpoSrU+w1gT3xrkBEmXTkIHAtCktBV8/n6V8zyHjdRS4g9Y6jPk ViAM7wpTd8azxqIZeqRjKl1dyGO9whqrxX9wvRh+bCPL3x3b0/f6mAbsg4u94K0U6nfG1gi9lpc5 gLyn8e4bo/sVF0tnWUglQ06hIC10INJBkwrblYuMTW0tMVo6H/8Uj1T4diXzCveh8FqgrNM5dTHo XvVcon56f1F26DuOE5m+PpH2X7XXh/yKgb587as44YDQW/51Fn6yjFLd3FNGxULcF2Kp08QlOtyR MwbMdwDiDBgF8nK9uFkP6LfoQmaaY32zqRu/vS0rCu2vmdqGpNVLj2fiJHun4k/pAFaeaH+wr64u dA9iezySfKtXVBZpHTSwEoRbaGeVHAMWQ8tfPjh4PEihXypU6uQTJ53xxadNSd0pOHpHDGqlq6Oc fmtoY9BK55I29tyVRlYygcqQI2RM5KhI+qOMGYdr4lhkjfuYk4A9Pmwx7/ex0t10nS1jI0N/XX68 +RzQP1OJHPY01FnhuPWLYz/UEPnU3pxRDRxf+BRJhLXyMyemXTmojvzaxH2EmiPobl4REMX7ZcQK reFxkVT2rWDS3WRdDhpnH/ZHpDNpgUEhMnShmU4UJEgHrymbt05i2pT66yN4kJTGRe7tH/2/VKZM bbk7IeYdAfyNQ7in5+Szv/YH7IyuRETRhCRBov7LjGQ3yj83/HHFGPi9xcuPUInXfud2fo6mBiU2 ERYimk8di6rXv9wgmcktmqYqve3WFB2Ney8YaBj2qpgauVAz31Rgn2V0ccylBUSn1guyTBm7AeKe YDLYk4ugayq0fFSUhMryc3Gt1QEFQS/jXQCSC8gHWaNMyv6HYN9k4nIUi5B8OIqCEreOTak6RwbM ooDNetaV5z0hkqWAhHOQnkGAyS62o6nWgJQNyMnnci7Yn5dLB23t9iOF18JNLuhg7Zx7T7aRLKYe EYmF7z9wrdpxtd6Lr+JxhyLlUTN32vjFyrIBYVjhXRnGkFT/tySUydgrBdhH/iY11YDJrvTZk0k1 WQPIqCVpleORXkxO9PYHZYJJW12tHB4FbYQC5uoVCIyKGOw1hDK2bxxAbJIEdPLXUW0gwY2tsHlJ 6a8ZjZ07fRkWwP05paeGTlOP52UzzR2E3hBsafQkEJfoXI4iIt1YksA4nXnhwYICqsJPy7E3NJ6r B4OiN2zGAZQSZOD0fTplwBKsXaMrrxpmdiV49xCM/5EzBRPkNBuGta+OcmLncuuDSBPNsLzXE1+r E6toDT28/PX2H1ag19kINtATD4SEjdjvx70ELiSNrUkGIVoPlIh1r+yhd4pVMVTONLHTFlO0ygso gtSOI2LXROgrbH6FU0ss7d6I9zJpivpS4jt1J2HVGNYkMjZKTqy9JewtcZtJ7nb2bD8PoWp72Sb9 ynH79DxKpTpRRzyGEdSJD/Z1ZOUHzYG+OaB3cxPsNaE0Id3AdpkeYRZMQzzeD0qk5ukIjiFKj4OL +dQ7+Hx+77V6j36A84O5UzuBKOdBobzOEyNA9//FS57FsTz/u7ai7etghBxLoZmjqKhlWyJLlHZn YVc3aAA+WqjV3E5EmbaUGOzIUj6mQ8ezE+tXIbYLizYUCvPJly23quNarGm9UN+0VZnEOAKMzJkf AR+Z6eLNMclwK28SvWvwisPte0EIAB1HVrtU/i1eXFjQRmZ74GMln7R8a2cf65RTgILT5SxQVGTb PrWcg0l4D10mmGkTCf3HsEIPi8lv5itVsiysZRtkFFBd7CEZ8/fCpL/XArRoOs0G62U+ZkVqHKFT VNKy5EYir9IeMOMdaPmqN7gcTOPgDNp1jaG39/KNneU/orWM6zrr2qb8ErjQ3xOK3mU9D6BDSdIx BysDX2c7ZmPQ1krQDNw6Kf2MsLIQEtOzHimfLoKzqE1G+mV5LWRc4tx51nCYz8aylF1cBVixFgib 3xYJzH5Uggjlr9sS7f+GB3IVQF5o1+kDOxVK/cc27dpikrcIkMWBifXpRII5TM2IFcKYaYlDU+8z AGwgPcn69CLBlNiRSxKqKMtf65iAlWljVUoPVd741eAIsb7tIWSFat1s2hiUz5XPw/gX22CQNpM0 gMcYRxCCzsHs0pKiXnzVpG4HTQMkyIM2PhMnYQtOHw0LXFP/dlc9UE2xWUbzjATeT9ETMVQCJEH+ LNrr7Sj7LVVciyf6xGX6Eo9Kl3JA1G8bD5YnPH3GkhWNrqbZTCkz5uR0mcadAf/66wIw36DKoXk6 yfhlNjuJACgNfVJrar7n+6xPQgy4/WNomDf3kYps47DSNQyseLtE+SHxKPo2v8os34GElrw4id+J hJXYK8hqOlX8am+l3omUPI3zt44tjgMY3+0YuLv9G9o1VJkQnwYuscnViNyiJARicQn8ztFZrpeQ OeDbdfl6wvbTqcQBynzPIAschB5qvOArtlQ9tQb/2A42pTqGSXvmxOtuVO6vhVjeasGGoxLa5CAs fiUDgBLZA7VoyT4T6RmLRPpsZk20GZV/Dy1ALowMWt9rbQEGCSPZrrZz1PJOfY+jMklgBSuA5fsp E8al8rzlR8CLjvt1w9K+XH5m29Mp18G4d5Dzjb0lnoTtDxdZGvJw3bUW5MGwOnx2CI7jBNl9m385 ZTAQJIh7b2s98syPxbgjgrjosXXcpb8wTcq3nuA1QWms1ZNhCTpnYgTsGHZPlKR6OZgFIKVTDMJV +grmvBKdnEbOy8ovkfUGyNAzt4Gpn/NDBc7ekMqsPKOAAe9NjuKkQCckFORp8Xtg/YozWZTkv0xH TEWD8sY/I8BRptwzeIgDcwoB67PJqwnC5SIJkS/d0fq++wfSGbyNh2yCsqwTYYbbfQNpCa+E6EPi 4Gg2l5mZMfvVQklTYnz7K4Kf4FFy4uXOuUzwkursVXA3AU+itIZ3ns/xooUX0cL5q/9rj91+yqrC Nq9uXNGO9mxoIm3P2ZQeG19e6kF1MZiXQXsMwzhZ6SB6rWCiwxJauPDa5538L/LPDpCsZrwxkq8K S24VeDSNphX5ByCVNrkafVU76ZicJY6K8Avx8z7tnTyoBGhGluRnNtpEDklXsOjM5Or9L+f/5V/k TXIjou3GjPAxsnY2FEnBOTBdhyq7HhfinnGcmJTN6Hy7Q0YK+PzWw+nySdk8B6aTlCyPu+NSNvHW wcBYsQkhkJwb2i2kvhOrn5EpReSKd5TCPY9V0nCijQnsbYG3+MsKj1lQWSqUclWn9i/5fh1VwcOt CdrMPubE8hHJwEmEGgPIe6Q4SqsPCI+/ItdU9Pz27ExU2rauSKuccpZJkduWKFu5UZUs9rKO9Rd8 x/TetnfLF/xPhgjhdLL84aN0hYTjR5WGeg0iOm3ootA/+o49BmPSJ4y+hSjT8EBaX6t6la2hKbyD YVyyKnf7dJOQquWDPofa0laYOEJkswOSzwnZ7Ebj5vNXJDt3HzdqIsRX9m9gUAMZYGyfQmGGpgUs ssvCUBrUiY3NxcoezgHJGfmymP5ag7VH1gG0EuEcm6daCcdO4dCGHfnBanWHpOWUhgHLEgflCu0Z xeMrZQUkLyxxe4FRbL3EF5TNtu708sgA1AtphE5mSZt+Be94DK8zWLD5n3m959uYqq+DV1QNVVzd YCmKFfzvRf/9U6YcdZq5LorqkQ5v5UtoJOzYA/oT/fPoFt0oxTdPYOHPIeReIJflKIZRsEd6csbF 4F5fqsum4Ypx0/14Xp+TSh83P0AG9/Csl4cYf9C58hnzEdPnFo+mktLHlT+qJL7zt92MS2+FW2ol t3wLJrAPvoUD4xspU9dqVgSxTpe/WOpSJCvgeG17aSVPLUlK4UeCHFjou5v9TquITSf/hNHf8nM4 Ip+rbbY7IseMXken7sdbmGusquJQ7Y4dba8sFoBUfdvgENooWZBGtceQrfb1X96WXtmKkd8a2p3S U3KdYG47UfZ24aCwYRyikZ7NFw83Pdvigb7/ACD5baBYRjOcWIRvxWDqjznGuunQQEvCm96+9Dl8 32tNcK3DC2gCD0Vj+s5uaiELIIS92/zI6xsnG+DEhAA7/NA1ssWvXh/2SPGEWKrGamUk5EMx/V6U bzx7u1eRHWsyrRSaf+XQ0ZkavqzzikHi7dKXt+CElpc7pl/OfUUqv8zZQp/pxsJjg1w59e+kUTn1 lrQ4ecXWECjFC3rxed7JYoZ1LgW+f39qDy4wucWR9kSDA17Nw8YaLX6qB3iwX4SP5BeQvxvW9K6I TZ/cHbSH/ExdzLMnJXa6nDOco2JsQzS1BHcb84bq3HipRXrkelzHGX5WT9VNfWiiXoU4/mPaaz6K 2pngP5ShailBBvjQuatah7FqB6nK96vyTvGnayjCPmeXJ0fpH06wU37SjGhBNW4IvywWtoo2iII+ aEKenkV/M0JznjE0pNISXr+DLgB3Tnrldrd7/kBUy/45eEkRNMAotJVnH34VpnCuvQWC7ClA/vYW I4YVUVZh77pAV9qBfBoqnx4Z3pRFi4A8qaLReeH/ucUAXvCIf+kgVqtMS65jTr1uPghkV6OlrZac BddXqLBSa3ypX78C9Jrzi/wfcg3JRjfA7LEA3bKnTJqB9cYBeR5fUv5YS6CUJkLaeiqlt+dcQraU hTypkMoyZpjxBsh8mYyQXfyfMRu/qBB0jk51Ev7lXJmHZUq4p2kOs/6adJJWo1p0/kWMViMHngtU sQN0YY2U3DU0LGNnHwlqrj7nc2emgpzGVjDxVKMTh6wpStNX1a1aSqH4cH0CAE3wIg10MDa+OEB4 2VEP2AwloeBYS7Pg4u6jD0R5CT9ETv51o1DaAamN167MF64k/etCF8jaRNMylI3+A1IUbMAKww18 ECyGcv/DjC7uDoNUbAgusvycEnVu6GSjIQvSR+KDhvAt4Gyzv2QHUUWSsEuei0Y6PeoRyzuOyJYB xO81uTABxAMm/lyTXYNUdl1LmivbQKfiGVFaT3q9BFsdUiBfHgOFqmP4/MJZbtjClI9+nR16bYHw Gpr1lzYT2utpMBlEoUJPP31nCRIN+TmyUAI6Iiihv3Hkz7C1kQ7RExjUo9hdzXwTGfyuz3JBiqOB m5HA75EkKqqRTlJJz8F5MJv7MGVKnLiMlocRyVJWRf6fYRsgoFqU/M5NAsTHnhX9PcEtDYB5PbAb 3eDg9H2Z7POtDk3pG+SpIVhaJkzfAPDvXDoHgXkGGxw+RfOGz4IIXhOJqhKAAz+zCfgRHntKaZ63 Q3wJfP8j181hPSEjSwT4qfuUkUmArrj9azrV100JG5alsbKErP1G4tHlGDGFXMCy2QzdvJVR6HLZ G4kjeJBNt9FLWHcK1JKEo2aJ2gjwdT1OtyEN0E2Nzln9Rufwq2ophUxyifjr7H0PEvne80VfRKCO MKi5s/WV8rRgM8KW+6ujqFX50AAwZLP8hAEh36Cl2Ipu3i/tJ2oIfKFG0/i+nml9d1P+y+LuRe+7 1uBeJHL4DkMCjcrruD4Oxa/wnOsMG7MqSS/LVDUqvL3YjVJk+/k7Tdtiy8LqVuv38vNDoQ+IdHuH z08RPFLCLDC1ObN9zhPmVSIjBKlL/n861bBZVi9o93fCXhyVRMAY8J7glxnSdq+JCRXo9XjfpS1c O2+i6y3BHHBFUEnVTBNwXFLwE5wQRVFocU2nwwQuSJiEKDCR2/MwDZbPZtaW6+quvcml5w7EumKs 208/zSjMS4HhmZefQTFUYXnoWmrVwBJ34hZCE8iVdouPwUnbxm+StFMjn4Qrs4zMRW3xCnDRHio/ kvC5FsP9goPjMFpYXaji8v6zSzA3zNaojTAegli/pqAvH04cPGqIrMer5i7F3vpwxV8L+CNIi8Pe 5bnimfX6VFN0LukyMu8X3KUpz5alt1wDuauPfi/4thUInGdQyC8qnSs1UeMT87rcyMRky2y97ZPO ++Ieag4abDkY+yTAdveN32tjIHY1yNeUpbm8saAOsHEDtCdkOaBQS4uRPwq/ZB72U7L1h4NqRGYG gdO9oVInWJAz9kHaZwoIptBWevgEQUDASt0Uhkvui8NHt3zE0zg1EE9PJcRCnRw5t4EnbGynKjBP heXVmdLw6QSOwCmNilBBpBXAyVyJ/La6DQGmAOv4gYou6vDip932PnTs/aCaMcWEklvya3+Lq9An 6SrxiIazTU417/r15tQ/KPEAXXmYU/37EzOrLbJdeHGv/EmScuOeIekZSoQvCRVNkARkFwXjsmIH NF7HzUb7MOy/fE0E2tKsguHppXS/mx00hJaW8suUtiRk8fpICzgALM3m68dR70WXceJcBd5e/POq 7LQN60Ra8TqxrVVS/w4579fOQ07d/e+DUhhsmrE9fXP7wTnKNKMbog6Sd4aRUBEdKftZ736Xl9Gp LQ2vhmZli9VsGCvDiEGbvW+ZiMye0kcNj4XHUx4M3v4C1IKpymM94zplFiGAYfltc7P75xOnHuPu wsCN5fesAfG+fltcvVwVuiFoQEMWxFkTqaa1S5gFU/q70QbeeqnvwPnYNpJSZcprFmeVCJZb8oKD jpoSEYGVaaTlNdxinVkxKGtfGVddYGa4rfnWiQ8Y7D1KoOhoEDsAEcH9JnKgr/Sg60TNX3YP3oHU IoJHnimfraTYDOT+2pNsYuClVTw44uIIucMyg5pA5M2gUNGexqYZsdRIoFv8hbJw/qKO/MpZMVNq bpfFq+dXvOddN8CnE4h0PH4trUWyYot2VoHHh8xChCmioRBhcF3Hknvx7pPmgvhF16mCquFTQl4m A/jn7ug0YG4VV7cA9V2HYtLljfNpp7U+PQszEwpy7bYFDwVJ6a7BYT8qXX9m6KA7hTtzQNxRCl5r UnA1NMFZlC/ccVttuBSJeJmDanf71MFPiHpOoXjoQxzouuAzSy9CLvN/HPUpB90fzGnvMs6qv7Oh n44YQoosv+BhJiyMNhnefqZEZgQF/4URpUjZ7ZU+3kAZQYGL1ZlY7yGDJ7EF/hgW9w+AhYbZ7k/x h980yHMQjA0Cnx4h+4jBDJuB9hxQfrwIa7RJHA6acwKSEYd+1rp2GY7TELLehDWVGSAZKlM4ZOkl KnhJsA+QayFEP7s2Wg57d6/husnNrh2unOY6Sca8OtFVYeQldZdm7CVKWP4gqfmW10eLzMaWZOL6 /Rq8gj3kSgX3cj4zn7yqNPzSEFczjFamSf+AoBBvz/dR8dTvqXc1TOWzK6/bW2zKA+0P9xBL6Ps3 6n3V9nuhxkkPp51jTfy85q7CP8nfV0RaFJme+gGPRMdSw2S1EXrQYqAPjhgsMeocZUBWuWPhj1BL XCziFZmreyqfEYd5iEYlzYBqNSzfJjo+wHS5dPrtYxAaK6coJ6XU1DozkMhwHb1ocuTCz9pCTMTG 09ZThukBNKPuWCZT59p3+jImA/ittedspuKj4wS6sQrccgK0XzbACMDg05QnUDNC12Y2rleff9Uw jptVA9j/OUsecVfn2+VWihEu9lzLw7N1zDqrw1P7LcYMOBOY3L2iRcytd1cGq29w+CuFM6AkRQXg EfEMZrvN7QZwJbMMWixG9sZa7KHW6+F+++UVJ8BB33Sq24XxE01GeGK3YD6uheWZQw55zeZhurTM tKi1brIoD2QSnA6tS8Z8tCV21i0l5FQY2TfR0ipAklAMw8ncWU89YTKzvk5RW7FVfRih6iSR5Vjq SYWhYJQl5dFB/W2A6P3V1K6XBqb4UIpnL/S7fkpsgzSe54xISLsb2sNA1Eh3bVVie19g4XHoOEAa XCRfPr4OdKcZ8XiE6V/lFHogBmZQ/3BUHjsiU13oO/ldrCpAP9T4p6PvtXa9ff5Holooc8HWPSq0 I9XKZMQf6rlTSuyaOF3Np7LUQHYP/XroLowLpEZl2jUPaMh2XjlLM/pJ+Qvr0QBiAumNPDl6gDMI jGUnAP+EQEo9VUUgV0N/f0celpRi2hWs1fehY4YjME4E9jdoJEZTDUqr3CSLlN0OHQBYFI4EWXO4 5uSLYIlR9kqaoco8F55mbLWfZsUywfAFMBXbEU7tjik09txeB5BkFTUB2Vmgi+i+pi3pt3CP0t9k FaUAxuvvfkLZ6kvCaJg67a0N+HpWxxqfm9wJRTSf2rKCvyN8CnRyGS6URYLe6f2Rj10H9PTGU6uV pLXWrLYbYLQTXTGxGl79lpDxb1KRUshu20bbDmcftgFl3EQ5mrpfoVdpZn5CVx4xPAME33TE6e9u cOY0kBMBM2Vao4P7plvclIQI0e2fulgoW4ewhaNwGSNRaOa0s3pHkktTE47dHdrbLmsi5R3Bz+px FEu5m759oVMu5nJ2O6fI6o+vPyf960QeI+tK5BoQD0XlYKK6zNoR/a3Rlnw+hDdkv3HZH+kn/y7U 0vh+CU3VbvGXK/lgfMUa+OiK0qfB+r7j9llvg+mVFOY1g7HamWUDS5VEE6iJkjC5UC4arnc4a5uK naMwT2Zjigc9vLSUeqtEJ0WaguF1nmFyeJU/IyasikY4H1IRmpLz+m06BApWaOwRPnZ4AiJZg86Z j+ByHJ8xYBEFZx7Hs/kmoB1H4ho3h1aemGDxGPj1Yh8nNbiYh7VkpZsbGZN0UuCaSJyiJ83w/6h1 kjlUp4lBr0+SJlr2Kwb0kYUiLmZ20E3Bzb5r7TNqM7HzDAdejrR/cCXd40BnjvT/nWN925DVVPMs LTW0oTTF/EjPprMW4faSbumuQr8j8GT7BacjmKthhTvixLjlOW+GypRnJPTAE69PHNuGizlgRzu7 IQVj/q/JM0V3sAuFJjK+xuyfp35vl7OYkAmAvT8/S6YejnNb7VlbxakXWEgk5hGxZjFzMvFmJM1L RnBvCDgXis663OWei1YXqz6AJAdxDxAZ7esimq0RhpKAf7JARIUSroQTi3Ov9I6/cACBVQcUlWuz qjsQY/r3YZwN0RuMWTiSzUk8axzF7hi5lDVaZEYqfBAgtCuw5EpnD8e2U6n6s1NlWzbDtPdcYIZS W0jrheGQQa3GNJYEBjT7Igxz+GzCufom2Q8tGXCV5xJFA7HsY/f4KRAOhgAlhW07eVSnmPDDd8kY AIrwJCNEaCsda/DV/K6fJo+GzZrg8gWx1heZjIs+6EoB6oP4wvTXrhG6u7Lj5g984fviTt+5G8YX YgunvkNHiT716ZIaAslBikBOTj5H9vud+QB/5/HqSwmy/Y3XwULiwlM9jYZOlTS48KQ6rFTS6CWA 2pyTQmtrIQDI6e/ljsC9dprI6ruq8KMy5QyPeGBScbKXDuipR1fOeefBA0HBeJwHujs/bSW+MpBM CDMfgHItdz1/y/VDbSxh+BJU1lINZ8rJ96221KNx5efcOOEL23xPQPerlxu2+/1XJqJuWZcvDsF+ y1NbXbvmoh70bGMLGw8cQf2gaRRxsV45e1e72EDP3EwMcX5Q3EuVwg8Q+WQggzLjkZvZQ3YGQ7VX cgx07HNqCGmz892VctpF/d6ZQf+/juLOmkMG1uWphqqFaLuO6ApUmUi2L4dIxK12nCDIzWIlR9CF r6QnEx9tKQQyd98B4kmnasxNUIJ/eMQ9WTH8RWovPMLd70TttIX4B0fjJ8A2B3xT30pccOw9wWxZ 9nvxi0TZ3ggxvO03zDrzlsEb+cmjXPdM8r1AhdNOyIpI/2sJEt7SbvTu0TKOBDj+Qnrk3xC6yz9A dym8YJrbKvEAV0NKTbJBTMsAfN4GxlwjfFjmAmldYkB+oHPOk38XkRMUWBJeJ8d/LZFtvclVVDg8 EEFIlPEbPkXKLdhYqJhvsX0FTZ/yZ+e1EBpQ/09lAsMd6Jc090NalRel31ZIFFhzgOSLq9BknNXn tchvEI0g9MXp3u2JQZNjJCCCgZlQf0ML/TMCiiv7gVsBUs69B0e3EbcC1QSIFXZeHPDe5yHJfm4F vUDmHVYHpEAAM4W1Fj4mtAyZjHTYsi/JuODYtut9gdBMVgeJ1F6WmD7kt3DlF9xa5TAVl6hSoPS3 gLbsEU75fKnDmwBvCeZ2PSmWMWgeWCvE1jilCCa+1xc/mLxpiK8hg407FGKlcpPEIIfIunQK8+jX w/nUB7x1KWsvNJeKtMHCUbMu88pZpUvUb0/gDifljzkLw44r2nJUzVJTuFQHCeKpZZGg9+q5/Ev+ TFGURBykBtp8iyluNdOe5sEgq5JJWBg8fK5x4MZMrcElRD/BGFWzNU4rGqsfANre4vEHiK3ICas1 InH3nNMiKfQAS2Y9excP0fthvp9tKUIOOi1diDgYW28FqZ2iIL+aez1F3tm4pNvLPZqW5klrjH97 +lIk0lJMTw+TjEzom4o/etixNKUkpwKVBVSGkRcHohbzoMZTz+eFtsBZBBUbi6eJ7FwwIXjyMCvF Qw5WlI0rGtBTljHCMZH8Hbv2kn+i770FUXdYRJp9XLSW5luYbw6XP7JbRcGtu9LgrvusSefSY6gY 7ZX/jrQS+QC+7ZmL5Akcgr+j89xd6QN2CyWKpfUDiAs58AjGOx6ej7YwUrxeEMa5aABULnghCt4C kNxQeqzKH6VYgSgH8QNhOCHGzclD9QToovDU1DIAi+7NkAGUH0t63XCvZEIX0FvdmbUhMjSlOF7J A4B3jf958UtvRMPNq69wiHB+a29rpALM+FKBdHcHx69xvYfxhDaHM/UJshTcH3WM1dl7WwdLPtzn mtKCOXpidFgpHIGqV+pPzRslaVZumD+Ls8A47XaFC2llCurn3kdr2MOHS7bjTBFn21SKzvWBE1tu mslcd0I7B0dhet4D/ShVu2aHE+M6ZIWXu3RieaxBEi58/AxBo01r8teeD9isr6RgeqKJRQpfCS/T tW7g7EJzQmkda5eQpMMDt/OaLXO7bPv2moxdexxF53i9rgye+TwQehECFuGPv99fExJzrFe1PTz5 LTqtNpI8VG/UubSsrYjl/d2vs0mU+yhRbtPBsnhyJ/5I6LoE85J/m6k7geCvEWqsMAUkq3UvjIWc UQZatx3FTSCPtImnuzmiPkTgfD28N9OSq5K/GxtmqIUoKqCJqJKkwemgeB720bcAkiLobdE5DYEe 8x1IGs7PAREdaVuTEvUofyI/1gIdaJ3e9nvcD1DhTzLH89mWFtWe3PRdA7wwnVXj2hVKyE+Cqepk iGOqi5iR3Ra6BZLFmzTqxOaDFUZSK6z6901u2ECZsR40hT4qz+LL/Zh7OMfLwW8kmq050mJICsoc ZA9myP1ZdE7b44g+eEerjOKk6l+3wtGqvR1qc52h+WjuotmRoo+4atynv4FSQrSBTzoNMP216dqf aY1ZoZRf47iQXcyJRr5kMB8OqUSTRskHaRUzhNAlpWLMz5q7G9CjSOC52r0EZksUvB18a1zJrQk/ QjM4A2JZztEHI8FWfGpb9APEiY600tTzgsBz1fTGIvYVByrR2cRgUetm9DjEVZvvxTuSY6gscdM7 aCpDspc4gp0pPks+p2ITVamymrbqF5xQDrX/DrdpCYFdhp0xD4GdxyCjD6vwfgGvzDmSfl+DlwGl Ol9Q+3lDYfZ/3+eF8XUcApIWmw220RY9kBn1t1AxK18Vw32ql6ILVhz2NCFnDVQJ/dcY9kt8KN+5 kP509l+IwHKKOToiJfIA1HIkGW1AwWgg6YbzvRQTTS/6QKNMW0oojocDkKzDblF6Xq7mAvUH1V6X 2p+gQx21/udlW/yNBi6TNvlCTCVx+oCP0zKvciUsgLBqS1z8j2vaqMNG4T03jMIll9j8Sg6oKm6Q vJGLOXZ/BRU0AgyDhxSmcaMz3QiDV+/HK+HsqNXK5nXd9M09l47s+7O7MQdpGZXeW6uNJTb+aAwY pRxU3t7udMwAP9X8YOENYzgGz6HboYEJxtrCezCeXzfWcukSnmjuC23t0X66WL8Bye3J/smfkZFc s3T/U5scgwc8e+NFSz5L0iQ4mQTTi9IPlvAPIRUB5YFjlcECwjOzJs1pqSW0Rq5anuK7RPXzMkb7 PIH1xgdx2tfnRgz7RzEK+DNLFFhEnX23LDorfFBslc8ZJ+x1equJbq8G+mJRXt7baTbQyt6Ay+cC TDKOkDzGm5Y1yXC0oi+GL+5YBLRwqFlU3LmlUloUSCIWjyA8tf/DyefANO7XhhhSTT/ZXGU21NAZ oqf7ZCxLL5xw6nATaCpNtILuDIHa0MGKb0P2rbFXPe5EViSm8FALcqB5D6Qq8nvQeROnhMs1Nnb7 h7CzRVyvQ3w2J22yk6+ppbzoiAtBSpyjxNQfQWZPyBK0Hvue2cXqCRl3Vm94OPjB0ZPLNdJeSeLV /zbH1HoSu0mLuAdnmy70oyC1mbJKJwRe7xKsmw1Lr098Flv2VVmM6zYOaEzo1nR1BVhdNllJcyjQ EMX+msvvnXaoDCpVOiDq5IfZMbhbwBAB6BUlghgSJxoIozniAWKKvVh16TV18mmL4hDqWUX5NG9/ efjYSuKE1+DEH98nHNMRFfM2DPfSPTWdfQXSIsgExG71CKgJlFOw2+GIKAUb0XvTpH0e0ywsVwee HD47o5a7N9pA77E+5rPrC/FbqKAn1vnTBECVm4gsWR4ONGoaLmSBv4n+hyjRM4AYh50NTxowPTSR ezVFTJ4kBN4mBu7M/Rl/XbQEaEeShlXlKzgRKTFXA6aLllhcismNGdtXSgEpR+Qa83tK6k+i3BIW jrWlBTkMLt145nhTPVGq2C7u3YxpPkfe9yj1aU99WfEnVG/lIdBpSIg44U/k0He1aAy8mHBojpea uyQo3ug3If9qFERXfK00bCec5w1ikxjW6J7jmY+isWUzfCH3uFgtsuL+QIYNxX59tLXH7hphWAKc G/1NihqtP8jS6n8kd3FsQeuVP2PSKc1/wgZDG+76tX+8HTrMAYe098wMJXt0kBRKoSMyuoQqWjiJ mxJWTtJ/Z/sShgCfM0UAlnDuoL0ggJeR+OfiVnsEkNJWG5kq+hrswErArFSU83EKSGwOp8XEgsRA sRwMx6+TtyJyOS1UAT2XSV6omkMl1GoNWX+QDiVobayB63QuPqKrM4qjHMdSddUO5W4w5oGHumgW Y4bvz6xHP7QHj+wyrbiSYeJqNwyG0QbjwkUPsU+m8O0Z6Yp/H8TNwmKRi5jHcQFuUEfxZR3LmeWC beox4MFrgYDirMxwCQc6bKJ0kAhAXKbxFKCEleT3cvIKIZmXtOiNZEnWhxsPSlpw7AWFR+8vcm6F x8L4wpe7izomGEgPGyX+5YYNN3TwtobKCob7L3ji/wlii57KTOLqQEkAtSZMg339ZhnsS7ZZYVJW f9P6ryawfm4hAp3mgSw0lUuS+3kSC3e1b13zN6iK2GNxH5+bgFMSQwuMO76KbalfgdIx1O5hpfjE c31kKS0UuQyf8Z4vvHcvZ0aBf7FV0J/dmp23XgRE0dpveBeLcFyB3qPc/IUxlg6/EuHkRMf9fQOX aInhzi4W/YQ77XnaCyDpNKeY3gJqnNQRzQz3CaOnWdnDHCmJgtqloF9pLk9pZiX2hsXgo3V30h++ GuLlJ8IIb65einviYC94A8yVUd6zApifoxuTNzcmWt8sYvY02GtCnaXLOcO2qj0NhOri/CzUeh8P yq+z4qxl62eIdFE1JO8gH9lX2NaJUFRFCVqz3VmC5lrbPFMTuu8+Yj//EfveMyAXcssyPXGNlFZp NWkfAsv7lx/DoSzPIpfY5Cu80wQ1JnuGjB+tdXfwOy0l+qrQWlCU0lHLrNvY6rwwKU5qmXMWEYH8 3JDYkvfkq+e0bzsuAA2p41CIhmz3b8u+ESC4BTVinFLThmXVT9wpR2CsH87vlAYFZwGQmnlhf+l6 TwpKhrmUPboaUjMyUeJgczINyIL7ru4/L9MuuBsusMw5lQsBoM/iv7pAJ0026cdzP7cklkwZSgjg txSuHg/RVO2YsBpcCVhkkiid22++9UT9Wot/7bqGqTS2AWt2hGbvIgdWBjM9gQdzvjwXMVDEL3fQ e17pfL+DlGlEAWU/IGSkWkMndEBycSbbdGXpG/N8SsG6nUo1BIrkypXD4T+3ulYiWztF1TFSRrfT 1S7EIvqqMyv0oC7zk2NnfvCWxnuCs2KYwRcz1CRxBOjuVf4QFSqstZ6TBcmp+4z8BnTTp4uZxH+V kkVIAoZmIXO7a1QdLKEeDXezLIkB02+PxKgwsUuTCETzu06MTqgBSymKxef0fNLzsQ8hjjBLkSD5 kLydY/grBPY9jbzEqPWU01/zYdJeaDzqT8UZulKuQteKaU1aG1s3O1exKR4GjUqeIad57gZIrsS5 8Q/LD8jkfmZopilpcS7F3EY9mn/UDbc2O6/Smtpfjsp0LCRRqmuFtQRPRydlbru/kIPQah31N2gn x1rTjeNjCGJYukm0rKPmF7JBul23ewzZ5x99QCFnsvFzwJj0DYEFZP4KTY3IHtGyOwyTJE2XQYj6 JalPQchFfiSVHMHkCtfHztEfXszekkrOy2QIHsSUV2hAlBChLFWPXR/o+mivFqsWnsWMPvy6vd1W 7HFQIq9Wl2WLpt9aIgnrDDM3LKz16dUieHvngSdK2tSJYLXTaHpE9NFn3Lqa+BS5CMUeROBJinij gBaHf7+M8AuvO+al0j8qcxShzTyUDvtiVTUqdi+QPSeFdOIq918xUEQAHIQpDdLFFpTiEr734UU5 +vQKLOcz7knCBHNXta61rqZgSNkMBkqmZS13pEDmRjgJkELPicFLOQ88d42oz9s+CN3cj2r5nWXL RlgBfh3kEbWBg11zfZqoOu+15oCeuykY6pvzBETX+9/IBi3DWfAfUBC6RGWjcM0bPw2jlrgTR89R mnIXAacIL8aloLtBfuVecBwCMIrvBXzkTDkMZnCihjtvFLOV2wzghfDhRZzwGWH13uft+iuFGPh1 yR3cqHaRQ5iUVge7Qi+MtGCbGy0Tvr7OsmFVDxzB4iTh8uk3cQQleTpQp98yIOt1N8q2/kxaBUir WxisZDI9XkspVaY5s7TCM7/cvkKeJaEujlyuBk6TpdGsYvJ/uOheqccWEf7Qo+lcG46v1/OExlDH zw4rasA35sDM8zbFfdj/RRjj7T1Dt9+5g2c46LAXL0rwOSMdpNN0+VPS7T8oOq8umv6rWO9YHzy9 IE0H8YFIS0qNUE8GOJbbGETs4H3e4zGiKGmOaSdRBxYLMKuUE7HXM7baoLlnaQyZb8ig7B2nP206 LlCR7OTUXJDy+ckv4k+O/jOAtnBjsRirIbpHmXu7/sZWr3uCLY3rz365AVLq4Kua4plwWKR1BHVZ 97PRuk8K5vKRRwCtbZoNDFjaSu5nGLMWgb5ixoX7I2VQSyVravp1BshONN/JdAQ5K3oubg3vNPsP XjpP0fffvZJKXYYu33JDUbj/344s5rysSMv3361wNnjXEZA1AU79Tj7oxbm/4pOnKblNSvSo8mQd y3SVaUOZEb+qW/bUQPyzXNIGohaXfmNp/eajYAGdt5JynpshdQ8oGw2qdIMJ8zV2zOBUc03Lmy/K ec3KwWMKlPqQ+PYwWqcJylHi1qqexYah52e+Tvi2MI8BQ/GooQh0I24Bd72XwoufTuU21p7ayjYz HD29p0NmXoFctIBYXegQ9f5EeWno0gAeYARbCsN/kSK9pkJwrT+2FBIVXbDmwQ5Fc3u+2DbIs13m ryeGvA01rEwkev+STyCDioLn2YXYGzRfDpNY9II5SUYE6UPN7XdrQHeQyfnJWs52WGke/ci6o0xg f5uG533fKUUSWQbQ4gxVnHdjfPlBc25DFLzkZsVxyYPSuTiLRWytYi9iM+WkigXIFsgDa2jLlkkU D5KjXRE0EAb+nC3b1+OwTcuyHEbppt9m35U36KfrPaPSG+REEefFEanR9Uh8LvRs2npIhHwYawVP OfKXJPmRk+clg38SifGjqe75Qp/q+wpwskaG3x49PMyROz/ImI/qRlCGw5W4Rdd9MJzRnAKOkONK 8PpSttcBBJ4SeKoUAV59F4YTAN65nQfVYT2RpCmI/YdqYPlVXCwO9Q1UU1x8mHH3d/410UwMXmgj uhJkN2SiABh0bmY3whqAu+3YtOg8ud7SpIL8oL/x54bStUirxs7WGjJvOxhFd+Xxm7MqXfiuCaTs j+7IfxMRpCYmCmBXLDroR0Oay+r6tLda+Vh82dlyhykvc9YJfT9+HOJDYSDC9gKPMkiiJl+KTv2j uaLQ1YYv12iPofPw6NMwijmhYxfPayn28FcubhkflvNjaH5Mor2xOvtXPKwg7vGE/zgoZ6nvHCp9 saIJB0lbW0K0IW3paMCP70G2ppS+dDNb95t96ypPI6I23n/autzVLVbvdEVUVdCpH9bW3kKAukcc /shwTflTUlajfTgcJ0Blp0HO0zi+kCaZQSv/4oaHD/rpBR1l26K+XOG1viR01hKonUmG2i6ddrJi AMwryBR0GO6DCBZuLK14YYT01KwaHb/I0kGX5L80Hne9+J8BJacbFhUWj91Z7X5ColL0Tsf4U4ey jJQUyZtjMFKrBrU9uDUJ7mDjR84nuJS1ywQSIxjBSj8l/6ptGXxYIsgs6pljK3PdZ04ibWEbT+wP S8ehBf6sJegYqFOO7Kx1gNuSxpPQSdoQloG7+28CgMacSXpVEcx8hQq0D7o2VbcbZR2UUdTZFuVZ UolMoryuL+bYOcDRumWuDGhEuxxmOG5D3E1nNaCd89Te1sEYTwsV9uog9ep5HxFKJfqrtXtP6JwE 4hylLuirZf5HoooQlmqCrhCquIiIxF7xpgdoa05vE2a+w1t6FpBFefBf/yI0wMZId9eQJDcNJe0I EieW6H/7P67Qet/GzYqk+dXXYJZnKahQtntXtHFjiz5HKP4JH81ULEE0uv52mYhigRBi1pdytZF0 gAH4SB0EvhMQE2jAMWRaU9PLcdb47p11b/K+R7ui3swRjxdBO3pyUQoOrPAmoeSv67A40jbFw57C ybWrIoXpoGp6G+LUQYuB4MPZUVerEeWLj3XfsnOBaa4eSIdiULuJKFH5mki6KlZs87l8hluu2Q4W l/XdJQl9vlma6s0v/Whus0e7wBIXhG5uC5LJ/LdehgTdaY9U73iqS7ENexLgRZ3PqokbzAUW6ZKx 9xIqGvEddAG2MXU+wIFcCQWW4Pwkkd54RbJ+FnvaZCI+VzKe4WP4F66l909974slRny2+AdV+P47 dsG/jKzgkQfIrw67WJN1IjHN+OHn/Uj395sa1Z+4OYLtymrrP637SwMLWHnrT7H2Pvtx1fJAeK+x YfLcb1w9uXHZKShCR8K5NcGV3e5CiAsCu+HlZyCgIjndaG1IwGuBk6CC8jTnfLOS7EgE77wSSZ+V 7r2vNtob6IZVZKHYuU64jINZWjuEb2yfRA1+0fr4spgGz1/tq1sycoC2rCgevdSGBNRXMGlDvODs 9l/DoFIyO/Ed+5XN9eDNnOvJv+d+yAzA78JXvp/8M6q1H4xijTNrOOiwcbHSWRgUXf55frH3URYx J2PtHqtO/N5xVm3ikM0O68mmc/Ic87j8eM0zRrw5q/Apm5ds4DbuByIkPo91pPHsWuA8MK4ZlD2m a4cN0cmRu0m4MyyoNlwt8vlNKklORAoIU2ts2RgU1R9CqcmgqMBmvUhjnBeN2wJG/g1qIa3tju5j ToUed8jZzO5k1DHEWjz7sDEXgZZa1wBo3cwcmSWJmz0xbtyWyu0wJSGS6prYm90Gnh8qBlb67XD+ 23PelBmeIJcM25OazmSh4ClCrINYRNqJmx3LAXWP2L0OkBLe+QgI44DPUYiGBPP+99iO9/buygsI n4ypVYWW6PhbEMukYYP8AwPLkgquiiKxL3DbeW5LcsjCvz6UyE2nwozVdp6DTzmT4j4Xd7DQN+Ng epfUhxjokgko53bMw1m5bLXQLyfbu730HniJenBQGg6TL8Io0mOhbzExQ9fb8mjnZYHVr8DAvJgu mycER2Di0cSkRYo1q41j5DknYqlhfTcaXPNmenTn9Nw7KQYVV+TqbxRewZvq8GmR53MuMYic5ACn P2wx5c7pDQzjraJQsKXxs5sgGnFsa7/R6njBpE9PzsGg7v12MiiMZnzWsZ+6z4q4pz/o1Oc0rvZ0 MoGxgUXKyCC4V9BxwXghLaEfaBjt67S6uTJ8iw7Ts54nv23bcyc515rkLnsOLiiHpByU44+/iCFp YVUvpQacGesW4pB4R5PzQ5PZCPRdPUWa8j/L0t7+EQGEQEVdfYNpZVqMIfQR8zaFPvqvnkSAEDXj bpFMqPpDEVTghNuw7wGtZKijnYPqwN00WXpDnLy8xl7LbdrCglYE1j/InINsQOpqWs6ktox/Jt2B bibSSua5SiYvnR1jnyp8MPLk9SdpB38HgmT465buXLy0BNsjFUSTiL357JpjYt3JhX3NaaR6jLw6 9O5fTfkkQOLUPOOnieOOswt4eqNIByEX2kOQzFps7wvXVB4Ue7VtMZqG7ZqMygbECmPPuvPhPTgy uoCAuW16FSC0pAnYdWIZMiG9eO8X12VyCQI1q7lz+B+8KjykRIUXEChQWVsnomAV37lpEV28Meue yKAhAujqr3QwQfEolQBjxNeCvxelqMV6mcB8SfE6xC7xJZQTFfXjDkYAQgoefWg/k10VxMjVE1Y+ xkZ7/e9IT3T41QLb8z1LCzrLKwkyJe9bCyopIGHfDJ88dQGEG3KVwO+2X7aRblzeEJ7WLODlHSKU c/XQOaXkgB1BreC2SOmh9uqeUepWHs4bzY235LdeWuB4VW0pOXT+WPxwQNTVhMKw7fSioEtb76lH Pd+Qf1Fdt0a55gtGMaXCJQ9gJ9fPpEx0dtCPgMm/1l0bth/rsOcppV/kqT7BVhJtKsqOjTmmlYAD nzivDth2HRGPxFGqQN3ZIN7FNzvmxXzFGHPHrjNw1qqpMbRzBO/EJws9VRi5qr+HJtSuKTXbUw5R TLht16F2dUXt7UuVUbbM7sGoIwaoNjjaDmbhuz887bm/QPljaQcU+kDFxvH9rKgB5tvyxa0KbZJF 3bOvO5YqgeK/HWFNoGsogtyrm3rpdohY11HaN6LMl7wMM4Z0jchKKudnFJCEzFsAO1JXPFGCxQFN gEjvds9iT1P7NJHJUMKtLjxjgsrUbxxA9Xl8fm7+MOjBK82sXMWgH7wNTIhRDjiv4FtebURWQpR0 TbsYrn79BB4verUOX/txTtGWM2Rsl4n4aw14KkuDojw5HQdsUONpDDxf6yfr9d1bmvYNKI8kcViS 3T/ZjCwrmFqC5u36AFfv+O2dhV1I4qAF+K6BJPQVy3ECiRGvyHO/4EEkGb5vuQjccDfTe7rOZxoF Vcjz/AFcOr4pXsLJSjNr6hK+3uK+2gg28tMG4lelz+0YwPxR8sfY5mpXXJGuvUUc+FDCM+FD/G55 5pIAQghjWcVKWG8P61tfZVKA9ElN0hsV3/DEPXIRJxO84W1QXiJSf65UlLEqf25f+iXcuQTYn+zh RI0+uxOPiQkoA1LmqC2CsVdS1DV7pHcDvw4sDrk1cl2iC12OoeB6aIIKukU8YyjkpDp+y1dWlp6R pFoVL+Eva4LbLD2DyxCWVNHwmhjz/nsMl3HnSLqu8HICIsKtep2rEMeBYfdQ+8oMH4dHXop4EwDa RXljB1TT+4BjT6JyrSKcLH2vsLAzdg9Veg/SNAdLgBDALnyzFdtopRl+RVnC2FSikrFe7Ek94sAP GR1KLtsfY8rhyhqAf2IN8LpN9spE7z4X81gkjmqcqbZF8H53mjWcSXIT2xYXH7oqDfD+cF4QVDie lGWblSMBBVeR1h/D8OARXfU+DUBonPcaxlcA6XqGF0a3ZpHxTrzvyTa7kVkRxzoAGpwPqei/kQzS /DTNK1o2Qqu12X7GnZTJfGGaq1WkzvCQLXOai5uiWlpskoDHbbnr+WrsCKALbQr76WX/rsJIr/kS OU32fTkgE8vBMbOeICR4k2dA3EuUkCbNMzulnwLbQcdkopWrcNXTF+hWOKDolsn8GcGciENKKzi9 bXLTKrZAIc8GBMbM1AQQZOUXaV23NOrz4pMDW84OzX7XuCPb31NZhetcCYqf9I1JDnsZ+8KTAH7f fmZTsnUi/u0Azso7GUyTIdV6GKiC5BRcDzjAfdceibFZ/Q1yXQLZLoagYSs1NH4lSY8GhnGF4XhQ a73gWj3tSj4HpozTTp5NenbpKaAh7cOtL7IbUjAMy+cpwyYZS3O0HsJ69Brrj/j8BBdWPudClWkK ew6KOq//nMjTsTQrw/qEsxCHXfdqglnyQoSd1gOz2Jn+dGax69iju0GZZDBAu1X/9XtfFix/J8b7 Al4APrNq1FNY9TBc5lHE/MzW42/z/JixowodcBOXjMFNhMM2zDZhjaMtpXI3wTB2Q/TDUzowwBHo LHsB2lFb2lXfuj1LR2GVrB/D8BebViWgXr2YcpKBqWrHUnu5WRAcLRu9Heih8MklGrIILWQ4GnsC qTAoJciokkg0veKl72q+a8HuMjfukfCzMfFFtFe99jAlR9jy9OqczgtOphGeOwNd/qurYpjDhZoh ZZVHhZSoUdMk1kVNgtwsElglp+S0NTNCgV8SVQzYYHLCkoOzllnhxjm6Tx60hihmswftO2ywRR7a NTEUKMvK6dRln88QVQK6Trg5KwBYBWEvS8z7wHJe6g55NfbGmZg5cVhuJYnmD2CH1VY8L9awneqB CcTxuuY3F9TtFCY5dZfNAEQ/RBKeM+SbA/lcPKZTx9EqEhgUunGcFNUyVWheIOGypamGoKAcxmnG VOKFQy8vbNrYgHtFXGMaTRZTasYfBYkYRyXFxDRRh3Xm+1v5fRZ3Ooy2t2ISWo/51xOGNYdF8l8O yfwuKWB97PG3IvC8Li40+NPQABdzo2xD5X81XONpB352jvnz6Dln7vDCa7yPAh+Rd/HIMFwblpHl TdadRdj5KgDVAXF6+FwJujz2lOObqX/v742LIsdF2wZn9vM54bshRe7rKAFYYZkA4l6BMJpovnte C7ANyzvX/q6TWdiAbeui+FZT+0Hf+cVMXB3cqQepIo5HV6sGycKXI3ov+Z8xixsptpGf3OeQYyk4 Zd8h46Zseb1Y8HkW1Vrmij/3RpcPrmbSd56pbnixWH0+t4tVAVuLFdYyX2RI5UtvI9ivYBC4zh6L OAXiXScdSh6e4fe76I9vhSo2n17I42Ei3YZF4ufjaxrK0WD5wke2rVyIeny830CTChnjY6jwZJCY d3qblRXaTnrqVc/RdC6akXdfpxvENmUPczPGr/Fr+4+H20W2F7a+hceEQYzK5bih45HlstT/oq1Z UxYygpJ6qozGj7s3Jp8sl1/Va0RQJQEirvqMh+ThoD77Tl41exlc0i/Aqy1w3tbrJ6v8GiMfoGSe ONFB39Wno3Gs/sjFMo4z9dUlgdig1fRLDZXB43qffvGTlKA7MgCV6Hl9lDROyeH0INDRHW7wu783 at7z3Y8Fz94+HtIKXJztE3j3n7OLX9WbRGIa58Xo6xlKfIYNl50LZlbEzPGc+btWmeiN/36uwiPf af/XWrOJ1y4EyEXXM9QU2iuxJP1X0rYdQSb/KqqaA5gTDxuCzFPfc9njmcfXlPQqUIMvWEPKUMzo ETnE0nQ84/0AT1fhpUJAyJft+bDuM8TdZ6DMcwfvAdiFMS2EZ315LnRKUi5g88hnOIDu91o27mzh E6OLEkfoICBwvblDSEtSgxNzB7NLwU91D55PNtZLTc+1V/Ngd+PY8IP40YbT/ceMvNTCY0FPTdws y2IigIDiJTgYJjng/Pq9wX0qE1kvASvi359HiyX4PtZ/EYsjKbLxlYx/bY+LmqZ0W4vIflUxfafx MDg520riNV2zTHatd3L1b8aW+Mww15NLmkd89XYrbbbbNtwoJ7xB2v6371Hb90sFbWBFSfhAAC/E 79L0xvmw5de7UYcbr+ywyfAx5aTMN9r8s4UTyCBcr1Pkryv+mY8N6/zHcW0MmKP87rwuzkv7B/J3 TZ8aj/W3LofXjUQK8RDTF1ZheOGAToerS0owkX9Lab6UZR4Si4I49obPcaWWeanFG27Y9nNiXizH O+vQCJW5hJuv8uffi7szEwIO1P5vt4HrjZVcoRSIA32XUi6ouwQGhPW2Lkmvxplyo0Fx3pvYvk5Z J+DoWo3yQAA+jaoDHDCUC0ymLc2hKZIuXZlKAZ2wJ3t9F+ErxuCFTRKyQTXZ5D4y5C1CU7NChvmG G3XmPsD93ONYAMWRmbkU9Cr8jfPGYgjuR9QhUCaO7yzXewB15u6XuGktWdnLffZCE+p0Wjan1rQA UJSUZ+Ac9lBzI2CtTkKXDqCMs6OoFB4V4UM/zSylsp0FmEtZJvOk1SyZSVuFZJWnjqOCFAk3i6Th nagg3DW546+4WC7CCLaGsGdyBXIp5Zk9xGCvIn4MxR9B4r/Bu/OYVw/SE6fuv0upUOn+5cqNK6pp H3z5Sud5ncdlzIWvWBUTTBxVlmTAuYBgwNQDHPr4/pFYAc7X181zeXMnkCkbNLsa46ZBbGqaNt/I kqIeA7Ov7AiMGqmHfG6fsvSaCf9kmtRDn/eXtLW2GXRDSgZcgZMqXrbwTaRW+HZ6aW0z764UBbHq 6R/IQsvYIDgcjc4qdEqkDeVtp9J/McKiDT/tgctbbsKVGJ6QlcDIxzwn8oPztVxxBZYb/wSPuX4X zplfAC7YTh6EIkR6+fgUqmGmqjj2pHY1xEbQHUgdnkCsyUilpUyTrO8TcYRqTfgk4F3HQ2Z7XgCP gL5cD6tDNbejahAz/+BqEyC3upHysDM8w7ExCAprrca8SEHpDa48Ik3xmo0HvjZ+9Orkacute/k7 jC6OnbgnMAdLOYRmOkuZnp5RXct9mFROHcrwW8JxRmXISTMBkRssl6RxIemSumi0cUx9L+l8E6b2 TdI77h5mabTgr/MgSo03bsh2SsPDIo9KXRm6J2gIUeDW38eOJRIbtOBkyUJgvzTJbfEAwqN4YZsF +XShhvAbV2NP21KbA/0+Bt/EKPStoHeMsgaFKJhgQGq0vYu4Lk2sJrOeaJf5Bi+B8wMnN6B6uoUv ImRJbdhdtiYymf2fcAhO0cEcKEnkiwFam6eaI+80kSG3GLZRr5xV0GjqXoGcof7DiksMI+iCJ16K 14N3g8GSQWdVrPU1MDCuO7kTdgEy+EjtnNcjYvp8DIpDnGSJKImjdXSevrHKs62FtZ+ZknFs3AU3 NNBNd3FOSCT8Nx5pZMBXt3TW4UWuEO/KV2bB1C5Q0o++DzwtuihVfBYz3+c5ImNUiVzEh0OiHAB0 mywCXdAeT8O0hT17+/cxrqNWnXDshoBV60ObfBndrBRsYFOO7Ehq/eAOpz3C3E3ZwMGpgEGoIQoI ljKHKqwhcaGXki75sb+aeR+su/KI3i5S6ZgnhiwZLJyp2Ikl99rWQX4fBI1Fv2OIsVsy7fKTl1ei qIKjhyOFlI8Yb6mE1h1TmRIuFfpz2ewabjuhvbIpGH4OUq4yVT5B4rFjx+DO+1DVBnV+TXQOaSIy RPZP9AprnG72JSdZh1KvVxIhZUEf87bpEpUrhhijhgZE1IvUFG0EiLp8XPinK3BwepYR9kcydIha wCu3N9cCNpL3j8Vai4fDL+Fb7KzDFTbNS8eGlsZJ9fc+ovW9F+sxPwapvHZBX2SLZngviRxofrvq +6xiM6RVcaV/6UACUbPnTT6sv7lY8o0QJA7+k9ubFR7rbEWorPaFADbpKMhiIwYjfHypx99GM/fS gcrLJThQWAvyFccXFVQ5hl91BWDO05dz0n7oIgb8eGGsxhf/hI7mNIXcqIIVa/0IQhfzSucqFDNT zAzJRlxm4JmourSttkWMzN5ta2dEQpUxmFepafv8seX43CiqR/e5JPOdfp93xGv8srevAJ6XK1BK R1FtUlmd7jl3GwxlqITp8/qLJoNQvhKOQWh8H4KQ40WU8ELmvODqPk/3/jxm3qYkse/mmy9O5Oic LKM+9HYbQgGcr/mnr6ybh/hHQCu+S+GmYSCNLp3uazntDOoZkGjgC3PlNcZl60O7sS8tW7sPQUYW ovnNRe4ntcJYO0qcTXfVVmDQhrtSxVEzWa8P6ANNmnohWakNBVR2U4YX7htgdSoNsGOurTNXHCc5 mcwdzgxo7h13SyPZePpvYHvwNaQ8W1r9LiXjNpMGB2VW2MrgCRdQ0e9RFHgEkVTYDRFlytOxJagP pEptDvkM8b3S2ga2MrasKMbynfiKqGGAVJ8OTsz15/2OLLUIl12DrJuTsG0fwu0wOSY5FJDQY1GU xzyz89635nc/JDAQ6wLcftHHiGQluLwzOgckxbl0ydxDFqEUGvijLVKz9DGaehV/wX5JMhf9XwlO cUMUW61B3c9iF9jXc0D5DhPVV4P9FcpZ1sQID+LUZBruzHPy9ou6f61oyjjZtbD+aE8p39B7hq7P e/hCbWU2oqMD5QCYvBY65ZNL+PdRtvziWxHTTWkRMw/hLficNz2daWzpgebfY30Wm+CMNkwHwglO 1DW+HcNZ1IQp9WJ2tLcXPSn8VkNiKx62KU5a1jeVHqYwSpn1q/KxvinAcdO6nsjqcz93QGvjKgz3 podvVaT/rb5d/JqbOUjMmHqCtwb3sbbwR3CszeU3n/tjyIHgb3XcH90dZ895udfgyketKJNbUACy UakUEL2+OYjzyWhWrarSMxegQXpTZrKGhue3fI2oSgmSKJo5vjsSnEtAElffUDePTj6IrLOOjseL w3NahhhrW0IGhVNheNjuZdBWkLMFyMSIx4LRCJCNS9LMigAkYQV6iK/1d6Cla66KYE9qcxnO2fzh 4HMh3i2SG3UAwG17B1qQzbIGQlsvILvUm5uo2hdqFbAEm073PF9fhu5bAWU5Bo7qUzSSs/fGAGxt aCGl6OueKK/cQGJ9F1MiJvj4qt5piRO7s8kOE1xZ6d4qAP412XayHrJWRR/VsfLydSMOV/9wCeGD 2tChk5OV4kK2gb0YA2qtpRjD8SU8be+4aRwCUe0FB3AiVBFxFYSl5YUXrIY/Y/kDysys1GfbZJ0i qJxLD3n4phrx/NNy6VK4W+g0dce5yx3GV34Dhe17OqKSF1oic/BtsZ0grGdrlCy2wA/0joCiMAjP tzKEz7wKAf+H9TmhCW34femVukIl+v2k7nVYiUbKEHxH53VAxHibT5Jv8yyM1iadpNHvmJdLPEvv oVlsoSM5wwgFGimEhclAfVt9TGr4U370ti/+x+7S65mX6CxpRAjaLjKiEY2wfol4MtvIk1WdmzIs Tyna41mKEeggOzqe+S/uIcEbH20A9HHThbEhgcRtYdzg1Mv/Ms9wQ4Vt+2Xiq6NsxHC49+HmXWzZ HIHth1d2yR8PjB8bUeZ3bq2Vl6S22QO5Agy1jKtzX9Ze4VMNMDgb+PrXXoUl7xVq2JIL29nDtvTU evBJE26kI7qm46e44enO9+CqdnLf/PFW2eoRG84qUembQI49ghPoxqOJy4B1LW4/RU8I4G9zJ4Sv Cm9qqE4yZELLTm4APxye0h3xhuOVxnpxpFtyZ0kFcwATrCjaA0JMJNh4GMM0ATP0o6wgvBb6u/1T g3iRTLKd4gOXxZuiMXAPyCh34rCEXZwVHvHqVm1OUTNPv0ry9V5s1o+EMjWJKHqPCBBx7NWnvd/S jhnhFNwa9sFAQWe3A/2WZrTACO8d0VJ7vEY4Qjal6IEcLanSrzG1zrWmadI9o3UlvD1DIrE7VYt2 QyseTOSKclc3YX66K4NwN68HYAJUGfDzJpEUr5e1KHQy7UP42N8JkT7VR3z0/Dw5hTtfPOgWK4x9 WJh+hrP88/i3djNtmyYZ6hoDwmhFzq5Tf0PKucYRvLSwGsHtgWDAWTvBAc6cT1x0WhQ2vv1sQXBb UGTxwZ/nOPWroT1Z84ezDvyzoUscY/f2jFGoXGn53Rkio5ibZ63/7MuiPQmZmHMP88+707lQWi5h cx9yxQ2fXmhhbcE4/cK/99HnG4EaWU03YX8BNzJJFhBomQubKtUpycczyqnNuIogZU/OsxfNLp1t Q7OkSiLiPt3d7ERcSv7vIL9bRcduBBOWsYJ6iTDKLRXJlN+y/mEWmZoGf608ecoVs9ArrIFiJaQN ZmsXShWS0jZ6lJvzx5HYO54hHP3/rJPfSZIVSkT/HjIVJXA1oiUgFBntOPIuoXX4bgV1W0PsD0JY vRq06ZcxB/enfALeyY01OZlzzWQEpp1YxPxgJfAlHH3xeWbE82xBEunQlC1qBZBC2FGNS8JQmW4m HCkbOv9Enwub197EVmDOTX5cEDjkDS6bZkeyzBmExUH+eVOa0Bt8UccVTIpTgygyrhs1YnmlZEP6 antYE+pEOtXiODnVVwrSlRSPJPOKVrckC0o3UKUb51DR5kZ28AgiK/vPUw/s4qhvBapvT5kPBQ71 GEEPgE/QLLpZNTk6LQwgW4Nl67JIpswGhjQ38cZXqLQh6mYUnChWWpY5+Sxy8BMi+Eptl3LfV5R9 H085ovM+SCdiFh7PCAvHSRdgRIWWpcK0UwngZ306B6tDcEGOfa7JF3/DX0FcW/zW1yySVHYMBrmj PysknKHCPi68za8VGTBjNgKe9tTpJgZOwM4Z2IXxYFSwbYXjNy6T2NW2uvoui1a+hcl0m2094MKE m0ANOrjQwqPHqPx+JK4178BIdCwzHH0YSot4aXqHvnNhIoYSIZA/iEKaedBmdq8SUhOnYYpGtBxF ERyjDetOFAMLAFMavBBBTkhR4dYo9KO8QtGJmx/K63Okv9PajbmnkNt5ZRpuscHvfth7aTyVE/oZ UF+5+gklfypOYWBlYP8edU0x0Sddhj5QhsFV4kqcKdy76V4MfH7OWisN2rGvfU4jbNJYmTckpixB DzB+eQoIjI0+5Rg+dtXCAHONP81jyj3oMcngLtx3TP5ch7MemfyzO225hwjStlMQZo4rC8wI/ayi oyiqfVjWPRHGJ7DHaFQHBixYl8dRJ0glHEYDpUwdZCJrQAcvXO9IM9BlwXHSjqcUcMmSi/CUZCZ/ 4uIrniyF1pQLuJOw4//GGyQHZDet5I/hWt2HuNHTnyoJlH0z0xplgrLBDN8Hs/+2foT5pg4r+4ha ThESE2fy4Z12uYdvVNHM9YIdSffqdl/azhgtSlyd+ZKIiqQAA3TgVdQgTYkGt6O66PROYPeQPOQf 8MvOQJzgtGkLRMmRD8b+UURzJ2hhSZeuurzePUbHCIv2ffVvPbaTuILDd+qlRdYv/yksuKlWPJk1 emEqw7DqH9SEKTpRhioSBpC5zrxtrCslKw8zD7aupWL8ZMbIDOdjM1NODmIhjJTar+fDAZG8B9yk meh12tYlHO9auloiZYePYpskOXqq0lsqK4KnjirkNGu8+FYX0RnwoOz6OvOmgRKI7udSy9Drn4A/ qpIZC1qZGzTZzeRpsL/XXK5YKFArk4ntZuLy2iq+L2emuaobZkeJubRN8AQOa9ztk+DrlBYQd99p puovLzu29KUsUTSOh7RKPfw7FTCOz7WWJG7siFe0aqgoAEnnlqan6THcRHZOwNMIa9T4fdFtdKhH 4DisGGB0VylLSMJ2Jnar/4WhXMQutqVQhJhSx5d5phGjfC07DtYTOhfrMzRTEuZiFuGLuPWlbcs+ aYkH145dN7Hnf3PsMZSI9ThL6eQD4dTWGj+BDMrnHv2S8XnVBStd6jagBPkKeKmWqWCdGfw+iWyx akU2WBztD5LW02J2pLXfSuhvD6pzc/7L6zP4fnSfwA19dFsQEG8XfPe5YhFW5Ru9THuFJ04Wsrws /MUrXKz2gIQSEZpOPcTmTTWFZ/LTGQXYlLHTBMWwuxilqjb3QpymZiVh/C72NC6Uq1a/A/7RtwPL 8p1s+ffiQrOD9FitnejB9h33bXYk/SatOvn8lzC+lR+K1Ae0q7Wu/kBEM71g4h3NVUcLfMU8VEnH BLnhhELIww9XVKyWAAgxEpYI4inqnnPbPZUSTKO1usF3o8XdO95gUkBGABKQwkJLgmtJ4wTbxTh/ kK9l87ZisXRohkBmibaGNsDUJXa90tV533otP1YS1047MdmoiedEhfMZLckQfuJSuVphwazICuS+ osCfc5LdaVWeQQ776M3O3HTu7cWcOnK63rTBs9SttsitWH+20x76Cd1O/PGHm5uW5jTtcScnSwYc mWDY/E9zdPTuV2hutyysS4TsqG4utB5Z+V8L8i92U3wpZuSxPVXZCaFVvARTWV3f5XF/QuNzIfLv LEpW/ajQzHrULvRm1K+F3I8EawTuFicvbANxo3DVa1Jj+1Jt33EONbNQgPBWy0Z++DZE1lGtEW2v eJOF5G/weW6oTfaOVcjkMN/ufetftvAx0S/LXzMuIQu/4ynO3hDxaEz6GOHflybQ6maCC0bxrFn0 2C+wXQroX9ZNtFyrkjNNFLxx1D6XRhipZVygjAWIJBrAm/4jmtI090N/xH7cLMKCYDDI5eTMVnhE hpT13Bprp4q4i8A2U6Z5J/KXSkfjMCpJOZ9n0MH/mRT9uKJhoe/tvmO8VLQ5hHEv5B0JkWQQObsw 7+A3aYL4xZAkXzSasqMJEsF6jvYvRuqa31zw4JFzHdnz2XCzP4PHECO+Z3rucc9jWGkjir5xa0QC Wd6GoDJYUo67hDNjt4OrTI3UKM1nAXKlVc7dIwkBJVm0YnPIhaw0szu12vu42UXyxVsc50WHkHy9 lX16MpsXKZ70Kf43NhDQdOFSAPrh4BD53DsGvdTk5PX5/R6Fab8BJQvbTiCXnij5+JjNi/mxlFuj W1CUPcc/QmYvQIvCvP09KIaxZEi9fLc6hohHgMJ4JbuWd1VmEq1VAGvrJFPAukCOH6ba6cqew2Zu IUgTdKfQqkbc2muPU0vPAX2HyDK/gOhPnIb82uuAbPqMUJgyb8eVexnz1BwYpLFOkRPGvBiTPPaZ DDYCUCa0iw9mF69/xeSavuYDfXOol2z09tssHkgmT9hE0Bx6ttlBvyV2VwU2cCC0GgBGCxnVbCZ+ jyAbA5EYaF2XOOH0j8LzHTznYlCNFl9JMz1K85RfwtB1XqVUnWr1b0ggdgFlvLkNUI0AbbSw2lty +O3TnHNgvR1uFfSH7WqzbZ3UrPwqzoVC5bxXAq2PGPcK8l57Mzkhz1JqeWIArbcoR/6GuhToHkVc PwDEE/vYpoQhvzUQjBQHHXghNbQLzzgs8rAw4Cb+Spn35+BTlO2FNuGH8eAGuhifqIavzQSjrnHu pqJnM3fsmWBQlvVtCjl9YG6r86A6hEnvMAZNgG9m3ULK6fQTXV/TjdxdsxlKf0SzFE0DSahwYwTO i4iFgyh3EFSP5VZ0epOm0u5BiA86YzC8jbm2Q2ik9QA9C1frPoJeDgT6TMEC0eJZGszpgQ4t9F3k IeI+VNEq47ivXI3L3PIsXHz6ZFX/5p7Cs3+IHmJ/Ep/Lc8mcJTPzbiP8erkVp6r5gJ++UsVNyTU0 jg/O02EsDTZOqFDrVFzsCgj2peD4J5P37vdlg2zIXW833DvCqR+gj4H0cKzK1lHz8qIwsSkKwp2R Op3H0/R7MOOEORA+C423zseO10O7tDf1lvlGWTQM//6jgYftO0RKGEkpO10RmHwRzm5lk+R+q+Nm qt8QK1QoBWovrfS29yq7yazaQL4Z6Z+V/oxksJMg1ixyJQXMpxS7g1KqRMEn/HVQTBeq+d0xTG5k 0d9QfmGnpPp4rcd8A5zTXZ/bDEdfcRZ5AwEK3wmkyZtR306zOOGXcLJKhHG3M3SoN+n3KREc4D6r Ldvy67qiEJjTIlP8Y92q3wvxlfST51ss4v0f3zt8lewrChRi3Kg1vvaWZf7DpZ7rl2SchGXS53Xn nQtr6jFGTdBNl9bpv/Kjg+rIkrEO8Xxd9FABdXmEL4MDjcmKRL0PsY+ErCjUmxsEq5u9AYmGNn0m WfI4pfjsM2D4uqcPIt6X4Wt8QAoKihIjDarzD5gDiatz3crEIL0UXr8t8rHFZbHocCZ+RYZ6uudt T6OnHdSzrMul/+AHjNrRrGsdT1wAv9c7J6huwsogNzECjYAD6y2jPUznGb4mf0MrQswSiSwNfOBU KWAVL7UJHPnjBVe3PRVHVjapOiE7kd0QAIIE20zJItDNgFD/XMIK+3/rkg5gEDdvzwmBIWGiz/Ft +x89zvs18I1+xAcGYogi0bqTEq5ydLhcstZ9qUwmctrhv2IqfAPTFM/wZaxzxoVzdcF3lp5KG37O sjbRmPi/ayku6zO7B47mfmuUn+USQ9RRdHrj7l72CWzw1TfX8Njm/qrVV7b6R/n5BbVi0Yop480/ nTKIYJgUHldsPaoyezOJpvfbsPWIIgIAQ99k9KVXiNX51wR2FngCCIquPRs7MmjBxhDrduzC9hrq x/jlbeywzaxZ634NBak+c4tcYukzrDVUzpSbBop0OGLfZ6AKFMRyxaZchBw0E1DrEkXlrx9lfQXZ 3Rqz1w4a1LlfH116izpxAhjOWXxEBU112A1HzQaIsa+5HbKyxIhZcBH6vAfb0S9H+NjGSFRSk9ZA 8EBpz3iWNm6HYP9S2vXhZERL5IYcaMc+wr9bWvArY3xihXEB+UNgfgTiDi8gY+qcVHsZV+w2J5i0 nreuM8LvV0c57EJPMrkEJsa5WDJ2pmLaFmAv2mE4xL3b+ya6lmu988QKRDVu2bqpPzVMBn+dzqFg 7Ki2sm9URXig63zX2Z4WOT0SS6L+5nBeVIGwEk9vE2No5bwjCP7LdOt0hLz1hNcwdDGmclGJpQ2n jMtqsRY9B+qDwGmtD3WR9FrlcwFSjzQGL1ltpn4LBQec0/JDMjaDIrooBQoSYhVXuu5X6aIh/DK2 8aiJnbqKBYpW1/yo56shCF0R66z0v5hAtk6FO7MHopjUo/aHYZ3yqDT6zmeyepamRQbjnfxj7KD5 2QxWSojr6ekoJxLJ3bwyDGaZBOmmyAQ1ODamIA2+m+CCkWLv1zi1mdS672TO3wq1JLwXuVfAHlYg T0Fz7LdFKgfxZF9UasU2cZuCek3dg0wbY47r86pkxQCzIhsKKJ/WMGV5e5+C0wRFaTwfJnsvOaOm WE4tf451nqZSpRIcOCUBJH3o64EecEPNji38lOn5qbi1+p0mEDe8ede2INHF+sOUXzO56jntZX9y +maQzx8L11qWLT2H9cnuIHGubEgYWto/1n4fPRej9h8nJt/EC2UXIqm41YIgLU/fZl4NP2nkRJPT 7v7r4f3fR4up8GiLLm/+6zYC+KuyOwOEcQw4N7GQBTsQck42i1BFrtCqbfPlZTV+MG0MruEBstHT sCSmyKfzePB7TyL5fdzYuR+X3TQ+7iDFvtNR7DQedPxgHJMGPdsrhI5k72cjhThL+FlgHxktI+1S AHxZZ9U4pJBz1bDPLpMr7MgttF+6m8kmdr+IutzQ95m4UhoYT9nNWf7Flc1AAr0CmazUfk30eK7z TnWB6IGCILy7oS3MScNLC+lZNH83mtyzYV+ViCVu9bd2jtz6M70aWr0il+rOrCNgj4JlniPuLbB6 ul3VwxgdI2rikn5u0z0xdGmHtpJkmkHtjJwJgSWzFboB6IvGZ4JWVcR/Z76ILx6cDWVtq/TZmPk5 60ujhRuI8IkhyoACYNLNSrPcD9iOV23BLAc3xfiibIkTdSns+YwZBmwHJA0f5QrX+7G5QU2MyKi3 oX1T9oIgaNtSJcs28Y7nQFHC7wk8gUJBE+H6qPSfZKJbJI+5yfrb08Fa+j3RSwZxVe0C5Kirshye U/jET59snulm6yTFMBGNIVmqceZNBQ0bYx8icFToFejP2bhSOf7UwOAieRXQBRv1R/r0PU3qxfQa MsHsQ4/tYT69iSsRa1uISDhM9CONUAK8vBCjln5/wQ86zVq2Jt/CQR9l/oeUJ/CxZv8e285PMwrD BGAMyTWbSt1xUUaPAOHtX+GvHojftAea/VivYhIkoIpSYGm1ynnQljiDC3k87kWGFTXuwgYOTsNi Dg6kxAlKXsMIqjNTwK5Nz1U1e/j1qK2IYYYmuF2fbsLvMoEpIlkny8O0YGuo8u3ryoiAjYSDp3zH 1TOWcToX2OibziSC9oUflQwFreET4VIgWbf64GyN8xoXpoWlyiU0v7bgZH2vkrZqAakBvfB9KSuH Wu0kSiTMuaBHObZQEAQoC7ApFzZYaSuU/eifMqB45cryUT+HGVS/ZRgsz7xdxPZBEnSgdYeJ9ERZ TxhDE/qPlj8NqKaqbSo9qlPDvwaDS6wi/A/cc4R8Ee9/6zvSUJZjzFZ3fjaxcJkiIN3z3nbboziN QwUQIV6dAwJKdwUwttAo9H4OBgXQ/jByiBuh+UnfXMszygcH0inV5e/9GNNPG22bIiipD4TOqIdz eWKVIkgCxnC8j/cbMPypLn3+n0iiwxvr64v8J0Q94EFodh7XNen7WDFZUOKSZ9bSTvF1rzGfbJU6 ifBypzIMx6Ar7IDVsTFg0UAQ9L+ajl9DPLp3ZETj6OOQ3QMM+65KTgYjgQarhRuJKAy4Kp4Y30Qj b4soxJfRFHgS2u2cF76C8wV3I4yBBfA7A1vTh7ZhEBfp1ZGsDdoeQM4FCN7kHKJK0HTLjTlVRQKE A4clkD/IBdb7SlmzNI4RBIKzuQT3cxuw3Xu7bZ/c6yvXOzZ84DsXEv6FUN9racaiEmjciFce/5a5 3rxufZsYrSrqk+P1+RirenuP19ixmrnh0GTa7PIute9P3QN/J2hdy/ClZFn8O2ihrKhfY23olC0g pEXnrZspyzi9bVDMTUKnuiSh3IgWBjipN7ydZjDH2VcCXmrxe0RSolMU0Up+x/KA4EygeVNz0uo9 xMn5QW9EQvkWWyW0FaDOmWo1Ya9DLAWy3umCjkAG/DAPyqdE2eDRutlGH5j+CX07hErHnESzSPnr fiV73KmNhHErhDblN7oFFu4aVxJtFWLeBXp8o3nGD1IqqsWLDUfOiPVn247cTyxXuEeBuHahkgpM Cd9biYFEsZNUSByjxDPUnYcQ8vyOKEhZubhKEkJKAoluO3N5C3mDijcp3MDmw3hD9WCfPS5h0hn1 hQDR8a8il319bQdFRtHRNRveJRM4/U0owQ/9W/IftnSuys15I1IB4r3Hq9/vuIDvBvfPNUF/Bsi6 JcfBfOujXcggIiByI+PB5vRpfgMf7N5nbiG+ybbAmt0PyEpeyGnmnTgND9INcCx1g8gEsAfOG06f JumjjI3+8R+M5MODSCjahFQwajHlsgW38aMfdsehOYlyIKlQ1VQ3l9ZGk77ZBOODTAl7RNAkDBIA PHDPAxhSua5S+f9v/qxH9SR/O0EmlcGlMliYx75Y6LopA+QDOcgKrNSzZdP6k+2op5OQZwdwNvY0 W2YnqQj3F5rLfJ+iOA3wUX+/yY6rb/HPIq4GYE3C0dXPB14QkT/EzmMApGXthb+3N5/7d0D/3yPB BIx42MftUSfwc9TgIODivhWeNEBEh1MS11vuyzznNejHfiUXM+YDa+qQg++DD3wNXpaIJndpYRkL SCGw9HhnS10DKAZ3oEhcLnqvsanEG44XnFPCJYYGMwEJzVTZRsDrTMEMFrqCScolkHSS4uSKzH/G h5cs1z4bCH5LBTuScTgDzJ+nWE2YeqHDRSmcSWz3Dh38icpw9YL+TfOzOR4vVRY93WHuX24za4Sl RkEILxCXL/6rUeP5oRVxnMgzHbbsiy/gYk+fBsHfEziBk31zGST50ptUFENqTmVe2DtHKb9WkAld nmYYY0FwHOqgHQm1APgBYng4ocx7X08xS4zKolbBgI4oCsrhHSE//mMKGsvVbahDLFXxiUBqjXCV 30getKFTLTxMjU9UjIsC2dk52dnPyfbG+756SJKw4Z07YY6GWhSt5oV9rroFsyZM8QgzUCqVyRtX 3jprUcCuAmv7UpN+u1hmouI4/pl14zQ6GL+zfBQgKmJsFZLdPifLsp4qrQ/h7CuI5/ydVeeY/zP+ ltCvzI0xj6Nu00aLY484e0R9oXCZ3Vws0RT5qHk9RANwkQ6Tt4cOMaRYg9bC/XwJD/oU+6CC1OS1 84k6V1rGhzNlYrD/6mnfXcCdcBJcik394M7AGNBNW8C5tsnPyYhUyvy+2MkLUT1S8Y4E50fIwhgL FOAkLEr4VS3RfIEvKR7Qb8a2N7GXyZkCTXAhOUO1BwUR4YEJy/slZLXHTLJhls7C5NdBR2joBL/7 3/Y4x7SFoJwE5Ikfxulno80HnYJxi5uEwUNFaRPvSxSEcc/3L7HIDX99x4p1DyRgpmG//2LJwFcp S+RPTI0viZS0AURWlTT7XBOklRJqxQXAdEkvVcilfWAA6BNYHmujIqG8+VDrqvfZpEfcD8US5uQA XtMssVjQ5a6aHVygs49ehNuHB/9K1RTv5BkQcJQujQuJaoZV9BeWQPrd8AF56au7LNwlolVX9DqV brl0aJV30BRkcNWWIBdM0bik2LTh0xVsPVSuxRFTZWqrKigYo2rW8mNoRGEQuWX93veGORhsJL4F VlXrUC1XPAAGe3wyaBvtt1RzPaSp9+ZwKgMn8npo0XSVeDRth+Peu7pzZx+TnnCA1xz5hV+HDBMo r93wTf688LPQzWpvq0fTt8RduWYLJpHNdg9bxzYw3YnsdCJL05N29oUiOWlGVt7h2/DAF7XapZhM rRe0t3Rdtl2lsXaTCz+4sQh+X8vCjPa/akueTOK8NmiakyN3y6evNBCAgKINpC9sbybE23xcLsOj 3SRPvoXxcDF0FbJsSzYqTsnIW/oTh3hOinIT4YtkWjvOjVXH9tcJXAyra3Aaywftv7hqceFJ2hId TtuhmrDMielcciBN7XAn/GI1RwYpib6jhh8QCrCTATQO2DhvAdBlAXpLDjlxld1LQi9H77mfqH5x Co6119gafSoOHjvUb5mqDQplbmRzdHJlYW0NZW5kb2JqDTM5NSAwIG9iajw8L0xlbmd0aCAyNDEx My9GaWx0ZXIvRmxhdGVEZWNvZGUvTGVuZ3RoMSA0NDMwND4+c3RyZWFtDQrQL+bacE90Z1Of7ABG OtGlzGn9xyi61AzTiymiqCNK1XN5xCycvIouF25a22NeLgHsKtl44hNNiDqOCInHseve4CIT44JF bGcFhpQ1rE4bCPSQZXh/JIHbWQDk/gHzX94boxpSzfd45quqmcdBafVfU6odvZycysGoMsVgmNFS oUIfKJaHhEglLVE0/8EfKB23SlMMItSlbTBXc5AiuhVRwF9+1Pa5KQP3muQecjm4p57MgqkMDVlf 1p9zqVkbDm1effcRUCChC17MnY05RyMON7nbyj7J7ZMwfaS/hybPCdbsyhBO2X0+vAeCBpQYkB2b JjwYVBDywIdOcyU4nMBeNLgOJudFTS0otsTSsAYXuv0jxiGQurxw22CnNaYZ2ouQGxqGXIRwT5Ov LgQR50982jAq86Tmd9kiq5AeNkUdS9KcZE0a9wyqvDh81SmwSxLZIS053q4teMeL89nPMCd1gf// ghVHEstd+eNsiMVbi9Drvl38coH1p7g30JPc4/vwy4w2avGHPi21JDofrNuDmmXac+6Y4mX6iHGK zFE8gKc5cJRrIOtmmZJ8ltmqMVO4fe9TKQvekO5p93WjiHvzQrxJPLvt9ODt0nRd/Fb9Wf/B+MQH 5UxpAFFV07j3En85+GrySbtWlblU1FpuHbSjpZ1CEBUy3+38LVMd6RVIP0GlqZnYjWgZX4TupT3Q KQ8LEb+a/etWx9uD+tpSkYmqI1rrrobSBrv9erfXT8QKSz2wxwT9brPdxJt6KLdtrEOvxl7AuEwK R9Wzi9izW/E/zBHveoaLuarg8h7SQRf92E6Ev/3n2kbyJI8UysZLLs/xVGtxL9aZAU9jivzijOG1 mKfjCm5g3kdQterGTT5u1Bu+OFqiG3h6Ylqu+Tg/kn9PdFgGy4Ax+xCXlOmdEJR992GTgWVsSQrk CU5ugkikPzgp1mPcrhUCljRYnx5u9702f3GNojhBjzaIELPwx6v8sQljMWpXNHX8UHE6X3oRrmiM MBN/bg8O1ceYoRsLSnj6riiW5ULCmsh5CFBZiets3W2Q35ndTzhbSgmt3Gjyi2h1FmZRyC4KsrKy 9jQduHFgpBCNtAVDXCw2cRewM6pzQU7LVZL2T12CMDjW4TjeJ945O32VMSBC+Hz6lUejqZMVp01R gwkwMLLzF9CVbdUfbbf+s8OrNeMKWMcJuNBPbw5HzjnfaqYbASgh3hrUD8HMpQgINrgjbxns7uk0 7dq9koQ7YNRkSgVrKML8HXPUQE4BOl/mMJhKYN4Zqeld+SKbjJZfS0zHdcdbX3tTHEFDvtp9rqZQ PbKHS2oOU6nIkWC+4eWLto+nSXlCOqGE2GrY15CGwzZ2okHsCLbKXCJ1DqTmsHBsLMBNQv8TUN4i PL2YsR2B81WbzQUzQv0k46/RNZgOp46ZpfQXXiMC5FHVhoVOP02kjsM2xAUS34YxTofvjHp0wJEf C68h+MlLa+wFetNpg9k4fa1YcYlH+brBnbdnIwgrID+Ub8EmSsRzR5eArIs0OAZIhZy7GQK+srg2 sTezGVprV6YB8m76SlUkwa5kC/wsocOTfdIyJ+5dZes1pCh1LHmR7RFLK7Oaw1A7KEqzYvOovd0t nwMHPzs17xnNW3Az2I9C0EAan9yNxS83q4CWllTTEkTsYREgp8l1VR7e73Unxb4zrd5lLTf/nFFc mP2OQ4EZE+xb0rbsl6Jr57K4VgTImEDK8Ml218dNZGhUkfWDTH3z4L/36n8RvcHsMyPYF4dzfwH5 MH9zZHlF2WPFYul4YuDqY8H4FC1J/5FUG0veQGbabPWpRnRde+o8adZAzoTnQ1Sc4ABNdRncf+3R Ks3xOIjkhojwu+Q+GNvio9ZnpgrjyRczakTyHMpd8bNFK0WnEASvGVTeDuXVcFY+5SLleIwUFzeT Go9l9eqY6m9ZvQr5Ysq+5mEaRQLnLlHxeUb3DO2bP4If15Dqg4j0tTQan9Y/+4hGWwOBz6QDVOxH Bhs9sv1TMI5I7wpBhmEpkQeXztoW+P1Exgyv+GMyHaAZx8//c6UGtyS43Gj/NlH+ZhM+7UgKzuQr NZ67+jTvlxJ7yyp8k327JquAdbYeRT+vgIrFZ1al6C/Aq/I66qvxa9gbXrI1zy3U8y1OGxZlC8m2 zpr018XU0FwQfAjmNOjU5uHy3H89W5R0dlbazSxPNXraWCjtUq+wdsWd/keVkIpsaSqp04ara3B5 Cjbqs20woIaNPVXPiOLPAZduBphznwwqb3MHkyu3xFwpxV7JKJSlP4VpkDG7Z4RBrMGNYwDJFrA3 bjl7om4XLCHspGFpGDvzoj71fseWCbFqQKGCsp3I/qkk3BsMoz7uE+MnB6Rz307Roeq6oS/mnN9q XZl/wulvdjSS+JErBU9yVcBuFvX74PyEbU9hX5G8wHj0Xmn4HO8705JyQv3ZZxE+krkJR4jcgbdn P6PUh13LI8E+5clUehP1iB/SOzv1ZioJY8s0+UMpAiD6arJSGBgrG49UuPwBDuGEImd1EIDvQR8W hiDz5ZaYiVoIHXzbhUJp29l0IqzJH2I7jnMkVuXmjHHlbwNj4af8+Ch6EKHDfG8vdzubs9AAZJv7 jbgLeRTfyPLTqeqoy9CisrG0qHQ6Gw/NlUxuARy+pgW6pDDQbHRzcIHuFdoFJqyQO9pT9C3Y6dB+ e9WUAuR374zTcBXqp1EBwz0ODYbgO6Y6wzA78bmyxXkB/XQFxEW0NmMXVKMRTYi/9nDQ5+oZd3Rn WLQfVSUhj6zRrrADaxWVRAKF5efIIj8b1eO/teNvsuNPErZrtb+wfEvkQRN0MrR64Z+pOFzjj8mG BE42WFCZUB6ZV0u1wHONy6FulEbgNZca1oKZ5Vm7p7FaQb5c5izIzhAUszSeJCxProwZl3jsKVKo GCAC3qetyxwmBdSWNPwj5/sHxtHqslmAdnFXCMAEhB9YLUEAKtuF+S1YPmqtVSBD/MIO6d+d2o12 p+rHAYKadRMQ677DLYkXpfSYWxiOC70aSaUa5MSfeA/LNOecOxjfsw6N+FM9++gPVfopPA889SVT UdPjRuF33OMBJsScSpUj7qnNfssgp/bd9wyaeNL1IPoJew6Cshp0H0I6RZIica15MXiNreolwIsK QJ5PPwEd2T8ScrqQrfbXRTzi9j0JPMg4rcTpgqsHsf+HbTpJGmqYezkdD1Hn9hJWQ+M7+214Gr/T B19lF8WhXhPgjUpjMzJS9xJgx0znZKDo3VCLRyUYFgJzRkYeZAouPg1DFDDXFxw4bfG7ljUTMBVV MIoVl8P8spHiyPF3q6skz2tkvmMxbBY7oKg30CPlt95fVVeECJ5P+cj1OCXzMofD/tCszI0e5I8/ NnKsPu+px0vHZzAlII/WM+GbIuGw+SFSywzi0j/Y7cVQMSGywzs58W2hav6BAV7wUQM8CNcDAEOR TZ3AnB81TGjBgj//VEh7PMnx8UnFm+02qUeE4c+axdBJbYdK50Ho61hXiKm5X70jgIfPp5eCyG1U cJDRX0Hm1uxztyeLmNThxD37X9sZZGcRUAekqRFdG0aFJ73ieAUPAUTTBZLLe1Xcu+/comBkJ119 X9Pqr+MlNeJfAIlpG3yOdZhTZiwUFGaWavuIOLF+8zZmmmE0LM5EY3hCzcCor8jBFXkp5oDXIOs7 G2lWW/+EY0ZLlpvDH1RR56Q4UsT0b4cTteoFEnC8dPXCYxEQowCvCfQl3AZBbExanlFXx5RUH55j 80c2URUJsbGizMSeTNnYkbrUdaiu18hMQz+foGFiztMLVnrY7bTRaUL3sGARigdcXml3JAxJd+EK uUjigqZQu3Z4PNsr9I4Mbi4tbYDhEEWk7eraXAR6fGAlZcLopWCk8TorGV0vOjSaQivjVBeHXg3w 5nnkJAmpXDbGnZBsN+tXF9aJ9SG7RcpLVYXMWi3vlJYFrZNY6ISLaXZktorEVzQO+Rq+E9QYxM5V kji/7qOsw2zDfY7BvBeQjb2x+1yySA6PiGAxp3wijZWKldzpPgBCDqoB2t4FJk81vX71hAInFVVT 63BzI4HacqFSHZIpN2HktPBNTWGmfjWOBfjCFSUiOz2X+nZFyh3R1aBNCnKUIp9of3yc5N04KckP FloUaJewRWbVzEZgMow3btH5vQzjeq3Ac90vDaea0ToJVm/oJPiZ1NQS4huK+JP/oVk1MXj0gwp4 fhx6/4nqeAATRMGB/Z2ui5RHoWVwMLoGI8lgQ3CUO7nqYh366IXkL5O67NCoBuanBuFilWhuTpPo 2/gRnPZ7zc4vXW6NgcC2x+jaqu6a5KePrDzHdOGmibRP6tPiJNimNVI8oDvYICVWKM1fN9hxzdFB gAhkf/wEW5bavjc6fQyYEdsoQJdCE9jl9IlYB5lq/cqFtaKgHBpO8jr/fknrjrvphpL/hwHU7ZCk bnA7OCorFICOhWVVSewakttWTNDUJFujQeJs6UFEvoNOEGaZZFBoh8wI0+547KmMasd+Q77Vl6dP 1QdzS3fhKhmALKb10ySwxSWbBl56UWlRfj6SK50KqF2YlJsFXhxVCiLZZW2Ke1Vy5sGUhCUsCwsq 9iTmp/Wj67UcBgzfSilYK4/PobjfPwESKShN1YUNoZj9vyFXtU4I/iJNgMqHlY2czCC/EVbYj8Vy c7wQkJogkB2xittHWWGM8W3qcnc02Jt6UssmoidQvlEIJb3ngo14pEkFsRQM/EETqJR5lnqRthR8 JxvZ7KJIeg4f/uTygpyyhe1akBWbd4nNHGpuGpFAlMcjOO63R5lN1Lio0WGRuhdQH2AehK5AUcMV wmK+GmvCUpetICdkCf9NuhDLg58M84hziKUqT3AeHld4P8SNpkkw98ruVpmZd4MAepl7a15EsOta ZL2g41qP9Bp9jcKqFPb/coNK5zM6+rZ4Asyyp7z06FH9eNoSbXKrog9TsC6Q+LpiCrzBjaCzKixi jqpomYjDGuGSQlMtYKbeXv2Wdg7lvGVmEt8SNVOQlaTONTDpgSuVZca3TaJJh+ZBtqO0ZDuCXphs SR3L+8PZOM+u0KEgHI/5vaukhOCTGLNxuPlEomhOYY3uY810jhM9K09kTkWbFOikZFqUNpmg2UpA zEnoI1sJGKOquHAVYDoxvocRId28XUeAMtmqK5ZpgWEqEJ3U7b4H+u8J8DAMNM8XZ6XoqTb2JNa1 4oJYSOLoKjRvgt4tny+e97box2ZBmCLPSnB9aW2U+lUGzNLRsakNddZq6563jHxa2UNDWjo3kxQo tNw497y/3xKHjDYszooariOeC3Zdu4Yf4gNE18YsGZ5fvB6swv+fmaYaUXHATy4KHTY6yeNb87ev yDhz+VEHVx4TQA8I5YmDjUMeJhzx5uWIkuz8cYc7nQ/b9W3QXM65GeoCOCuuVPVVPTTvy8HFO9Iv 3R3UbQye+sLd9Oj9GK024uDKGY8duGQo0wAtdhELxIIo4Kpo722koiUByyJ0rAxV9/d85vrmpRfc mtClS4tFc1MH5SSWmBDIeD1ee4Q/+SnAHTBxkoB7f74cZAC1q7DnwVygcrntRFsLI+MCJoYEuO9j CHJ2tRFESoa5vs0XgLEBx3gVaRjTGp+NRxaCUoeuQZmgaHs4KL7MVKJTY6cVkCWXs+TA/waawlZh NHgsOFMJJBcatjrVgPYSZjCCsLWCqyebZ7E9fXkaIhAiiuhITCl63ivE2E1qTzlRVtC31DP4bpET dLtH3SRFzDEQnuIuMfOCOFmcleInQaw4EZtL5fFDghIaVkPR5dErS+jNj/HxyQwnTUPSC4G/pR3H s1zRvn3BCB4azgYa/tCDwOgOtMhKmgDgUaKlQGEHTSh93Lp/DZIEebRsjo50dvCNsq7SOjMr496s rJHRJVnR6qAKIauop7oLfCyNUWdT8NpezGczcjiqItFzzzFb9N2BaOV8bXj5pE2r5dArJSnlJ/b4 MZFw29Yn5Snd/gk1sPTFUFuL6+PEMo5w1DcaHQyixdciGgJPcZEv8/IaH2LI4F/WZP1Sxgs1l56a cWhXXe8r58obGBTy/wQAM30+9eA0BrSxaIcCdTDMDBDVOuuQ9n91X99Cg8OxqSxF663nQQUqOUE7 YpvYtKeLruGMiTAQAXeBWe5rj1TP/RVOPLMeas0O9ttA2FkVuuFQyKCLuzs701ox6PRYvHJ4e8BE wNDFq8eI3ej4h6Cf9s+a8T+Y9PDZp85CmI+pxcacAL/22A4ZmjNYuOx8xDp80T/U1qTrTuaDEs87 A/PZ3y0/ZhJzyUgVVbIbloQTjKDXNNck18suSBHfoa7/CBqTACXTphKtnJRBknsGTOajRHu/kxkI yj6l250FcNTDy6GsSH0943//l45AaGB9oVZ8O9Mx645tYNeClEYHXghu1fVfWgFGTdexjEgkYYYF ETYcBslLq7V5OMd0X7gLgNsSpmvnLOvbkC8FayOm3Ucf8vBDRyH9BBdFQSiJ8HDfoEtl9jc0ZW2j fXiN+Qwx/rwB1J4s8Dnry834BvKEOMYs1opEZR3RoSLVbYKiDNVeJhDL0Wx+5qZtp1gr0EaaFryo vnTTPc1SGILCNSCcIB1wdyjxyxC2ergwPdJALYJjaGRTUT/MiA17hUZL84t5qaT7uVX4QRP0MLGU hl9juJjDvoRJLRhVY4+VlxoTo/35OaryAGdp9L7z43ds//DZhkd1xNQ04+D8JsX94OXez9uCr8QW mHzA2vFIQMZN3paunAut3UieyDQEqETtwQaBdvify1oaGw+KmMDcUuV4JUowLYuiMNER/3y48rVd Nv5JDa6Musn5RWimVvejUUR4N8Qj2UK392qFkfvLuCTBwcGKPl0ufaqI8ZRPFYXDF2Zgf4rKtdmH jriO+Oa4Ims3QBFvrtBxBSj1rzIQqgr0yErdWThEYh/91Fl5o8lfCWhy6ktqix5lVNPKDO8i/YFU +flDvD7KdQRDJ3gzW6RdSA9COKYNb3H9MNdt4W02jBP3SHbTD0Z+h08vVSEhCGJ1kwr/Lxm6eTDf F/8s9yngAzz5c7CleJK131Nts2HeWCcgPTPzy1FCbgnwuqo+PaLeDk0QGovW6Y31IHz0rvFo3RIt hjC0zIMqaKeUra0qC3v0oIP2ZmeyFeTZXRQwKcqqvpSHUt9qz/ssmkZiN9Zj63+Khk2PdMrT9+Ha hlz1r352bDmliWg2sgUA3H0XXHZ7DxG3XCuBIaLEUjH8yWZV4QqkPLdb2qu8bXYwFwFJlfyuLN3s vMzTB+B8UDY2U6qInHwHQifrvxm0BPJQPr8fCwDx/q7yz9VZi5ETXsmPLt+0LEukAYVUsOGO/I3j cky1oEW6Ae3oZSE9uON1sQLwz/IZwKXCkYqljQfj/LUHKC5bQLe9eZYHxXJ9ykGM+pOSNTTGdJhZ cgVBlu0okvJ99JWhdfbcNAGoMRcTeIy3g7UkRJmDUnFsVPCI5ILmzsRzcKeivqS69C+u6KmPVOgs 3xEVWk8dgr54RZKUABCI7/h7+gLKRE6pGpC1KH0nPjHLhJcK9qY/VGnS7KaDfuq470jCRbj3SW/B FMZPwMp51qsmBVLIq+/40qqNoNPItyxEvxgtjn6ZmeU4/crRixNwTQZKxT2YFohJao1vWdDUgfvJ t/lodloX2Izd8WlrEHPtUcZ/7hn59jahSZtzY4Pvr/eMhso6Zd1RsqP37dHyYjqKCp1ux8d75Cba CACSXevfKW9qy95dH22Sta+yJV45pqtOowggw68MksYHjDnJezFIqnoUJ/obZl5vT1gV416hswL5 yeTh8E9ojUw4AhooyMlEmUeH8+P9PmQ4LOkST/Jm4pIpGKDJ8ZQzI/v26dzRgBdlY5+21co0oeCS MfFZL35ZPdfrgFT53w9RETXC25R/q0xkfUAbvPNCLGRwmO8Hpa7zxss8bpRLPcNUB439/Cjx+yBA GXkNX2pTaqQ2s/CJENo5N8z+WThKEFNQ6m9524tfNMRnn3AU+k9hKiTvvysFujCwcjeQHUSTPYFL EwvgSaCeg+3SMcM4tnT3KE9XScq+jDY95NfJEjAqZedj5t5DikOSOjzZNIlMGSEXrrRH5KjFL4b6 +lfHJMuwxo52a1jAuEvSU9Re1qVcDi4ZV+AdQBTvjtrXvfzvziIRHLNIh6Q91oTRdB5oiM2IcVYr Dm27e8FfWkc0+XFf0XFHRCE6ilaeEPIOcf/ur4VY81BGIhObZqaz09QdLdiwcCGwKmqjq6pzPF/h OBzAVFVSTg+igDWrCca/7P2bpYjE5A4hnEtj7ulG0JU4Wm6fb+pnDdVczRIrUzMwa7MkSEXzftCO /6Yjr9/XAmUfUExT83rWl7FmYCStUGndljhO+lmT/LhgqRwH2WNql6rh+ZLIYn6nmhNHN1F6OFLD WkxoPzcxIaarAniPpK+fm0Xra2ssQdiU25vp1Tnd59ygMmr31fIAK/hpv/r0djA5rsCqRJ7SJIoR Ny3+ZATR9qdt4kQWdQWbId3qRGsnpPnZZcSeflLLbD9vBHBnbN87xlxAf6DZNPbH8P1bPo8qk3DX WoRaqyk0rQdfadA3GOwXlJKUAEJ2SeprivgytQ/gV0732oBG82+tChaoCWji/xOKSf0e9SsJHokB lNCr3+LokR8VIuGfe7MKwbFuL3CLnhFIY2bsMmEiIH3Q7l4SNb7yyP0ZQj4uR7VYFGJDBMXsANr3 io6td7m5H14AH/O51z1lw5ulStODqzWvaH3BtUgiN/TjqPIMzQrnR8r/OhjFsmWk+JXlOHvfYdG6 Z6rgQmOmqkAkkJw/GaPl/7V5g0CYRfaJOL6T7weuA0KB15tDJNpgMyM/LEgfyIhXxUNEIvZteSzM eRCu2O5RbhxkZpCWJnPNJTZB4W8nhnUsQ9oOWthZ/CkOw4qjL/SOMTitk89tlaIjLaiYI0J7bJrM RigMFkMrBYv7Vt5q7a2zU+D6UyYZ7xSLVA3Z/LFC5ZGVy6IVt2pzqp8Njobj4yv6u/x/nC4hkNpe joc2KLa3rQSjStKb6HRGdRkr7o/AsfMepncUEjUerPxBNea9b5jIWEk3wwRGlIj2yz60ly+AwAo0 V1CkDrMXpABvyC6BUEshYJqvwYzXcqEF4Pb4AgIjAtnajTfbx+2rjwhNu/QhTGELwIExot5x3mP0 Cr5FIpPQSxDVES3Gb8caU7OXY+0kovyR45Qy4kWS1/KarqlHn/cUsd4PVCU+Juo9b1Ld1d1rfqWP GAK6nkoxZHtaBzoRBtLNHxSQrnhy/etgtcbqrwIIT2neIq1vWgUxHJgtlLIa0MCg6aWLIb7zuZDP 9rXJXrYicFjK9iKgodOX7gGJiewsyTs35/i1a55wVYaEwtLoEZuE+d6yn96ZVm9nbJC3TXQBWQAW 9CXHpAb44oCAp7rnzK7ZmNIurYjbZ+1pKEtBJGHaDAxUJ+3jx6KoTO0Gu8N/R8XShIZjyrhci1s4 RbVHimYv9Amy7XcTASzSxVqOw6rm03uwqDmFkey2jsC0qzLo0SlFp+I3NooEeIS2TtnymUIc6hM2 cmA93U3U/NhXrKERtIWY16/6zU1P3BOGJbGfIxBwqG0hETTGrpKN9wLFy6eYFNNxO0vyS5U+O1AZ DRvDsqP5RdNVWON7ILi7dz+HyKkH8KLbthKU0sR+5AEf16MeFp/BGTo7SAwq1A8cmv0VY+5CaA+K GbT0jtBjCW+MKAREaEoV28yRqd3J/6EU1wmjZXq2SThZrBhw4eaw07ZC0hCNpwSTTWg9dru7r/g3 0cL24uCc1M0eVqwLZ6AGMLw+xU75lxiN+UMCAvKiI21xo4OoqLUA8I62/jCuybWrcYfwJw0RmWXU V6375qgYb1BIFMEosAISV3u3Bv1yukO9zh9cdZINa3fom2Cfe5cwSkEGU45umGj4oiX/9UsPN6+8 dVOeVLVTN+SbhvqYeWdG1xW7Mq28a71Wb5doB/phmIKByFTBUkCo6S45TYTsKL4qPaStLKbVrL7U ZEfZ50ooCI0gVaM0Tv2ZbNc+5zRH9Pkqc2CXdN6ZCm5hgUjx0gGjmm3RVucr1MlKSvqsnelgGPIj GXHGTgdcVtLyLbsqp4WiM1kG9mzvlLcSK20XMLa8Oyrv3rXhGqDntkY9oUmMdMTDhWuaZ2BZWh7k tzRyu28FSQSSvkccrJ3wh/WwRy/lBYU6D0jb5DmDRjz3bv5GPSLah+M+QltK7vK/104B/bCjQP3k 6Rs2ILbCFTGKy2xOgr7kUzKtblfvT0/PTVFgjTmglKg3a1b8COfkeUm35HUTQidBktC9WgdiSbx9 EVzfMf67lcHn62I3PgeunNDBm7gQUUsc3eRtovShZY8m8wPZ65+gHWrB2fLC1BeoTDBKqGUu1MSo Xprk6qqEB4EVFTbwzN2ErcDrako1H/yGBqxvP8/GpjVEKwgVmUY16r1WmxaAUFbye+ujzghcZguw VmGILZiu30emfJh/CnHbYwXTBXEMyXqk5YnT2+4wkZYcDfcoqofCH0wtDEAPaZ5fO5j+DhV21ULc OVi3SX4Z33+AcjfjeGTHL+ZNAJhybx2+hNLIDDPEj3J9fO2oyHvnGcHBOqKA6Yw5m310nVgpR6K9 RwFDPuF2Bs4CbNX7Up8up5byJ1wz1jIGQhJQ4/XFGflFt0K2MDj9EziagUO3d+F1A0NSsJELd8w/ 2c0cC8x6sa9SqB9nhpsHNAnHhBAa60iAFSKkeKSBxW1zMwpBqwvTAVPRTEwWUd7H/bnpaxd5EEZK 2DnYgS8r5Q24BCT6cDevTjFPWNaVYpjljl9cN1w9AtLNpvDvxqFE//U1b1SRFX3DjNGd1CkLT070 oiZzUazf2sQFLk0nFGoQBRkocCEmbFBhx0BRHQwLhZKfMcIBbmiUMo0oM0LfEoU7GPIGcCgzq56f //fDs00uhQI+kc84Q+jXj2MIXNv3mJaYtBgfzMaARBSdY2R72TsHWwcqCCvVnwAAfyWD7cXDR85C DBAOxthJ7/g8CWkGyIkG3+TDBJwRx/rZhKW25/uCNKKUAY1WeEYJLKNCCB3OSy4BvE8YPyPar5jT fN85FEFOKzlKE3cnFVRfBaa0Uu7FVPy8c4ANyFr1RK5HunWZURgsm4v6rRsVjOC1Nzu+86z13mIU PCs+QX/xBBPddQ7hboV6I1HluU9OMRFqh5Z8jYc08DiNUpSehFaByIRO5Su5Z4CaVpYtpC1k26ye oP773Vqs4gRMgmm/9oiVciA3ez0lYKEiaqix4t4L+3kNnbT3yxH9VKzgOKP6rs0jrhIS7gZJ6AeP nHDwNZLNYtF+1a/AZUELExKAWsQe1Kh+CXoCbAOmP6XaBspEVOT7j0YcAajHvcZ43DMpU2uHfMfg AIe52uLd8P52Z+bAN1Nxl/Uj+78535hn3S0HmINJup+SaYohE5GnyIeALdtifJayow2OB5UtHQqK gER13Uh2ubzL4Oa6MT489MfRgkSYtMzolerNGJlqhxrLGHRlPS+yhnypIGBkxOaIOSdh4D/VD1OF 180SGu2XflzrDgK3GzxpT6c8k6ue2TNsffb8KHhWp5tQfqkRyC70BPrPPwZO3QNO053Zbgsz6tpH HToqRcOsBk+ccPQYRchkleFExOxzHt9XmS5GfCdz8VihZkKnoIUGgqY2RcC/Sh4PYoa5xhphDurr iqownLLFswKJlksULyibkmehMh1e2YiYKPmapZTOdXZv49L+JRd24GqLeB5g+yluJvQvk9DNTCIt xz268d+BQbkcqEoiwgAKRcrGFud63eIZZkf2UCYXJxZ2Ah+YQaY+FDQUFzFZdp4VGYBv2j9m1mrn ArUkD9URvbN9K28YpJ/m9XUyStQxzcnzHcUelVb0rbsHDiKwPUhlHCIaGytGXZGaHIx87v/CmZjm 1E1X6uwOfuIb5Lgm5ZXfL1bagIej4iI8URLG+5qW694g9GiqUQxsQRAbzfp1kxtv1SpMjs0wxY3z FCo774J/zmjaLrP3vAzC9m4BLyQ32KqNPnwp3hhGT4rhaUeAI/P8T24476sw1tOPStFxh3Rtg6Js 2qPYxHmiuIfyqG6p/AvXF2dDCfpdjzLb1L0GTuh5xGRPgPpMxDh8cy6J7bT35+VB7MN8MYRTTxk1 iI0Syf7cYeqTJElkfpFPz2lx1Wg9FO0bTr+287BX4rxVjRVr1NKXJdX/lx3bZ1qtxD9VrX5wssa/ g+8hlMtaNknuPhYCHADuwVttEGxgzm/uUXKT14bZXyXxUJXaehXzKs+TdiRVE3A1vqoxx/koFWgV YOOnxO8DR3Sy46eoAyfSz+h+RSYX+bDuxt8RYDEH+pxm3i/yeczPGll0d0kv97BVKiFmnUhBBmpf u7I//Adgh8VIVi9eN5L4WpVi59m1DwBTRv2LzSNqGx3t9q7cqe58QTcsxkzdsAdEzjwMdLVuX9Qy zRmL+4M5LFYhExLZpAaSlZxHNQMeNqnQsnkS16SQ4zL+oPZ1KCuCKu4ecgs9MZXedZN5Y5GKViTi 9l4il1SYxSxi/FvPzXvAqchJDd/T05D+n3PH+f2euP0nRzwdKG03ioL5jada+4rylsUZwNUFrTZ7 LhVtNQijw+owXMuC9ALVr/ljyackOwlLw94O9l5hjoP771pF1KLBDuhjyYP2WFdYIaSYmW2ueyo4 Ec5yjyDRmTNVrcw2ryz5qBBCH8Q8aK63t0o3KuPshzrM4B9N3iEa1Ny0ZqE73Hdd/dcX4I38ROOl SmlzRUI/Vj3V/i5mPsuzEqQUwIwQFqFNRosLvgzzhXeIscyWniwJDIh8fG7mqflPF35rZChLsAPg AVcetRB9qyjswl9z3IfSIOBL/0Nd9jX5hRueXI3lbONX9Pq73nEtUgFd9pj0bAurBg47gl9ursTO d9rhSW/2cZ6dMOuxkmnUO2TZWshpE1WARsytWIux0vssgiSQEtHcbRlP6YepG5gcYEZqVFqRxI2+ A9eslriQDpwFX8+sJehQFT3tgulMkYcsMPuG2Mr42wzuR/sZWGlEckjkHtOB+uVqojS2FuH/EfHX 9LGo4FbafG5B0Wi/e/mF0q3mdMjggw7anbZIony1zDs6/jBTI2Npn+nmtAHbxND9IveFyK3ADFNp P1RnLxqIndXb/UafsXLRpFniykuqFb2VUX8ZkrPE9ccNVyEKMJtWeF1xhMlkoMF+e2xNTmldgz4J h52HdQwyJK9EuKTEF8q76cttCqiO5FOb1yf7qS9pfXk4HCaDNQOyFZsRa/w80JrZVuseujCB8ifn de0Fp8kmdOPS5HPCfP4RtIA/wsl7yqvFHR6siu+AIvcMR9nabH9H7JFLAm/41OTPuZUxSYhBrS+L mqm5ctS1XucEmQcjPj2ilvLyw356tj3JQLV6EbhEVNuuP75k3Y99rrok0VVM54M4ztXI8DPFfs8t f5PpLovydKvPppvRx9JMFkzom961xLAP2yQ/vFrVx0QRoe34Jw+YO4sWtb1K5rhSLiOXoa2nol7/ hVbQKbAojtYNa1UWM+gEUc6wv9gx4qWJwEWA8HvQML5Iv9rp+oezHHDVJBIKfTko+nwN7hKttQTc kDATjimg1/HIYyWaLZkOqrr3+zxhJOp07gyHlWULYE5ShG/LXTUI+XxHwL+8V/nrdLHNyiqs5ld3 fbC9mgZ/UsxY865jcgdRySVlHTPt5p8hA7bzFEnvvxFmk1DA50kcSQLriZ2574QaODT65trbhIPn 4FHDF8AqetC/05L6bRACzid4TAsxmcjwyhO/n4tzCLUZ91TMAgHZp/ccbrw68Bw33v+jbJj3C2/f awpWH5siG1rDBv2dY/Pkgxlu68SqcQsiaz/DwQgjN3zf5M9pM0KiKpy27yaaD0H1mOOsd5DlSuv0 Z+2WVTqgKsBbbzp8057rMqYZrNJpsWOK0W9yUks0S2TW1tpTrJdedELffvkrQFvY1ehfJAeGBXXI hHyX4nPtkK5wAAs233G/szhwU9hPKMG5A0e3M2m7nddUSxH/6hbND7yQ3O5eOdl/YUQ9ao8tO7fp c6OugDHUubh0auvIbIeeKTAmwTW7UhG26zc6SPiq8qfjNDU8FJGU1P93Rr+L/BYOGOGPNqixgSRF V2HcQ8pF/CoFSV8UJDxGTXG6/QTB87w210Gig9dBX7zLsosZIG37/OE5m2shTUKaEVF4okgyxD32 k0wY44XeXp2WHnhejMs/T1k+BrKGZkNxKv0rPHuembsSDIo7Mk+qMnhTncLNv3nMcK1I+ZZzHCMO NS9QUbp+9nL0g91PvtSAytNerKEcBoPRrY26UE/aX3JBSx6So5cVNjRRysWD7rjDQsPbMsboPbuF cs7xKnfOvfzlC7ydFMpNuRDxJ6IQfNzL+y8aneaAZs5l0jr/KJhbe9IW7TXGKndPw2zGpYe3XUhp PRb/c/MK4/xu1nhGVIGTGuZH24BHRQ4Z1W8YOwHcGjyMsN7Q1moHmnJ9hwvwD7/eZnCUfgpX5ewd c+v5NPol7MqGYtW0n+4606rEd4PccAUZn97S96XPUx7iKYKeEy0kIRG2yYtVaVrm6upT/g3PXBjR IPOR8bjh8G5kjj7PsXTE9eSqExLKtwCkPdt9ck+qksxa3T7RtrKUJ61rzFv3FFeAHPpEMXnhM5Gt iz9JtsONY8EHYo5IU8NOE8RcysZ8tixNeUpdsGOQOWMIwHwANFQu9QsjIler5+fzIf5rYBny4JSs vtiL0/kV1NScdFlZNrJSfc9O3ZQgdZZ1K1qBpzicgGbDW8XgSUsTGLZqJLGFOf9Aoyv0iFhzsGRm vpFo/IWwhbzCUgftIdzHy76SuivR4F2/+SJx4380RX3VfknvMUAyyh/zEPqtEZ8ZUdGgzub8r95o 81CCBBx+8ZMCnymvRSYn3kEzK1q+DExbrAcYPQTjbkjMLD09vmTBA9ZLI/0VP0q6UFRdf4R0z6VL g+8MakRmKU7QC0ve0LlSPZyjGWYvrP3uf4yYp3folfgcY4eezxde3o/ythBub85Q8QHPnZ8f3SZ6 g2Xw6RB4vwbJwQRRHvudhlffPE5LQIMbxz7jRVwnTUdYVIV+ekITB7f+eP6mG7W3ikgioKW3F4wM NBhXZu4U+WnpoIIgVD62I+uoOP7iqyg+LZMyIqKckdlX4vl7B1cyisvsDrAcrogBoGEv+S7BkmSP 4c59QFPDWpXvD4rRYOjamM4Q8+sKDBN/IIirtBW9whi+SEGmpU3hlwxq6FX/J30rVom1Rax6rj78 QZDVIGDJUUEpoJuUB5kuzVwhFjws6Y2eLe6rhzm/t0c+gCViWBBO341+CoYmkPtGACDglgP94V0n bPpdLLvFLkE87sgkvRk0uPUiTn4PA996zExeh4WI+Y1SiMAaEAQpVI8kFiMmNdqN0Ck7De9QhybU Mnv0LDKI0fqtxelkzL9hDxNCm7h+bOP7fSY9UTqXW5PjiOToEwcfFIvV2llTOBOjMhYegdT9+RQf H8qSTKO2eO0xmw46ZFZsQvpfLMH4D72MSc95dJ/p4RHhIeGvz0Dz8ZwGyvIgZJ1B0trh9FqUhu/p LLtvkjLjtBdGKnWOfmedh93WGnmM+rruuqeUEEcIqMKDJzy2sNYAwP3K1pa1CMg/+R33iHlr3PlK spuSjzTGbsh3ZwBNVjOsYPHP++oALpEoJezpYIXVOloWrrmSVdrrTVyEMNK5ggXbDb7oIsDWpz9T NegjKpQQk0JEi4dTrTNRptzWS25Y2z/wTWPjEudMt1+28aSv8vklq6tkv3zsnONPMhKXAcRbwlZM fYubnxvlpFO7UCqzau6yPbnDppuiVvvDbbfOEyf6HL4Ix6IpGXDo1Map7SYGAJpApK+x+5Rvc/1Y o/PiC+R6eXEGvB+2sCS48zgr4WFmnQCa/T8P9wOGAM43WJhnsOLkCwLZMbVrlTwh6I7gf2aMYay4 B9bKzwV398Rl+JlfMFs3OYY5TmERWed9yYrgB1YtpKwY9TPnY8PglOuGRMdwgW2VdU2hkz/VuwJP IjyxyWb/N7EE67eB3vpsD3zdDXXDWdwm/eFf/U0ol5ly4p+s9BkRk5UTe5VRrxj8YHoUzZJGX6NY dyR9ZhnSqTzCbrzKecfeEO52bpRoWqb6TVbT71uvkzqasTq7pK3F17ljMT6I8sRX1MkP9HMXgAIY FUlTHiQdpWkOmk6WllnvWQ3o5vN/HVIpQ9v6t4OG7Kfr2U9bCDcEQ7vfvmVLUbUgKoC1ukcIndCW 5Uw0m0BUC2FULjh6reOTTBuNx3Utaf6VxTXV8ABzfDmGCdRsi1tJe+ujsoaVX/LkOjuO+cEO2CdV ZmRdDkm7Bjd01ppsB0h14MwU5IRTA+d4jacB4Ta38rI1PVmp/5/bD1VrYzhNnfSrfStegqwBW/GG zWawUoCBnm3R+O45ZJtWn0k7eF7P0DmmmIk22/ZQxwCQXTNLNpvtIWYBW35Bq1Z7VgISQyB8b0cC 7C9qrdQCImZX2wzYF9MH01jgU8VwaZ4cZvEnPiTxYYbGF5gDldZvzF0A82TTpCQ7Ffs9p7D2VM0G oQiMZxmaEEaxfo8bPmJAub8YeRox8M2N2tLJbMT5ejXuI1UMABeKg1ygILhdw7RmXRlm7GUIeZom +RO7Lk8boq5cBxSDIYkLhPhf/2oEyvjGO8aFHBCfiEoZkYzOKdDvhaTaUw/JSgNRnIrb2UixrG12 z3vgEOHxSjdAoS8iPffi3IOaJRpZvn+cxd5NKXzSagRY7VptDy95FHRSMtM6p/N6KIrhVz5advmF RhXwdjj06po8jd9V2ml1YFuvbIQXQIlVORp21s3WfpXKIkdL0V6ThETvePMFf43GTlEs/H71qq1H 1JjD2cOTyc4EkwcC0AGbvAN61cozrM61xf9qXBp3L6E3b9MJa68xtQR0evEDyQ49T06LHYbXdENf CL2PpR5nYpNZPHw6kWwlr7rsxT2RmOf4x3hJhtTSExwudQW+yZtYppY1i+sQNRLztKEia5up2qjn BT8A0LlghsRRVHim8FBthXjXt8BkuW4n1gLMtRLBQYtxAc72YqMERF0iTuVHNFxW86Re3f96DuX0 AARzx9TUfkwib7P59O9VA0T375cC73/TRq8Tl2spMWqu2AEQYRnq6SFeegIySMP99Ras5uSOswKj hkDvgBM78aYNOwBjRFuBgOnN1jRFCPLg1s2SnLWGcWfrJk5BQfc9hUYLv3yWVO2PxcwN6bj0WepP UGIjLrDr+bTPKJkK0qt9w9QGyPWJvpsUwX+OhjVCOQQrEwIf8IiIUZ1gYFvwLbPGq2WsD9tYaqEl C+2nSjuzeo5++/j3OwAJoU1DbgMeMFphlyVf74rlBP675EJyOby/4gvpScafRadWFsQMGKoHp9Ar J9zwlha66Ohl054GPZj1+y0VOo79VXw6OsWKryGWahvJ65W9EtsTGcyFR1k/hZ27U9COfzlUe73p R4jwskhBtP+49hHYS6ch86uyzp2k62TXRDZUOpySDqNkd7VXqU7DQ3VfMsU7t/3L83/rFnSaKCDr hEzQFnLTUmMWLG0cPL3VwHDBz0YWkkwggdoLPe3f1ECrFie2MlkLo2ksvGy3p16QXMVz3aFYJEeC VSKEgJEfpBeabquotdADC3KpIbQ3Mwm9YAcHOBpsz93MnZls565vJ6h5bxg5RyQzD+0jonwYL4c8 H8XwzBlofu/OdWQKf1gavrUgmoGsYM6ONw6jAorr8JJUEaRuI7tw92x3OjNLtIcZvAqNyhZqtb30 63DJvkcoh+aBiXyJeBvH7961Dk38B6xc8UYbGaxsu63BYwyQVX8iOMJ4sy4rbDM8l4Jgn5bs9b/8 R76NV091Ju4rpohGhlooaD31k/O1gTX3C+rhVyjXkfwg6EADOnVx0n8qIgx+HkqocJjnvLCMzUlq 8SY2nNIraiOZubJAf7yPZaZCAjxZg65bsZloUdASoBGbFwLpWo9lrnjrLXEm/qu9RA3Jx24iZ88/ 7gQDxk5kox6SNYSO3phGp85eogDLzDdJ30LApFHH8XOBCWflG9X3RcTGN0/nC+oslr9oU4kEKhuF HyQ9lr+1NtgnERg8/Tptfn+bCChU6TDNvjbYcuqEmG+Xt0UBqjXIsnaj8SdG3UJaxQf4g1saqDCQ zdgfz+bY9m51PQXfcHhWX4GFwGGe9wgUONzvxm5pZyKAvel+p9B0P/XylStm2UuUcU2BD/ojTEop lQ7+wZIe2cD8acylK3w/i/pTMpnHZN1tpLRmzF3VNtsdD13HzGuPadJRi/1wOHxm8ctVInA72Njo EWvTnZPeXb0Gb/+SDHBYmrqT9NSBoTUzK4tqvnZKIScfO86C6EsFbaO7hzQWy80Y47Fr9wMFDdzV zA0hfI+u3FBVKQDqO4PktKNqUBl8hkebW/RL9JYMghQvhPM3IwS9uneEAq+6xsTltrZZ2r7akmfY j+Gug8V4j7JS+gJSeWDDoBMEFxe6gzhZKWG8fdfI/DNZ+fNh10O2ut8HCEUdNV3l0O3vMADaXxtW uaUk1aLAoiQq4oKVs0qJerw6PFEv9OF6dk2P86VDjXqqHe2RwPYWS5jDHf7VNdSSntYa9jihDx4x +g+RLmyQ1WdnzTfjsO8/x4B2Zfl/JULiLJnbS7sRubo7+0mvMZd97098OpMd+x2mBRC6PDEXX7yN HVTrrsDodYkaWgcJ1+83e3Cn4EhbSDq/0ebhRFzrnf1/yR72JQkod97NaOguY9zkse1bQxtOsVf4 yLlf0zLCpGUEP2LwNkGVbEtCJIIpqBxAfVn9G7xBe2yYrLpaN8s2rK75doHgwmZTG45FkqBD+XQc H4whEIoZAhcu/F51aMyQfKH9dmp/Zk/zrwij4IKzhM09CCwuskXLglwBmyfvFGRZQb9bOiXNXesi fVvYCgOk8lBn75cfgt337sg3qHg0uyTY+1CK4iuNlaMLvI2NyKpuB1p3RCu1jQDToBzsQGw8KSpn lJYMPQ7SaFNG6wF6EBIXibnvt1JJ6nhbJ6ASonVEdFzKvyZsNjatwK7cqw0Qj0rlj3gaQf5JVorp Nw1JGcwv04LoizPYtEMqnB9ZERlQKSBBwxE1hIN1e7EqevLbPSKw7yy1utxn8myoDc+ogVCLt5Xu oG76iv42LZD87qktcO4LMtC2HQTmtBh70xxztz4NhOvcObGDu5SbQxn6r8mFK/RF84Gl3SBbjiQp te94s8u1H4KVH1ZUN59ap2+fMWHICqykO+3KzHSQvcOcKWQWiMpG1h/sTzQ/gdDykyiKKF+5uGbg gvfzFUIToFOhCBFCBDK8VFntZlcM8OdcCJOI8sUJ46XpYyEcQ87Kci83AQVc71yIfkz1HFAnXdtl xnU9XMalwvv2/giskXycl4p9b44p/+MsrWsJxigUEzbnp7zLMEoAIdJJb9pMXmZ88ayfY2otKV2v DpRumTzR8SJMmzl6W98+rOCddNV88hPLwQVRv7XgE1P07487nD4IC1nI6elve9u4tLHlR+sfGDR0 ryHLNjU/ntUHANbnO5bRtBu49zmtJLRjLASUzLA7EqG16djIg68vI9k5liQ6kC+3LD5+4Qhs66Ws cyJXp7ABaCoUzckz3pV1dqj+Kvrghn0ADyrTOpBgiVeFW0TsqFnjKkjMUHKMBdXpLPmHiNmwTLje yxDhbwaCqaZ1tWCf8jA0SQJ/JtGSrKxvLxF1eSdBXmkR5dN+w76WUswn71NyqZpX0SElEYNxXyjJ EtkCA1ZcDCUEV43S1InQFay51e5z48Csg/T16SRQQR6KZKRWrpqToaKks+gMsaK3IxAKMaovX3QG RlF1hvn6WKlMD9vtWUtfNA466GFbtkchmHopsJUtN4j5IwpkaH9Xatz3UJ8Tyb0mnhC8THumoT2v 7Js/e5nPDyNK8rB6CWwpYeafx+OB99forb1CtQJUVICmmhtzottNX0n9wDgfHiW2j01IQIM+9Ir+ qq/0ej0vaYhP8ODyCkOdXIMldm3GbIREdcAj6T3phHkzSfSqx80B2LwVRFgOpYziW4iswHr99hD8 xl4qMGa5cttTY0hoUvbAotQvsEcWISfCAnevPTBMzwz41B0EfxXYnqD/oO0widycvQLd3Y1aDcrT kiYT02ZI3HHzcY2Iin0vi698u3NwuqMDcuHiQMnQURmCl2Q9YmGG54kF2PuM/u9jziGdSi5ozIbD c+D4CJW1bWIXoqvyFq5EINFSpC/6Zmt4xBqZNj9SJV+hRmeZyRSq/aZa2tTOHQ5TiOJBkEHOq6HJ 1kE6xXD8yFZ6JH2tEqiF88KZJXYDpfxU2hfK6nXc2ZoqbkRaqlVCU+J3QW9o2AInsEKfmOEfapRM REKqImS1yyofNb4rxlZ7s5qowpXSo6tx3Vr+jL+EZkNqgXTfLBjv8cUPmkVu/gq+3z8/aQxDtGJJ YM0J/pSh1tY0KKHlya5dAcJ08O+evPSSJfAZhLWeaE4sqLgM6oETWCVNwK/vIMp3Xnl9nZZu/3nH 90PQK4Kx0P3vitPI0CPCRweGV4gbhfzXvvCfGDZhgTsPz2CcAuYlpQmq2+DwDqOHHRDSMcBYxT6O z0L5yAR6qDcpmoI1UmynmnTFEWb7djD2tYf5bnxUAzJr/Xf9RktRIKTV4WBsGiFBLqqOYIxYimeK DmDkC9DYvSlMgnAbhs76XN1DkVV+4HCNd/UA64hcJw7ljcseABEbsFOksjeMjcS5Dx3E4FtXz5Tb yqY+0d4CNMplAdblRY9lQ124+8GB5w4Gp3cDST6j/iWzS8SqIBIa1AjtnzmgjfYashWf+V2OaBLg EwZzBnp38GTWhHIQaX31AUjJBtYivr7l/USIRFH5xTlmSmyShwWvKVNPA3/NAtuvcDyV9T1+UIfu Bci3/Plf/NCVQQys4X6fDAjSfHnz6glKY7Qv8HBFSkBHJJ71UrxuLxRQd6q5aUj1yMasoyIlm5Bf me/YNB/tsSyAHWfe9pvj7MmpQRkQd/xJrsB0OZCJXTn5ZDoRnwYF09VvMOykOys9xXrRgDiPFrgY uN3YmgBOIpNT0Y5AsJuU52UjWOYRKpFwTc7Fbit6/MZrXLv1HNmvnU0ypsoMpvnjPbe1dzZ2Py1x YcZhHhwtTlRRIZCLeAn1oscvnzQ6SSh5qZO/Ejv2YPkUo5r7rIGuYDn0c2oDivC/I7Uco5qoyhbh F0LcgzrdaPcy8NiVcV7xNnb0BUTWie6XAQOOYppPFeLmcovXbdpme30SsnI47V7dSKsYNrZGtGYe TVWv+gtnhebybpqc9aoMPxsBNzmBGv2xWdu5Ms4kakQGouaOtFBjX6Q3a2RkyIDmD8grMKKzrgOm TYyRCCOo7VGy5Yxoseg04NOVle883HNw7LzVRr4b1bzGpjupOXDl0mxusoKE+kY8lzuE2IdrsBi1 gRBadRCppI/gIZIBmIQAH7+hLhokM4ah7UzKPhL/mXzAGu0ESbJbxrCVXMPDog9n+zeINzhlk975 CcB0vrCNf5vaSuryP5dfvE7eaR5OEmyneioJQJDsrhvn4Mc2Fgx3ePZEtQ3xb0dO1fatXxnZFRJ8 2qr5Qzh6OPHDP3dYLLiATr+HAZeUxptFISbiuGWmxWU4UvJuuwwNmnqsnd7mYvQW1oB7grvfpX+q Hw2jnC6RlctOkK4IWDUEhbbtOcKLqH9DUy2q2OznfdrjMSw/yWjl4M0UMJ46XvRi2Z35mDfUFdMZ LzCbWlL3ExPy2PoEfYQMjr/EZUJCLn1R1asVpK0Q7gsgnzlNQuSBDXXKK+UpwUHNPRhyYWM8O19N zqMP40dYZLY0UI1ge1EZzXrWoZhaeMJBEcHQ76EY6RA6BH+xozBO9Xw4eRq3Zv/P8/LfkAy5GsB3 5oD/K1gmSVDlx3NlnhU8+4BjbQYvpJZ4LBS2/9c6BegZBsDxq4PW697urGxcud2cRDN/zflCq0yk FzXEi5GzG3f6kxmY/pT23hixi4/WP48FJXHATNdscRTQYQq7pghon+vyl0iWAEt/yoNfqgGGO6Ik VRTWJZ+pLD6dsLKgPceebvwegWivKj+c1Hial9Nj3+X+i2v9Ugy7x2zw0L925f2hhQHMn5AXmPXb km4eUBBO2tWWiR+EQzCPRmL6t3Sq2g12klCbIQF3tQo/TdqVw64dgeQ7Iu4FI8M7OfA/E0Q7LZ2R FoZnDDP4qXdKyNULvy2964n3brAk8PgdGv1bR06QMb2LO07TM4KwuJmFH9nnmAL38o9e8VLqWkdH 076Y7S7RTQlQ+ds0nVU9V8H2oO4M50lfw5hbYSI99c6ZRi8hb/EOEWz2jjllp7/tPkGh//kYIHhw b01BH2dm9XUgLmQrN2m1EALXyZE+mjVpsihJVIsjQzpdgOwxLPrlvCJcE5kPSuJUh249Zb9LR1se 2Y3pDEM83nHyOKoDAT+zBCB+KjJrnOPCVHTaEh7Ye4WX90g7MZdmjxtBagVxL5Sl5qt10JhuYsFU 4pMJOS9V9uFN3ZcJNm7ElHki4TSTN4joTKUVH4sc/2p69IOsTaKEfBtoWC+s+WuWpv5b7VPaKz+0 zykWtRZVVloKLQjWKH289kGiYtc4f+Jd4zvTbMqqL1Iy0A/RV3A+ChldawlrNxJeSItzdBp55jkj E10IMzVcyw3EtIrF76t3Py3HqDKOOvGHjt9NmW2LrUV6MuuLxadURdAeQy5MOKZ8o1Woo9k7GDWG 3jVX4M8eWypRIFZTRTcQRHmfJgDdnU/UWSMNX4X0vJZh2+raQrRuZjW7/i4lFvMU/UcNytrPm5Zx +CEdcphL725jDUyA2qDeEU0xSCbkFHaI/nxlepgtAlgD/K8PsonpOJ5U9YxE6a89r7kDcZkdri7/ leRf29ZZQUBMbtIgJIGvpY/61FaiAG1+DSArOv6c7cyiBpkXVO/RGtyBnJhxpimvwZXM19490Zk6 4t/6Qp8Qkafgn1Fo9uAgbk88zzYzthwNcFZgPdHw09uFbL8Sa0w5wZlkFW/CflAof7XCj4F7DhHx UgBEwXjvEpTt7sb9Op/+fWS4dru5NX+m2p3khQt3HKRs0nmiyDv5/d9jhIF/pecnHnvaAkXKIIL1 X2uhCRUIrxvgI5ss9Gon7IcohuPr9WNScOIifb3rOfaFNcBinvRDxzUlj/ZA7fU0l4dAIuufA4bI 6dZXBXWqQ8CwnsqxIzs7YUNTPwxWUerUNzy/B8NQbuC1CfwRlmxc1qievC6DYT4kgpp3wHspPGMy GdUC1k6zrxr47aTb+PXosBKNF5a9hCnD+3PpI7Ob4K7sPLSGTVcldSyZtlPKrtY7Zrdiaf9JArql LFNsX14RfKxtIhFWdX85S8vHxVrsA4gPlOC3YjXIWEJr5rjiR/PuGX0joR6fAC3BZ2VeWZZlECRW oyILt/rjM7gt/U2CmjAmriAs7wGUezTmiP/3CtqZvbSxiUoGEvpThYkEoaI7HbR4+QI8FeBazZlo qyV9NDDf74JlzhZm7DgE5Jx/0r8qQIW2sLMAkOBjNKbUKNW1DOUiqBqsU92YsN/Dj+p1lPzYIAwx zN1QQ9KAlC4jnlSuimXpluMY0GYtoQGeL2K7WVwKw3eEbqyLpvO770w1pI4BMNG+TMVAsHfbOPzQ Rq6r2wk6pXMy6oplsyzl67qCKus7B+Qkj50Bip/A0y58wY9Oxh8vjy8c+zrFxkTilg+Z6THzNSBr ztJrb7Fc+XXBp2/12nsKHtYbprHlTvTSDDzq3TjbzkinRikB8C3OaOtUeoba3InhZBf/Y7CI9ZZR op6mvkILePLQk4M6ydMNtzsYTcCYZTgTbI7OK26iwxjYmPeXvlMnIzT9Nyt2sBK29WP0Or5GmWIS qaq75iEiu1g7wUd0cuR0ovJ+2bXn9Ayv3id1o7QswIQlANEvjF5WC2QbxZ653ZmAVUyMZx99wm2B zVY6TnqGdqIXQOss5Y0oYQj/LiPeHdZ0rlnN/M3LjtgVvGWd8zPK10G3amCF06WIZ44BLMykQpBD sKzuGfZvyNY3VEAYMpBFRPLUJVGCSfyKoT7YUZjCrYWEROs44UHrdT3c2znCtfOTGpGROH+Lt69A 5ht3/33seKQDIogiZXEVMCkWooB+IzdvCy1ZRF9HlwPCcniFqHipCTEeJ18Ikk0e8dCWn5jADXvD OgyaX3seZoAdcOSVW5UeA/faWg65AiokVGqaZBCwW71EjlWToNunO/dyE97ELoM1RHuULIXbA9cC vMlZ/mNmK0OUbqtCIJHzefEf3OdLi+IRer4gMbGaYdc5UWviHAN0udkeFajOwlROuHHLypSRszA4 5wsikjbHYNrzxl6M/NKdtZJU4ewbTayczZgQ1PWfcdS/vEsD/ErtNtJeoaP4Dvd6MAXweF9Dpdxc /hE4booFZd6Va442xu6hAFAC8gUs9ujaLZE/4lJHh4JFagVFNSD7xvA/CgmI6FTyLZuE/dAVZqzt 7Ckd1Xte1AxogdxKN+3VW4oxX3v8nKAeqhTDDhajLyNW7EFgxmmWvCzT7IVN6ph6VJmnd5D2mdIh qldXUbcUj53kmJY9o7V3sqKF5tRo0PLqVhNdqLApe1Gvem0HWnLlJh8gEB6/aHH0+uy08YAXB1dx KNFZCJ6tX84LAVEumjGpvW6B22Ga//IYGdxs6ffvMvC6/XdVOEDPCcMTfdH/BU8uNiNMOGLpv2zm prXgoIlVpz8/wTMsuWu9R6Qbig2Kcv5ZyzQRMdoDTdHGnivXBwJK5AWmnBhnpFzmSuh7WsiqJPAl 0wsUO5G00IN+G63Z4WflUfelpExXTMTf7Zh75nexV1Ay+Ysr/hMVo2ndV2Rz7WvUz2L9TvCYAigX Lrf8wGxOqDJLh6h5Zs/6JcVyq2hFmO6xcalkNtQO6H37q8w3pPt6bm496RSvB5+GM0ttkSFM25Aw oqN4kKItY+Hk3gejXCxeEclIEGGwbtvU3BVAfmUkUQfgvTKwh1TCP+R/6QRSyfstfPezpXC1bsBY cE+ulnE1kIWMLyajRjKGqhJPsVssNVrDFn1oY4CniNZV0lSZ9k77B5cLbsMA0NjoMI3x1yZcUTL2 mfyXDoP/2XvC9BB8X31Fgw/8I3gAjo367HS1AHnxtcL/tOWEA6EWzMtgBzjjIgeiGfBUpcyhA3/C IlkZs9CQtF003SFlaMc+nXLHvETTAY3S5U1ohkLld/YCTPLgHoeFggi/fSwJVYjMOdp4Mk2TXlt+ wVHMpUJe8+7JO/EzEIBm4Juis260cteuysQOQ4CcBZaZT9QNQP7051SKqDo9aUD3+uBbQ/NZTKAx OJ90l0S3mxFKOq/DFO68ZaIzrNE/Cbto66x0QdjCJxsVUqWDfXlk0iLRnGXUyjb7V/UeVyYzVyiV jm2SF+iF88VOsBkGJtSDoWsP5cg8e5ukc3BX1iwPaynlFGO/CAy2mwEJGi+nDaHF/7/1O3mFHPRd cWHrPjAcQGLpNme/TL/yKJ3cn/PgqpQMWBsgh4fBavB3+8sbMq6QVYVrD4up7KXusquSbss2i/JW 6EFaSsQ3OI2ORWfJlRcq13WWBFWW+A6fl+lpHTjVGc6XuGRySeB+o0w41vf7MwhpUXq5FVF9j58C BEC4rsU7NahXLi3ORDKX8BB1p8F1MqS0xzIYgEW0ENRgyauk3tO10AaOxuy6XJKWdYXykKnzBYAP 6SyhI91P1UdEHwN48jaDRp2d5VkufZZHITFosm1LXJ4PtHF/DflTdQw6X9P/USuVrFipL7KB3Qk8 8JcKz/YZuV6PXfqAeM/jlMgsETypO9LENffMTOYjr9/htimwNJJ4SXL/VI7Ag4/Sbku4YXzbDrb3 Px+bKSd6o52jQ0j5C4sCVKtT8PR1MN8sWgY7MJ5nDg/f1FafyUalZ1BGdKKxMAopYHHSunZRp8Nn YNXFfTwQj4gQWpTePYWrG3wJc2rMhZqU+gPCUBUsL3Q9Tes276ziP+N1AtSY+VIhVRpUlSK/VtFa yGw276hXA7SZ1EIQyQjgKxUHh0PVNmdLk2nhPQhniCjXBLtc6TkJl72DHd30MCjJXUTzuKlqpl45 d578Dm0tTJzop7w3GZFsSXxELTzG9q1tgraXJgXbT5QaI0VQy4xYPIvHQJYIJX6mBPm8kbfa8q0O KyygHbuYNd31qIncFTDjDxFtjdXqzWN+vrFKd2bGpO8Hkbc/wpXQN/Bslibf2cT6LJ2e2YtmHdEH EYsSN1B/ITvK0XDnViS3YD9eQiWu6tImMu9Js+YjKuqiAojiDeUMLhH8Anjs/RdFjQoKrB5mnup9 pmAL+kWXyInvuomxZfFD5BiONnzyMwI6dlk1Qt13FUM655+BicxD6UtJMoJd0JaJkA0rB5SmANyx AtiVyM9GUNnX52XYSLOExkF/7f5ymledyvHOAtasEHZ7elayfbmwNTzsZIvSkUJR7d486DdkWIY2 XP4bkweMfVwW9yPg2c5r6RzvsGyWZ0FaOicJ+PqBmqT9R+Xgj1KS7UBz+s6v4rUrDVsT4uu7h/bT hlLofKeRSMW2JdHZ9bfiy/QZ8dZfqVdqeS6ke2AFJUbBwv2QXW+jWAa+tvTcUKVqcWwp4/LSC2M6 yNxZm5n7GquXZIpNwY7SPco34rt0EkFZDGViu9h5Fj6Coe7CZKYtK7si+iqhqHHM0zarHG2lqWMO XTv+ELPVr62zD1NMos35Pq0TI1hwXm0OHWP7LEFL17+hck3LuANMTP0Gfou8fWVFePRfMI7F3N5D D//vG6LNlVDb98lHBFsq+2OFH/JUXw3maehopYafayVvej1TF/Zc0AhgCID9LN7ek7A/uazO1W0k fEac5kaq1MD5Kf5J3jdeBwzofMJNEHlB+0CZLWBxPSqA3v2Gt4viyJVb4RFywiM/M4ljLl5XdDIc U4eNYud600/Imu3WZafJEL8mEx11NxlZih1BmZyqLTozP2ArGFM0Gxs5PBR4szmv8hb7wRcypBHd Av5yDF8D205DVlbaZhXng1TblX3LSIPIQXL3L72T6+GE+irlv9o/66X2X7H5awBaG/j6/b/kntzO 6p3hO/WDM8vvtEAHa6yLn0pPTDAnWqHogqxOpUyhL0hktUY4qmxpwrZaaSYCQABTNYfDfvtrjEVn PgEKyONArdbTUhnf4anHF4rE2bXg/SycQwOwPOpzVP0VQN043lSSFd/hcbXqH8fkSMp90SZ5azR7 GhP4Vb/SoIb4gxxdABD06xYp8JaAX7Xs8pnw9ys1pq27+Yr2HsScUWPGgbtyXUUFopohjTS6bR4z 9qfjhylJSdZATQa7nLhQrsMnV3BlFHu5l9uyhqcaYfrxJFUCgvsBwwL7B7Mz5UJBaH1S6mMuYa8g Lrs/LEBvBCmWMCK+GsjngHTp8zI2ExJ2KnqGha/NNMRKDxjvuZHzE0Wh0dXo8xm6trOU9O+agsvW QshQ1e3kt46sCirHQ/uQh3y/t3Y7tqg53ofkeiwh0P9UX0FehX5dJSKmlktPi9LaVIXQf1DZ9UZy 7JA7dCZID9eaM8WufEIr6VwhwAGWePRvxZdmWAxBkJK/yCIQfanU2qbj9lWyy1VS8F+1CKncHLq9 C5pKpbzLKDU/k/q/ao/M0Cf9oy+Y0GgdZ8yGEVil8xvImfZ/R7aJr7UjnsIsHQBtRr8l8Zsps0/s 95PGKDfkW8xmBkshEaTlWzm6KzU+khMXyx7ZWmizgsjhtKZY5xUhLFKrZGKSoIx18L0wu6XMRFGW SMVnxxyV1G11PVxx12Aj3X/rVzGLtHA8ZZhg09VKcUgBlCudbuus0hUK5zxZH6gB8EaEm+Q9/1gi DhLAK2SFqg4S/BCXfuCIYJ9t2NpUgGaYIdkQd/NKldqyDKwVTS9Uzhs1v+TkZIoW6d/kbiCnDIAN eO0whnzOXirCpWsm/cbMJ0VfX+P8/AwDUaqJWb+em2xVXjtSLi3o1+jBF1XXOV9X3h3mWuBnmYyP Y6NC+rF+kdofUkUXnYPkGW+l3cVJB6MDwCNUtbOWIabuq6xy/Z4FNANMcwlpBTG5kq59OynFf3oB bX0jc2nTnaynG9J+S0yefeRxm9TYOfEiky0490M6teW5fRAczI5MhD39VK0Xd4aS2jKFcUeHBjoy fMHzLn3vORHOfSHpS960bK3wPt9MUBW+6d0YtHSDRI4livHNBaTkK0Nvh3NOq4w7H25S5/4Xr4vu K5dZgkfi5Ltghm+Iu8IMocPLd0xoshFDRAJbX6GcQLVGxYvUREDwEbvOQiZJPuRmqRPQ9vZeDQFm H0PkcQalJ/P6lGu97uhsF64T6tiajAFjYMJ5Fwoai+p5bbI+X89IBBA01p86JE0WQRHLIKh7F3lC GhLguj1SkpylMhEJHTzPScgW+mHzEw1UY+WyxIsvvru9EKuI/2ck+XILdIYF4N5We3uF3Jz6dqCK YN6tVXOwHAdvcylXnu0id4IeBL9yIqB3OFsqmOoF+FIRyZlJo5amYBxfb5EKtSP6K+7TGEjcK3nL 9avahGuu3Y7fhWUyUB9pPQAu2R/ZrB2srZ0l0xeihbmSNTXgTxX9aUdMMraxbRQa7ANbOeO6rVCD WWPM5ILAcrT5Sw8/Nq73940iGigUIwA1cKD4xGoBES9A+G+S9B6o66dmRhJ28KSl0FVAT7u6RiS7 aaui7XVVsrZRr1cLqziHmC1oRGqaxYC17PpHwCGNbak1R65mqNvMavS7BTq8h1keRTIs8VjzZqOh TK/ABzeKs4Axa1NR9tyfDmrx1CexDlIpNbZWWL69afqFUJ6pdKv84v6kcVBmFwKi+q+pH7wQZ0Q5 PyA/qGqDWFR58v+XyzPkQ4tthh5aJkb7E8S0zPghV6oH0m0kV/cIXeXg7/2FHywo+iUIheN3Ugzc 890PqQx031V9A4mlvQ3A4u0l/3ABGRc8r8S95+KhMu3GcHyUIWlW+xbAMq/mKfsxjV6jL7LKAxB9 J1viFxT12dV1F2lPcBMGliW5ckF/4tUsNpqN8mmKwqAeBqDgQ8PBQUOp4VHosY2iZxzdNBkMc+Y2 U0pAnZYX+vY5eC7Nr8kxFpN8rCOl9IBr3Nit6OsBZx+/kmH47McCJUkCXJKEoR0VvJ8Pc6GXII0X waz6S3c8gn2sCs1P8EhvdScuECQrC2+zObyd0OMKWeh+zTgVCjD/vOttOpmQ5HTXIvwygKZt7U0k piMHg80L+3gNifcRmMpnSQO4HQdEFV+9bDrlzk3tTuvdcP0lVEcF4I2QYbBrnmoy/pRw0h+sN5GY S+Zq0J/yvDTErUKtnVQ4VPR8Ut9+7Yw1EZ3omFdajNltMZTFtANb5bUGkxw+bSLilga1/LWO0fZc o8cwP0EBLqW2FfoiLxaz014NbaPkG5JMcpnIGYVuEOYCHuZv6jvp2niqU9qWi/vgY+EFjur3VH5N gd8W+H04DdKVA4pbYRsffkmA5LMns0wXNdhsDRzNPpjqHlxCyxqsB5qRwMkEVAk8zNZiQcOiGauA tN35HqMaTIWbsG8RzDpyACx3uY0JjYNkG8yy6LsoyzSLfGPEV8ORNjVajywxJ1DhOJunqtBzZaB3 htL9WgPXnDQt6AJX+Zltz7hjYAjOimw2HP+qPpW4T8iboJmGzwkB0GzYo3N/8evrSCxrJfdlJmXX Db3erAoA+PQOhQFe+PW3Kh3RJ2XVVrYTuSq1YSwmIItvu4QAe6ayyJl9iLDTO0K0KkalKXQ2Wy+E Jxpp21RW6WK//gwf52Z5LyXJlZ5PosP8FxOdzaC71hcDKQZgsByH1kfxPvoTwaYSESwUvVI/w4r0 48r4+6KETSH1JNooFDOPsonMULU9cuYzyXcShg95xBz8kLYi8OsII/De3H9NJtE6BBvHLL/Znf03 POErQku+Vvm44u34zSFJPVWp4mI7+qD21CKleRfr0L5obGZ0WxyrH/MD4a3YJfwOREDZdOYE5VbF awmfGH3ihMD0ZgxmLOmgmzcWQZJ0cjJJnMSLSBKLkhE5W7XzXYUrFiJ1rhpPpWiJvtuT6w25qrm5 DEyJTj43b49UWi/0axcnjEq6NtW642T8ygKZ0sc83ChQ8hGdHKMc3BOO8CtQMVTVeohyI4TXT1GC hgwyQFBTXjDrByeJ0T9RoEum2kzBhm0VY9Qhc8gyWSMhcpxGLRRzsGIE+Wp5ATgGJBWBthuE14cK N/pLt9gaAOh8yx2WjamlpCpFF+6U+YxjO13YZsROf4I8nUhN5u775/EBWGuqvfvZI7rOlTQzrCpT Luk6xi+QFQFujb3nBWaroFV16HFJEg2KvilJ65cuc5eLHlBpJe92WsWU0ChX2+jQcGnWyJ0mn6jS m3hpLe6AiLAwIMkCCNwB7HvEdtyC8ps1tNWkyDffvFW85YhhyN/g5lN8ood6UlsfFYVzVam2+P01 CMNocjoP8NAxOk0Kvht6fYJCI0Q6pWQm2SRI+MKk6iPNq75R6QJQ1EHu6rCPOuuIR0/gLcgKxlRb BxDDsqbEpTlzXUS/dlcSS72YW2DZNsReRSPdIZ0wY7a2uJO3tJ4Npnjm6KGDeNFS8B3Zkma2vn1s qmni7hdZxXPaXDIlPkvJtkP71BXWJ2FGp5gw2lFs9KlxBJ3U0E4nKiSo3/12xsRuM7fsKdo1glDA 5i5llGaVcY8pq3Rlwt1Mnl7AjePERDkXgCCuwl5yeOhvfkIuX0RlCB8GWBHqFV4ReWGvWdhQydSC Zq/Of5hcpaJuML+DnIONDJiLdc0Tn/5PGpVF/U9yBgKS3Fdr4t1DzyH2h067xpMa5HhKbwMbDjHY SYDP+jIkgT87uBvnl5UeTTJyfUAJss5Isq05PyGKN4kiJAm/PTdninNf4k2EX5YEf9sxaGSXF4Jt Oi3IRMrHaaIh3tOQ4y+6rP8ii6UuNe9pHsbZ/RZUwOPO5ThWjQ/WIDpMVkbLjWS1M1At7QZYXEY6 GZ1X3+o6wDp7Y9V2ARsje192SGd2bJ3r1WnkQMF/ZaL+ov7euvtWXPNv5LYlIdp3MVhLMYKqTQGJ 5IW3Pa/Ie2JQ7ROWJE8/9guwkOBf1cwMuFsxm+/FyV6eAHFqdldiB8ntTHok9grRJxQxJfrXQQrH 4V4wtXUEQFuCCiggGu0PDfZ8ZOdEaDaw0ffsSCnhwV1HwnjDmJuN3G3eiBELkn8Rqop4SAJDpSTO Tq9IC40otnEiFWroBtPYQuizbeUjKqmA7nyYn3Joovt4Shl/pzF0S8NhXNKn06DVrCEnQx95fda8 7x9D2Gf229t4Iog///6n4J8+SZOiUrOg9fShPXYWpFuQtUlKz0iKi5DHKfM615oWT8nHyZ8nQVEw Xq4ktsHEBqMHOSt4EmMa4qSCzWxzMJUznaubunQTeCfLd931GiYmNWVJKss/53w/HwJNvMtyTbB7 2nONWHFS1eQtYaY6Koyw5OuarAGd8r+AFmP5XrmukUeeRJfIQb+DQ1Jn/kS9lY2DHf6aUPwM2omD KAjeiEnLa2ozTZrVP8asXepA0hVC8xiC6VrPgexTj9NrLLqqBBVIs54IIo38Zq7jH1PCuxgN/S4y e5Cp3bV9upIg3bwTCZ/0xgpSCJNrwr2QdQfP2ldlfD5OOKCbMn8VAJaLSGlZ/frBhQS9YiqNjJPU Bg6suVt10Cw+SCHCWhdF10s/c9/5eA1OPHVQYds70Mgyt2umhlOQB2xX33bWqJYVoK9JgYyyJRy4 xBe3M1mVMUAGWpYDKwn1QZIXXBiCi7sOxfhdv28ae4+BEmdDox+XwhT/igXX0+82Znnl7b4A+nZN x55TMWoDMVRJ4+4/6EJnS3rGCiooqS0e60uRlnJQ5E/8e9vzE+1CETjLadHBsCOHIFCYAJAiSbxg OoVH3UUgmxtm5jgUNof2xkyoTnIgMjm3PFCCGoGZDeQnCa2TJ16HqnBa9gG7oL9FRK0d/dnIhpEZ 5hlIqa1N22iP2NffylTCTMD8J4HAsuJ7qomEAuIAmj8zthWJSHdlxPL5YBvnAUWl95yTKDgmROba +fJe3yt3Vzm+Sn0Gd3imrgnZPVmW5ozlC0DaX8pPNtQYk9AoFG0ptjaujiVoWbsn8bWefZQGpZUJ uhIZbFcd+vTi87olh1OaLFT2oLo1bAeWCbjoEES5zAmqhjwps79ovFNuM75DRXTS8BDn3LgEtjkc 9rhKMe0hc9zmwPIMVrj6pF3nRPDy5t7+E3PI8CrR0ZCYfvmqSSZpl1LU6AbK1w0KZW5kc3RyZWFt DWVuZG9iag0zOTYgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAzNjkvRmlsdGVyL0NDSVRU RmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAyODAvRGVj b2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMjgwPj4vSGVpZ2h0IDQ1Ni9UeXBlL1hPYmplY3Q+PnN0 cmVhbQ0KYMpiRDnYOV86APaPrztCylHXDCmYjr+hAC7A1FUTY4miUpYIpfSFTeEYzQwTB+wZ/TJE pzKeesGuYcVVa3b/ZRduoSGWeVEDfNGBgz6KFimgKSWh17pxlHoZTbK2sQ3XsKHy3irSmeXUmI5f JzrBFryOAmCFq3Rvt6l0r2WvGVJxfycmMIQnaXJWOoi0gw4S8qdO3dw/NKJBHBMhtPvcGkFq2WLL MqakwUnOWjZNu3rzvnWjLJRCbODD5tF6XGk7zm43FnyTp3/NPWrlFdoxWZu4UNtQXnkoLZjLC28Q LTYdHaHMsi//L/+M3oHUd9uYC/+eY2YubIaUVKvkFlX8ECswRGY0OiCjIumbIM0bupaFzdvCSTZi zHa7vgBk0QtBWEYPyvW3wlhOgRQwMwEIGYwrvzyY7XlZT02ihLQv3ttJQkj6nsk673l8v36CFf1q +YkCvEHDyT1K0GDbYzspl2qRtpEXiYu12/Sd6InjvZ2aDQplbmRzdHJlYW0NZW5kb2JqDTM5NyAw IG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDE1Ny9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1h Z2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDg4L0RlY29kZVBhcm1zPDwvSyAt MS9Db2x1bW5zIDg4Pj4vSGVpZ2h0IDQ1Ni9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KMaD9+Mj0Vh74 5CmD8/17yA0LRGDFaZbSQnUrwcYLHzBBBpW1TbQEtgIVa5koT87bzZclg8KIXOb7xRwplrSJ/4rz MKRafNA+IJ94mEz6REBizNU07f0Q93cCJwucOpWfZgw6C51TbK4/0uOeSSmG/hkCgwIAhvmwrVAh 4IrcxRoSc0BeOe1FjDDz0kguIFGQ4RakLkgPH9lSCsW7yg0KZW5kc3RyZWFtDWVuZG9iag0zOTgg MCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA1MzEvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0lt YWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA0NDAvRGVjb2RlUGFybXM8PC9L IC0xL0NvbHVtbnMgNDQwPj4vSGVpZ2h0IDI5NC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KCXZ1bCuT L2+oXPNKB3BrmqVt3fM5sFVNxSVn+fHE8O+RC3M3cdwqo5E0ntcWOjmseLfDpgyQ6L8DyOg15W8s jVHQyYSs5IKQdpXx6tk8yFALhIkaUxAUa1ELUYqSRSQcnc0zOdq6f9X7WZg4OYjhvF9AUvQeBWA6 2wMX99m/OdtMMHij95TBhuxreB5xUaXn4j7myi6pRc8Q30odLXxxUn+Iu2ERbiVJu4KnFqDa/u10 NZzKNmYTm/mdk3ssXuTM1tAD9higwdU1oL/WNLqudtFbAf/eh7YUEVC8IhmcYyGkwDLaETCWysO8 DN6BhjZgTD9gyTcGgBkWOqkZMKT4YtZrDx5Bx2hNu+8BBD/veRNVzHM0v7Nenp9Xyabj/Z4JiuR+ Eo7SndXM73Zv0j/R43Oy1LmXyIYUNLZwwoq5IBjHijv7iyEIlxWwrZ7dzruTUoGf15U20vpf78+g cidQLHTs0LnmvDu1QqNNwIHFyV6PXxuxDbxnbaSOdGwKuk4FhMTT8m49zaJXZ/SbvXOb3iYUb3uu LnUhD5WpyaxfcGFLqRGiieNr6ZNF3BBEzJk5Krrdejep5wWmtDU8CWNy1ikkSmPUH60WdDN83IDw KahKXO6iPscwEUYH1yGxe6DVYG2sWiBfm9AaxMTOShmZGs4G/l4VZ3+W38MN2hZWk3PSdkTsjCzK DWZpwN5Xv2mDKcGmDQplbmRzdHJlYW0NZW5kb2JqDTM5OSAwIG9iajw8L1N1YnR5cGUvSW1hZ2Uv TGVuZ3RoIDEyOS9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNv bXBvbmVudCAxL1dpZHRoIDU2L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDU2Pj4vSGVpZ2h0 IDI5NC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0Kz9FbNjPOAEA5ZYe6AAajbhBwWzbCMLUBOyxBf6ME PWgQirxxFuq+k4lF+H021pbe6AZlqJhNL1Yu+TL1i/VpAgkqlKtS9uM3WNX+bBN7MiV17tfQgUz2 6D3gI5hdB9dxZxHEP+lA/Rq/hr4gdg/LLwJ2uslXfsT9iSbt/3VysoRuDQplbmRzdHJlYW0NZW5k b2JqDTQwMCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDM2MC9GaWx0ZXIvQ0NJVFRGYXhE ZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDI4MC9EZWNvZGVQ YXJtczw8L0sgLTEvQ29sdW1ucyAyODA+Pi9IZWlnaHQgNDYyL1R5cGUvWE9iamVjdD4+c3RyZWFt DQoNiPRDiX5eXyEKDg+XIWNVwCVvI4HXAT7zQe8yMdWTgH1GLYRuVtS4jxo6e3lSjmRkX800y+CO 0TBKHGZGZXp7yDsJkmBiOUG/ndyGC/dT9QfmC94aB6G4xEGAJHdor3Fvyz3lhnQF0d7/CypX3x33 SXtrraG0Ltu2Sqlu1E54W+BYU/8LsH5npTOFRlyO+dc7CTz5YX/m2TIE3h47f+f5tm4SFSEZ2/FH D0CV4yJVVzWo3UbsuFwcFY4r8c8NcmjZgHEYIgio7ti1vUugj9ruhc9qUf5nzX5GH+n71cQfozWK ZQwW/vWOfgrKQxBkjWoIOjCFHYAG/UXPQBvWS2Wu5UcotWavh9OcPYrzu8EkLZp1iQ7/awi2+o2P pwHeYxGPr85reBvqqnbfKyGC/AUrki1k7DBdUOj1t8x7T0s1f5Ch974g0nA1teucl0SyqnS4ViKc aAZesrKQhr2bGWviZm8CyKnoJXgNCmVuZHN0cmVhbQ1lbmRvYmoNNDAxIDAgb2JqPDwvU3VidHlw ZS9JbWFnZS9MZW5ndGggNjYvRmlsdGVyL0ZsYXRlRGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQ ZXJDb21wb25lbnQgMS9XaWR0aCAyNC9IZWlnaHQgNDYyL1R5cGUvWE9iamVjdD4+c3RyZWFtDQq1 H8xGidzecxJKAELJgNLT2w597ASUoD/I+0SyUcg96n4a9Vn5nVXLEB+BdcJj9TLb558e6IrSAL9F 7zzPOuyYGlgNCmVuZHN0cmVhbQ1lbmRvYmoNNDAyIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5n dGggNDUwL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9u ZW50IDEvV2lkdGggNDI0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDQyND4+L0hlaWdodCAy ODIvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCvyBgHLfAUHNi6/uLFuvmqR1YvhmKUPFKKlik01THzOm RfOQZOts3K477dPEytVui4dfVeTPWn2X2swutyeEG1PU1x0OhI1WsGm9m0JTowYscIk7DT4/Wud2 2HTgqb3+BSbhTG1hLVDUApsAvun+jf1COl62SyCDzWtYv0guIJfgnUu8J+LkteGlj2rnW2POMk6C vyZ9H6Gt/7Ixth4yIR5AJqiX5Uisejgeuw8olRzBUW3Z4OJYC/U2ssjYOXxLi5QmEtWnSXLxjXkf W/Wf/yj5iWguHKFK1JPmrQtTmbxXzyq+Udyh4XQCRDEjnoWcj5oM8LxJNYFYHV9Jz+/E7dWNde88 9bvnNijeNLF8ezFsomC4DVIOvDgOcrLZiXbVYj649mnC8wcpfCkcO+XJd54mxaF51BeydTrS+V4o Qm49mIgYLOaWbw1VkV3rdy+22hDvdpoqjIQrXjHi5apuyWJdbV4VuOJeqHYigEzbwWV64b27RDKe ud47OfSEPLSG91mx3dLecS+kc+cW0XLoRHnAbAoJ0Q712H5IgMSGQFOXzLpmHO2mpK7clQe/cXox bc3ZcvmnA+a8SUX41VdCO2lRMQ0KZW5kc3RyZWFtDWVuZG9iag00MDMgMCBvYmo8PC9TdWJ0eXBl L0ltYWdlL0xlbmd0aCAyMjEvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0Jp dHNQZXJDb21wb25lbnQgMS9XaWR0aCAxMzYvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMTM2 Pj4vSGVpZ2h0IDU2NC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KOkIQthXXPXBV33xl8M2DQMAASZS8 TQVr3Fzemo5E15e9/dD1dMqLXeMydQ/snaA6QdnfPzsa14vG8Sy/lRxjM+8syiKongz31IDlE9UV DSe+SQAwp3sCgtog9qkYpHIeFejnqp/C3asS0NSMvTgHatMGnYhG3U7adyJUXr9hA4VPWUQhnE7x rG2w7PirI7/yHlRqXN696Q2rNxLRhNGyEOLcdtHGJWqiL9eIiQvjWACvwZc0Z/iPNtNLbMV2DAk3 gTYs74aHffUbyGiXL/dSBlmq19hHlvxqMmzorSYNCmVuZHN0cmVhbQ1lbmRvYmoNNDA0IDAgb2Jq PDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMTkvRmlsdGVyL0ZsYXRlRGVjb2RlL0ltYWdlTWFzayB0 cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAzMi9IZWlnaHQgNjYvVHlwZS9YT2JqZWN0Pj5z dHJlYW0NCucpDjOIA7AtcCnuZRIY7BAVkDoNCmVuZHN0cmVhbQ1lbmRvYmoNNDA1IDAgb2JqPDwv U3VidHlwZS9JbWFnZS9MZW5ndGggMzMwL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sg dHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggMjMyL0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1 bW5zIDIzMj4+L0hlaWdodCA1NjQvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCkaYDtail8ioxFouJ4at XphxlWKZkG5vp8I/61HgwDeu0MSws7ZuNKEvlFioEkt8Ldhy/szWk/IjSD9hczOm5ttX+xAcV9yO z6e/Ya7bPBNCMI1ZYlSm1RMst5bqOEBros2aq0zQQFSkEe1baHMmskSM0r0pj6fU0xcTnfJ26jF0 /NUNsITaIfNDpE6r4Kt3yxky7eVoLr9Ia2Z1xuNXPnOriFt2OpoDb8ivJqCmXHb7cudDipZZ73NF owuv9xGfg7qcxZFXgnZUY24PMoJlc4+UVqL4iaZP3h57p7HW0mVbRlYP0/+5y+iDRddsMILrzyqV PUVpZ4twHId/9zTb65zSHMs2BVuBNy9/ji33aQVbjGgBnSJSdEHUQAicwvjEZsCuWqN6qJjzzZal jTtu1PBTgY53a/EmhJ8pY9jejHx2ZNK18qnB69T5oQ0KZW5kc3RyZWFtDWVuZG9iag00MDYgMCBv Ymo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAxNzMvRmlsdGVyL0ZsYXRlRGVjb2RlL0ltYWdlTWFz ayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA0OC9IZWlnaHQgNTY0L1R5cGUvWE9iamVj dD4+c3RyZWFtDQoX2H6J1qWeOrWXGOPJz7YngETTEnG7RjzT6aRuOYnVeFeYKE9fMil/AGDfU91Z eoQpB9V7vpHSPZ2jTL6IWHegZAgNB2wB3looqW+Jaisto930vkuNuai24uDCCvSam00UBw9kzNQf /JZUJuVfEUJH3ba52tCEg6QCer+OwCbEH/rhkcikbvr45a5viyDr0M8xYPwk8LMuw+Gg+axRUFJf WnUD0ClrszAiyw5ayQ0KZW5kc3RyZWFtDWVuZG9iag00MDcgMCBvYmo8PC9TdWJ0eXBlL0ltYWdl L0xlbmd0aCA1OTAvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJD b21wb25lbnQgMS9XaWR0aCA0NDgvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNDQ4Pj4vSGVp Z2h0IDI4OC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KAnO3zkww2Qq0QM0Is4LxbdVYAX4mVHPpI6/d /oH1NuGIk599av/f7WiT5Ou2KbgdYIfJ0BXZo8k9yHSe3HIlZjQQB2okrrMubwVFbuVekQD/AmFU L6kuptp5NOPy8dp6J6dtHHoKhtYB2XPHyxP5uifQVllRCeakunuGdt4OJ3qkGQxfflJBY7UhDF9Y NB8ecSdJ8IAC740iwRQQdtVQsbJOnx9ru7r/xAgklU9P3i/6nvMQNgvxlIawFjSZsqGf1d1HhTJt 0/nqrdiRg2z+Rbjq47FmMgpOBN5m19gtXWWdXn0ZZLE1ThpCcc3eE4rz+pVqrA3aSGe2H5kvCx+N SygbCysKImITr3hBGX5WeIgMwWB9WvaqT4/6rB71gXM57p8cNC5VCXuTv698rEA4pPNorA6EWDMJ oI4hcUjGw/L5D983F0hqFxhJ/EpjznGlEyeYzXglshnlJyoUeugTX9XNyVsOzJQLbCPPe9c84q+p LsAPdRxUN+bnJCw7JpaQDTw6r3eeNoMm2CzlnPNd/5FdO65SVrF5kvK1S+pEObMoIfTifu82CKXv FmgtzwdMuigM/DXXBxPjUFDz4j43NCJJp0AM7kS50EBw3YiskOtbS7vo9mbMpKGFN2zoNCtFkO79 z4A5DRWbwx/mwGcH5tpRjriEBWC2cHcUsrAavgxxfD4wOIMZd0kKj5CgRwQ5UAmbFF1UtIfUALb3 MaKatELq/XDTf7FSryIqu6YjSd/IJgYsR6ox0qB3bTbFJOPFunlQoXDJuXjtJzAzR6wNCmVuZHN0 cmVhbQ1lbmRvYmoNNDA4IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNzcxL0ZpbHRlci9D Q0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNjgw L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDY4MD4+L0hlaWdodCAxOTIvVHlwZS9YT2JqZWN0 Pj5zdHJlYW0NCts/f9VqqEbBFa/BkZa2a77V0f9WWeMPJGogUZb21FID1Wa0BnM5ngM4qlwip1mc Fw9EBvaSXMn4kPQYvNloqAz9dtJnhzVBH6f0QrU95aJkFiY2FDCduiF1mC/OamSCcjJkkVGwOEqb Qut3OqiFwNgIf5dxRt32PmINDCnAUe+yt9aeCttoiBfV3dQMIYB2R936HYu6E3bg+TJIFNPZLY/o Oli5vSoGNOjWUCDkYcTlMHiJJ/K/8PuEeT/I95b3wlpVCG8E7SknQVWI8b3DazZu7nQAMgomLpW+ J8cuaQPsRiwO7n0oovUCaRW8Gq7IwoTdoHjvGC5g9/ss4q/dgeLTBam64XxfdVaEUvPK907pNtWC 74faGQq1DDoS3CrzeHQBCpjjru58OUJCuZzgyOaaHBr6Dts0cVHNAAkJUjQcUsyBjUNP3Cybwyz3 B0WWLLB/Oqf9D3Ie8P8gSsinAQKh+D/psBNR7rDhF775suGirFb58AxNJpoAqTd41IHm6cd2L+rG OwG9LPOGbcghuB6VRwVhan5EfGashYcSyaoFpy5g5RuKtMWG7P3afmt4vpU+nmsiwqHe9qppcRQ/ GWuzvBmw/GvsyOZd31HKxyET1H9KHwYf3MFDAENZ8UAvqSXsHltRGzZJysvCGEUIjla96Z+FcQ6k WbbOaDU+urynQE57Al2khhxdSgl1GXMMIStWPwt2mVgeakKqzwVLU8KjspKV/LQD2HSI+MFH4TNm 0wftVFhEqoYoBK0kFZgLyYxfP2eUQY4Pw4wVh4lma6C3QpYo98ghMB+T26Pkqw3OZbfbgxyQWG+Q Vp0QGoUQ+yfMTLzHQ/Y6ygutWVU58++w0fgbGg7whhbwOV3kTeaZHUe+29/suP8A7UALSLO+6BGO MlkpLVsVdqDPD+PObLi1SoyNW3iVIyDkg1xzDWSusCUZz2994jmMfyiRRW7Lh9+x7p/PJTO2Nuqs vMDqQizxLyet+KuaS3IYx+SfrKaD/JjjaD4K3wJOTA6DWm1dF6r0Mw0KZW5kc3RyZWFtDWVuZG9i ag00MDkgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAxOTIvRmlsdGVyL0NDSVRURmF4RGVj b2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAxNzYvRGVjb2RlUGFy bXM8PC9LIC0xL0NvbHVtbnMgMTc2Pj4vSGVpZ2h0IDE5Mi9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0K sHrRUrpQJXFfqg+stGO4uRohoBfX6960vxLPh3jIVmlkH8yCdp1Wzi7IXxlrpsyM7uZSzy9CnNy7 Q/UfILZqV2gQJ/TDQUNfs7AxN/uOAlVQUdXmy0Fqgvn17gqA+4EbD1PXHds+Nq3tPngU36S2tq75 Vq59wqm4cSwEIxoE+805tnnEtzFlc0j0fOr8KO574YilHtQjdnS6ndmBtUFHKcYg4+oLGRiGdw0R mth8A1kTuUXbg+WiwUJRYkseGTx5DQplbmRzdHJlYW0NZW5kb2JqDTQxMCAwIG9iajw8L1N1YnR5 cGUvSW1hZ2UvTGVuZ3RoIDcyNC9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUv Qml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDY4MC9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA2 ODA+Pi9IZWlnaHQgMTkyL1R5cGUvWE9iamVjdD4+c3RyZWFtDQoR/RbmV7S4tzgtEr9vCSLZB8iz /mCCzpcb3N69ed9jdgnMRyWBcUyhFp8jE50LUQkO1WMhZVEKvI8sNc+rx7ch1QIcXwDuvd/wh4Nh Z+VVTGhWD4kyH7aJJlRZMNc6LSw+yziF2ykw/JqKUQBiBoQkqn3nV8t6Cor23s8a5fy7EekXgunW Coi4dFFBeHkjx6LwCzlZj/gJ1YS6lbHTrR9h+w1KBgJHS4N/BgMHAvZKWWhDdiXr2PZK7EiDI5R7 OZmmEAnYlzB08GTufY/44y4mZiBL4FduSdMstZga4yBaUKcI1VoCmFgFb1rzbCDdVXjd8VG0wb1y xXV0dQ4TdnKP01LBzPA/6wwTN2uMgBzau+It+NF/AL2RvJ9X2+VDJ1y8jMmF1g1qANa+6PbcRBHZ 9bg4QfvL8+Op+aAeF0f3KptI3D468iCC8t/2pUGsfqZq7Hm2Hgtt5KztG8SaytkYIOKTfPTjb7Pv xt7Bdp00/8obVYBf6zfDcxpuuNCeTyFK+2mqQ1pas/so6Agn/69Dsg5iD8PPOVu7vw/UPFXdiISs Y88+A4hUJzyUIyPpG/kmkUJmAygFYE/Lj10pl5eEkn/cDeLgrw6mKL9mg3jf9guF3L0yPVVdOGow HjtlrTOEz4Cdr/3KoiuukGnrOHVhSB0+daDkHgm0MZwXEnN9eTNcXo9qN2Z/US7xlheJ0O0E9QFq 0LIbmcWGWYFPsf0hJVg3/i1W5wgDao/DlolgeUTpKnNtIdTQovSwboPJCQ1ax2cCqZGmgWuSChpi GPLXk6gw9v5dS/Yvwg2nDvqpu3W23gEMThFA67ZQspJLC95sq+Rft7eug70oBt79GMS1HgtO0t52 KKEZVE8vePShnvFF/83JBLz5aT6tO0kmDfwEqmS+ICpJBtY4XbV61odOkUa04gxl1FTVqhVnOAif 6WB8pj02SBu2az22URF9GmYIvVBvDQplbmRzdHJlYW0NZW5kb2JqDTQxMSAwIG9iajw8L1N1YnR5 cGUvSW1hZ2UvTGVuZ3RoIDEwNi9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUv Qml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDY0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDY0 Pj4vSGVpZ2h0IDE5Mi9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KiD6dUQCkSB4IGnWs9d7BXn09iqKq iycW8QuSa5jaHxC27+HDQ+F9mu6hKDZYPvtdh8fGfy//2HuUVU4y/ZC1ZZt3um7VPG9S3t0Mp37g FmyXLIhdhgOBruxSHi3ixOAVEzciaK7xr5/UWQ0KZW5kc3RyZWFtDWVuZG9iag00MTIgMCBvYmo8 PC9MZW5ndGggNjUwNy9GaWx0ZXIvRmxhdGVEZWNvZGUvTGVuZ3RoMSAxMTI2OD4+c3RyZWFtDQrg S0aToO3f/ljlgj6Q3VQKA4wrOweQNm2Bm322yyW0xkAl801qYsfe1WNZTc3IPitaP8R4Fn0+c1JD ckrpxF9geuZxLBqtUAFKPuiG/jFHQG9TEMguKoYa0k1YmcFZnJZDbBS7chYtCYugaUzZFqTLSvRY gs1RLF5KS2aadkl5CT1hp79P9fBOEyinK8WiFbai0zJK1VL93rnnl0S/7rv/IXZO85KHAIFDomLN w3hj7L69ednTGeM2j+IPlYNqcY0FPy/W5mlnUaeh6fZc9hPXC9GmC2lXuOUt43wJZv6NZFqiHP0U NxlhoYRc5d3v1tHlS4L8g4QI8oqLwRG3/cyfAime+5BlWadMuSJl3HrHbBtYtK0e11UZIqRyqgRN z5K9VbBmL/wHv+YQS5ZpzC0yu1PIkvRKMFI54WHuB/huIMWmBxyeA5M0Wu6Gc5eKL3q7sjd1gfcm ZCEIgs/gy6C3CDOBUVEFZcuGiXN7M7wNXNMkzuE/a512SJZmLJLp3OuDLYir84R3Ygbal4LCjR45 7C7PodFPyFIIHY0Ft3IcvbMck00GhJb3tqkR2hwTMmm18d6+2gTbADZVjoFQGVTZX6O1Q767Hyzo SMyGUAslAdzjL2nuUke6V+rArlGbngBApRG6LR3chKk9YlATQPb7jtl1nUUpj2wHN13kPvu/5qLA k2Wy5klehD63dlwkQ73E7sAlbB9OZKLSwhCpju3MtSOd3jVKxbLUSWVdNPTecU9GBS9EtQkGGYFR 4oA18VfWeI3vEgvkSKT6B0iyKDlZvp0ORC5U5PH5TkEDN0ewAzTYiCwfgLv2OkpddgAx9eAiiJfp 9PpSouBtpo3Sc5nf008BoebLC7BLXHpcdgPgf025yLVOtIlwbgXMmOaBoITDg/RydPd0qkyCIB4c S+gMhWD+YCq99PjzSrnHcCLNBsrclqHnGxGCsxy+ujj0HvZIaqRPD2+CUwsR2bPYk8BQNqUVgn2M z+bFZYZWMK6hV8eZc4EC/jgM8T4AOt+fhxfq2Hen00KRwzYOeFkl6ZyJbnqiPSH3W9wdfyL30lrL ac+eqj/t36glimJp4RI+tkmCoDZucpsDX/wdej8rebL9xfeAS5Dp081Jpx6yn5XWycQ3TwPKQI9g 7YD2HgxOK9a8tW3hmUdGvjb+aFVUzL/GiNYVwpknqYJXdirm66QuCrZdlBcLAhWeMjnyI4/aWonm Rk0+aILCH2Xgtz0zVyoKz3rtuldEdu9BYbv5cALhFqPxCSJp9oLo8e6jV2uB5p7Q+xIAgUCZDAzR woYKYEi75FHRJWHX3GZkMyKprxSapi0QG5MezTvWt1jJESQI1e52tLOZsRbPvWgq9fcWlEUwZYfX 4N9rY4080BAK1BCsB+NxUZEhRlHfqaYjZ0mZUOjq1ObWqUZiV4ZiJubXjhHu3392xc46bu1i8V7c IWB25pFhlodTX6HIrJF8Ygp4Y0JcAAwBi5n9GZbqxJKWAXXBwCxMwfRHhj0fruVIDx93/jCURBVo 1SGeaDF9hqT+eIAiY0fQjiTJNjdSfuv4IJN0sppaio1OHWehiktNnLEUrbyyray2xJS1v567+0wE LBvFFMFrDZtBHP5wjk95HtpTfWs5cxW3rOVw+mJPEFEmROwy/eSkizww21mcXaehO/2UCZQ8M7wi w9ofIERXtVQjZPEF9YOa1RX2qAqQB0oLfREoTgEj2obu6UqJCr0R1tO6p35r7kalQtaWKm1TPbEY XvZBNJD2TxDSNrObiM72Pg1/QC/Z/htKXy7JPbhbIAX2ukV25tyKKesV42WSkeDCtYHPMLS44eco 2zfJQvIvP8whf5FBeBGV/8pks9EcZaHx32saLDgWermc4Jv103IThCwH6xi3JZhE93MJ/zmpptAM EX3jzMPCtD77boB9fzG/5tjt7FkfpX4u/Ug8GR0HFKigVkYX4mbGAOt1vs12z3AX21+UOKBgUW0h 2Qo1BBUDrIBDwKY+mwpoLjd5EBf+r4keRrK/RRawCNLrqRjqVDy6gHLI3mrGutycfguPSw08HgMW MZR+NN8SKeff+cFxp1KXO0m4mZwjMseJMm0kELphqZuR/qAUHQTwrlQ/1gSOSf6DLLo5AJZMCU0T 8YWkUp47mIc8EYRbzd3jMlXeVs9mCBXaCWkpjdrLFeakzomo7ULMwlAt7MwA58bRWDwtknAG5R8f 8O7U+P+1YVjwvmD2IjGXKIVSkwDHE+4z9FFT+KjlwdgeJRO4MHdxaPsmGjsil2P4z52jAo5PWmhm lC0tRVsuExcn1P3bk4do4ILYifO4sJNISVJPjOBBJfXU5J34UkhO1E+FOXooCrdkScYrTyWPYDmH MnilwemHes0mh+JNDPBTUz5FP0NJVyhWWjALOiv7DseZn+yD4a2kctTJo6NxBtBpIGHL2v+Sbb7R G6hfIU9u0oEdkp1+Q08JQhnbHXg0dB2wHFvEVlkq8WUePU9exDIb1WlIZ6mWzueTX+cNSFqg0PWy irkRFn1UW/pa6i2CfmD7isj6iKSBH6+3uN1idyMw3LEk+0rgGwwWHaxAJMmUoytT8svtZFDb15T9 4V1rs+iFMpwp4OzbdjvNF/1PYG+vch3NKVjJ62w9yvS9tahEFlvqlBx4GMmBQtTQzDZFCaA7VB7f lpINKbLavcQq1kRXyLkcNhinyR81xeHVJr3xvEaE0LvdhS9NaoV9xCr6lJBJ1E5gXh8fGQ14It1X gexQsoo9JJ9XOn2C5kwOzVdpsAvfoB0gvSlk1p7olEpjtewWKzdSmWAdMkq7GwOYWfw44t2uGXnD tlMK8XCz150r137jmUWv3S2Z+0EJIidEdWobtIih2GneBmkGS/poJyk5H1tufTGD9oy1NRb+eHvw oAr3uyz13D4tqmbzD4x4cj/0xbulJCsKQT3prG9XywX0DXGpDaLBy9LSEfERWX/cs9y3lorTobTR QSeF+M86/Ne2hCMraZIxnH0/Ku6tlR2+NJLz/4ExyPCApMHdENaNSoIOYUR5yczoP5za316CtQoP ddYv+a6muWY/roq/hG2APsaogyNMbh4F8ckbS/9kN/zRLyAPKpLdticnI1S372G3lKUmu2+tWAA4 47V8T51zJ/Li+UwnzhZJTJY4hyioa0Z9KyjhkCv03b4vsdHkxjmCeb76uEBmNkBH7G9+iAUQKBAX S8YWSMyXfURUOqHH7oO9EzlkQNLCLLVIpBMmcG+QvLAdm7k1rM/pXLHggCNzfD9xSqBk5OIeE8Ne 5sal/4xdS0mtb3Mg7NCVhjkVLoVeo6MX4O+tWoStEseBMSC6FtRRPAECCGyr2qA+fbiWq4qsSjIK orIETJ5dCdvOOAd64IfHZTDU2WFN6is++IzDvQ1KHBFiPvpG7YzCJWrD+zGTmZL5K7NBv1zGmyHu Ztlk8bMXHAKeM4V4n4kNG7eZ4TtJY1L33HZPNDwMVxvm6KI9FQ1yVeJ87U1zU5OA+RBlr6sYk4HK Lql2A16NvWvPjDF4SwI0TROP2zU7gfat/8IOUraHDXvSjYMZS+umcQC7liXNDeQwtijKRFe7Z+1l V0wn1Z7txD8Ns78UkICk6g5PlbwpJ+WJcVTsRd2h1phyBoBU2wg/kzON/3CzdvGPvzdbZ5HhyVrB 7isXiEZM4AeyL7ckrLH0J2US8xifXtIBJhA54Xd6AHIGtDgzxTZ4iO1VpwBxaSV/mwD9r264MZ4j HUobZvVb8IeGJqCQj/sqmVAIKDtzi5+uTqKeW/8qEZ/tZixg6BTFX8Lw6lakTT3XaydrOkXmkCRJ QrQdLguEl7mCDq85NC3JUEDtfKCAiYdC6shnYowvpNE5C2rIEQTszAPmfZdZl5r7VEQD1fUomyGn Zlz8NnT9Q1lMXPUJRXUGIBnQ7hVkUgPl0k7UQhiC14bxsUGG1dM3UoTT5NxSfZg5XqhVkUvm73zu 5xEXx2GJCp73LLLGJJi7LQaKKXMHnVS/+KovFrNkyec9jWSTbVYJjucvlkjWJh7j7BYrSHSqsNmZ ezHPV86c7LW8vOoxPtIFcaRknQa1N+rRY4ktXyLoSYmygZ+Xt5BsMq+ETFRuPRoUSEt7rp7VOWKX 4Uw3KHi62okCJyPljya0fLgvJvUxta+3ANw4x/jTMY7CjgCQD+u+wJsJAeDfj5O0JzSi5y1FUNlV b0UluQTaFnGPRG2c0t/3MIU+hron/x3gzaVfHU4ex7VGH/WbXwlE4CaiVHWn2FYsiG8fX0LUfLO3 TrQNyHLRJxU+5zW+ctq7ASJmHXOmMb8vzBVjQFFCTWH7zx/CtyPA5gFjDRDlK8h4U1afZIje3VRt EwjUqGLTpLpQtnFqC5GodXO5cv2D2F+mezgpUsesyYphQGv7qV0QEhFDLsgli267z5U8Nu0ZyHuv 54soow/6R+eRe2OzGCnmi7/z847WMykfwS5pXQ2ylIHSmOMxJqqjXpEUJBh/0eIspv1Tji9NtOF4 kGBRdpWfdD4RPhUQE9ZYHc/wxg9OrlxgW3QUBNU8Jio7MHkeCWk+gCFYldP8jK2BMpgIXp10U/As YLonqcQJF+40DS3tLVpciGsX08+8MgvXTXZPW1beZlOVd1+MW1T+4VylUIFO9DPfbxfbDVKNGFS+ OXXoykQVHty3P1/01cTCWBxhPcuBEdUVGdYlg8FXoSVr5LCxfCR+OjGxTAhzpsf9JcS+QHctsNHp DMtLXXdcU2ZTxFKBHMgD5I83fg7ZqaHO7DJ8oNOkLEBMwv2zOvDpnrwL+1W6GwsLrMlX3fCD302b ojml8vFQN0KBcO6RNz2YgYFRKrJoBSGBsFBM+ek3fOIehOC9mMcs63lXqh4wJ2BZvt8E1CcAky8m B2fHXf0AyFxN/o5F7L9r3hOxKRR6fvA2qfLP5TINp2V0b1uW9impuhbB70zPLJrRskbASIK6lPEc jcIPoPMc+VM6XuG1bHkq+beP3c/5x5I89waSJRHCYGStREuI5FspXl4brDQO6FB9hX1t8Y7KdX/m 2dJLU+tUM5nX2Sw1xrGURUQLUvpJ+jPyWA+amb0J17A5j9mmYosDRtykxPkpi5ydHmqlAJweZafm lPURNynLsyNwOopSc+cXSxi23LJoz6cVHdVoMSSTHnkRk9efvx64dOe4ukrqolMns6k5hP6A1lj9 hDWy9PwtBrtWdI9T3EfLbU6EdgkAVZRFh1j565r5ElT9wpFnG49oMNOgfbm/ggSL2baV4bw+jwU/ dXvnSqN+fLAMmGj0xDTEzIR0Tjwu2Wf3ZAeiY8pUT66WhV5ZmULXNJUu1fbf+NhfoYUwwkXhmvfU 1CdqdGnZU8CZrqQTneeG4MFFgm00ZivLwl6tEEpCh2nP82M5f/N4FJr4eO59rp/KyJ326dypCeCv 5ChElHBgMivKZceABBTR6VUDs7WgKc1nbei1vup3+uyWRLrFCknDjODAMRJtNkNBwDOxI0tSiJBs VSekdylJmXW6oBbuhelc3tv8Iq9u8m3uc4yaZrz1KGiL+RGfhmtyi9Rr4zzMld5S9EwAVCMzH/62 CDzwGOqeLlEIDRefrJLLSmwi8Nta10uyedGGdstAxpMxPiZxvo9CljR1FsUUYWAn4gXfqYxjRho8 Wvw5c3UD0xYR3wwsjw4I7ZCzvyKe6dmcI76sXucWOpW/QgBk59G/7EDWimWDHH4zH/GNRfNC1Um9 RM/4PnD/eby7b3dD1tlCzP65qRmbhSsJMxpSpdI53KO3LsmGqLjpwIcXmxdw9bdPla0VRfGfzzMc 6Yg60fr0ZLtkSjviRQyFoplF6Pz/L7TkQke8qfY9gnPivbu6BZEt63PeXua5DmwAOnkSLK91Knrr eDUS/qgkj+1cGvaMJna69v3ctljqiGyjAm8Dzvowu+uuXrhxfM6OlQyWZ8Z9+zayJMr+U4gOBIdK 2wK9wt3+94mAETJnx5fJ541Aqzgo7UfN21n+wJoiQTeorKqhkQNT4HXhwwsF3mweKroYiejPPl3e 2LkKZKl6L9FHKevdnFFBcm5FJX2Q5anZ72//m9LY8KxwzTxX2yzv8+xpVGvFNz/TR/zEc7da7uE3 eQ6UD75kmCKBbl88sCWG1gvJFohxhZLt/6hgvkEPY6SRzHqY9lKm3NBjnRd6G/kVvu47ta6GyPpb +ZSJEYYgbrALe70TxOxOT58mAPKjsZCanPKO63T8gO2ui9PmvpRoOUdqTiJEbEFMu9ZNJiQWmfSD 4AGXEu2kIJqgT3J+rHXdv9L4FvabjRQ2JvH+MmZkdtL3Gft2p8qrncm9m2Cr/dzXSliCkvbW+KjU Y82llilA263w8HPseNdEqBLuOKz9QL8yXdRGYZ+tSZwvXN8I5AkVQcH94cB7TsviZzcEWjh4TCa7 g420gDRcwyrtnPnmGLvxy1iY4CgMSIPSIRIyoBPsV73mGgXvB5y83X1fdQnBvp6JgQjWsw8btLA1 8iKX1oMLIn6/7NS2g6zB3zTV9wZp4qXxxNbRHBGHBKpXK7BYpyViFClF5TbfGeexesSIpZvZR2fg g2XhR9AH3aurZDVyPwtFBTvN+7BdYxxk/nijA18fb3ZsFTSBpRjUHF0y7dRnl78UkWw2Zs568Z0y +GjJY6+QXO2ePbOYza+rzeaye6SN9rDMHQJgm/jyBDlu4rSvg7BSjm7tfs+7Pf/dmWr+g0qGJuZF Zp8IhrwR4Je+N5WgC8GCcjhOFJusJwpkAog0lZBzB0ccLFJtbjk+0WZb6+tfv5psIAqlo30vga7U DkZpyiw4gBj+djQh4q41LhEEw2GhIF6mfGtLd/+65ISocT0ACtIXkKkfdgFQ6egDhdA7MliU82N8 ewtFxbgg6aGrmCNS+qv7ERP115LS5mDxagPrkPpl90okPfxJJ9CpeiYuA7shHNdE6sCwOZcGPc+P WHEA0mtmnpTcfQ/ToPpQ1y0yUjWIkW3k+3SRGEdU9BqP5DPvML8kLRi0i0FL/IcWK1V9QaiHAKAW K13kIw5wumkmb3ChLDPENt8DnB6fvnWiu9P1UJskublXJVN8Fz1xQaEUDrtjNoPsieNk6e2mK5hK eqnybuqS6fXwc/BVdDPeenrihzVoVVBkhTDOhoCwV6bQc6Ti1yu8huqUvO7hovGPe9cgolbww5rx 0JvnOpN+xPLLRelGRB32JgWp2gSwRGgJ8+UrJDwZMT6uOpYB7JdVyVMTZfaFeR1W7dmQbk6L8dFk NzHtl1Uyufr1vUrs3KVk3RcP22RME9ZNVjfCpebhEhGRIwM7aJQvKsdsnXh8e/sUfeQVeYgdeoBP JAkSy9hAnFs9lL6PXp1DgjtBUCZHU6YAoU/CaKzANpnOvIHZUxmkHFMeDyZI0R5RGGwQYdx5MJIJ Xtu3f1TUAC3JNdyEmnXsI5nVmA3cVVMO47ji+2FYun1M1xAUnv2XcvlqGhkgq62tC42d7zFI+uxo L7h2BDK+pRt/XnisSS9kX9wmr8WaIh9voyzHoD51DedlYOb11RlFEy6142ZuKSUq9Zy7tI+257Jq H4nDc9h3IG9IG6P+OR1KoU0LhDEdjsjVR9gEZOi9EwP3xQjUz/Eg+TvEqwqNa+4qcSa649SCptZh pF8xe/RtJghSHtO2QBZFj0oqj0eA57Aaay0wgSVvL/3BlmwX1ecKNLLRb/Y+6hMYdKwU2KnFxFA0 Yu8b7/m4leKWNfD7dM+dEwp8Zw7KoeY6ekEH1zOi1eHne2zzVIAb87jtJD2IM8L5hfxitulfjYVO js3oyVM5lZk29njzqKhSu5PArwoOepOsjqs0DDd9Z5T/2xvXlKr+fLY2TxjNlyFA3pg30DAnstoI TLZllDl9InzABZ93yvCdDEo5r2vsBb+eomS/BXNPnZxc7AWPv8Ydao5M1UqNYPDTI/UgvUTTypAU 8PYHfw85p+tUVDOhId+mmdk0X43BShEapFg7+etEuA9O8zThW5cFUt/x+Hqea3TAb3vsMqW/gBsf asd/7F8XSRK9sxwnSM7F+pmS8myYfGzXpJGOvv3G40K3g+6dFPLcTwCcMIQuRRCYs6UMP5QGS/U9 tceWxXrPgsn29FhefHP67NgZ3z4LXoPmRARaZACIJxeQD0r0MK2/u5SzBMyelEsyxSF9XAFl6M3z iZdW8PLbMdx5PO4Tl8XT48uDPFmNcCN7jqzRtNxAHShALg/52fhzuMXaP/fI+cKCGY9K3e13OeOD HM23ie9CZTEQYU47i46yiWsPrUa16PRieeGGr11j7APRGEVHKjXdSafQNzlXYzuOZD6KGu98OOkP 7MS9SZU1vPyqEBCJBnf63M3Pdkw4phYtpuJik31EZ5HP/H+di2JNx6c5Emvb1bRn/Nml276J231U 50bSCvU7o8fVrXYfVxxBqPZWJc0A0vS9ghDRPKLkewgv9+s1YQJ+yhzGXvh3NnffCS/bZA59jNA/ RU2KuWrit7YiMw5TlmWzy/+TsL8ZgDMOldSUgNR4grgrJHgfY3Rzegaxw8rxR5mTacC2zFQfTYNI dCnClfQYt7havAKhf//zuGf7Qqxz3LXRk3IU54oyE4FnsXQnnhbcEh9GCq07I0InowqCytUbo6ui eqiEttHOHV6TZkVeRrYkPpk+J3fN0g45syQheOsT+WaAgfcsCpUpb3my2SZ8t75gJaF3+ZhiGZVz PuWwcrsjBU0NCmVuZHN0cmVhbQ1lbmRvYmoNNDEzIDAgb2JqPDwvU3RlbVYgMC9Gb250TmFtZS9G QU1PRUIrV2luZ2RpbmdzLVJlZ3VsYXIvRm9udFN0cmV0Y2gvTm9ybWFsL0ZvbnRGaWxlMiA0MTIg MCBSL0ZvbnRXZWlnaHQgNDAwL0ZsYWdzIDQvRGVzY2VudCAtMjEwL0ZvbnRCQm94WzAgLTIxMSAx MzU5IDg5OV0vQXNjZW50IDg5OC9Gb250RmFtaWx5KFX3vRAOKwYRpCkvQ2FwSGVpZ2h0IDAvVHlw ZS9Gb250RGVzY3JpcHRvci9JdGFsaWNBbmdsZSAwPj4NZW5kb2JqDTQxNCAwIG9iajw8L1N1YnR5 cGUvQ0lERm9udFR5cGUyL0ZvbnREZXNjcmlwdG9yIDQxMyAwIFIvQmFzZUZvbnQvRkFNT0VCK1dp bmdkaW5ncy1SZWd1bGFyL1dbMTMxWzQ1N11dL0NJRFN5c3RlbUluZm88PC9TdXBwbGVtZW50IDAv T3JkZXJpbmcoeujAeDoRTrspL1JlZ2lzdHJ5KHLoynQrKT4+L0RXIDEwMDAvVHlwZS9Gb250Pj4N ZW5kb2JqDTEgMCBvYmo8PC9Dcm9wQm94WzAgMCA2MTIgNzkyXS9QYXJlbnQgMTQ4IDAgUi9TdHJ1 Y3RQYXJlbnRzIDEvQ29udGVudHMgMiAwIFIvUm90YXRlIDAvTWVkaWFCb3hbMCAwIDYxMiA3OTJd L1Jlc291cmNlczw8L1hPYmplY3Q8PC9JbTMwIDMgMCBSL0ltMzEgNCAwIFIvSW0xMCA1IDAgUi9J bTExIDYgMCBSL0ltMTIgNyAwIFIvSW0xMyA4IDAgUi9JbTE0IDkgMCBSL0ltMTUgMTAgMCBSL0lt MTYgMTEgMCBSL0ltMTcgMTIgMCBSL0ltMTggMTMgMCBSL0ltMTkgMTQgMCBSL0ltMCAxNSAwIFIv SW0xIDE2IDAgUi9JbTIgMTcgMCBSL0ltMyAxOCAwIFIvSW00IDE5IDAgUi9JbTUgMjAgMCBSL0lt NiAyMSAwIFIvSW03IDIyIDAgUi9JbTggMjMgMCBSL0ltOSAyNCAwIFIvSW0yMCAyNSAwIFIvSW0y MSAyNiAwIFIvSW0yMiAyNyAwIFIvSW0yMyAyOCAwIFIvSW0yNCAyOSAwIFIvSW0yNSAzMCAwIFIv SW0yNiAzMSAwIFIvSW0yNyAzMiAwIFIvSW0yOCAzMyAwIFIvSW0yOSAzNCAwIFI+Pi9Db2xvclNw YWNlPDwvQ1MwIDE0MSAwIFIvQ1MxIDE0MSAwIFI+Pi9Gb250PDwvVFQwIDM3NiAwIFIvVFQxIDM3 NCAwIFIvVFQyIDM3NSAwIFIvVFQzIDM3MyAwIFI+Pi9Qcm9jU2V0Wy9QREYvVGV4dC9JbWFnZUJd L1Byb3BlcnRpZXM8PC9NQzAgMzg5IDAgUj4+L0V4dEdTdGF0ZTw8L0dTMCAzODEgMCBSPj4vUGF0 dGVybjw8L1AwIDM5IDAgUi9QMSA0NCAwIFI+Pj4+L1R5cGUvUGFnZT4+DWVuZG9iag0yIDAgb2Jq PDwvTGVuZ3RoIDExOTAyL0ZpbHRlci9GbGF0ZURlY29kZT4+c3RyZWFtDQrkYbVrjY2O/wZ9fUlv FDkgMvGxjdLI1UNykyT2/WKtSyC6YsfVHLWMz7QEGIzE/I9dLqe7sGU7sU1DEnTUetz96+//9tCf he0GNR4yG9sDuM2LSOkg1Ppoljmcs5wrBdNm7+rhtJzoqK9sgXJXoKjrt1Er8gLUgKngzROzY8cX tqzJZ+ItqE9XW+E+M8n/kLTE/XrCEC6OTb3CUUNLytQWbcJPqRPugpPB1BXCjzY9ppObX/RlhKyq CF4fVi/8u7Pydt81YMmdlJhAsIrNlv3ZX9SYrKEmiWfnVHSK0SXbrdRROeJobpUyiwOiAh/8cile 4Xg2ewnzgFBkmmGvk0NFP+2OcZZtqMtmUkCja6aRKAoM6CwKmAgjd/RVpaH+/VtZDeK4XaIhiJwX KbQpTVSmKhpAmIexWQrKYrwJhNV790XWdR78wfPe1GDXuXYnk3+mtp/Gx0R0etpieqiVwQnd3sN/ 32hEacDVlSb7kl+sO8X/i/+LSBmEwfDKoibQSF8gQmtQ9onGbjq0MVa4nPMX1Ki+rtVF9TbYSMAg NLscWRJTr4ZONhFrYYb65bld9zw5o0UF4hkIBhaYbRs736+NmEK8Rpowj1P6N2Jl5WRtT8rQVQds 2NElFWtaOAh6l8WIlNeTk+cXqIhXnpGEEughQEg15N9/eot1+Ino5Mj4ZjYvmlGhW+u1xfO0+rRC F2fTO7SwrmsOmsIB06WyTFPCnUXplOLMWvS6gQbalD1f5rg0w4Ws1qAdnyftCFONhPc2b1/6ViSj h25jGoz2MNAuj7F/O+Fbj70d645qgyWoxQcqfD5tFlewQ7sb0OpxosnL/VXhr8b/AOxQKR4S7eKB 3T4eF35tgryLeOVbUutZ/3uL9CAp3TES4X3Yu0453L9Ul4Yzv2WlZ+mTz5E5NtwvuQ2OzGPjzqKI fDQ8KY1Tt1zkFIB56o97N1KjIp9/uslnXoitGTjubW3TzxEEEwY3d4T808KdlyEO81DF1nacQQTf Bd1U449EYyQkwma3crsGtoGlI07ULTiOtraBDs139YdsE913StLSt77tacDcJLcf+J3PSgZyRDvk DfkOXv3JnYCHh9+tAXz2FHVYEyUwxKXBhQiHEVpuCA6Ko5/n1+4rk/c/BcL8TVJbCoe50qf6Lkkh 4y6Ib4rDB3GglpQTlpYHVeiBs0y+M+sdrdQCWISa3svrWsYTCFeyuGODs7mnF2UhLcIu0Jc4mT3c JqZkIFkHG76mF69kxfm4Lq1MhuQllb2bykNe27Z01ogeu/L7oO1c5ZeDRgug81n32W2ZGwA8H3LE 34BkcJRBFpHHbrqkHeK3RA81EUKSBLKVp+EwI2KzsbcIO1R24EHVCpd4xRIAq1p0XeMhkhSOS+Oi skoMG0gL9zs66jyZpk+7B9xHtQxdXnqn++mn2HWV/dwMAPVVsNodNPZnZ4eTemzGsXiE5QImvUrH a87E357hL91v5tOSr9mMjaWjv6vTWtZS3h3U0c+58OI+ev/rdHeYit2bevlnRAhDcQaCP4mv8y0b rNEvD+zC8XxzAFXugwfuOuTmwRvPGByyU7QSz7NVMCEGfTYhmgpksxlWR6V4y6GoASRm11zeiTK3 hL0JZK4fFmb2I67zTgHGeFtlKRvpvc/NJ5QhBwiRKjN78lgjvXQT0nYyIGPDRnRM229xSoQNIIsP zT2aMHfRVghUUPXHWmpFfNNgiC6VbR9gBp324BKEj4iOUecp1RjK2WpLPZs/hlm60DNlLrctiOjT CTFaMR1+yZDsdFsM6YyDzTKwcJwbm3sWWc3/jTq4/skjN5q3Koup+WtELoSXzRmEJtVaxjey3toD 6+zQVmjxEK6Uln+vi7lx5VRDEdWi4XEmxoLdMsJiHplEGi+xX85WmJ8JeUB8aJWdxzlNEs2YHAal vO3/NaFx9pQLduOwnD9p7DzhldMeukVbW20Gc5+HeyIXpTEmuMTaU1IukCPvatjmLP4qju5B6Nhx wIdUTXGoifCm3V8UQ3eHXOKxt5Z12NtOjtV8y4owtAPcK17/GatEXifqyPItShRCSte3wIuO6jyY Z6Qu3v+pq1Mr0T6rOK4u9JAP9PQIgPvveh/ezpWnjkQ81UR6UUBAdIKnVbJym7kClFHT/gI851SW ZtSL0h0UdTqWiuywhTt2U/L8CuC9AqnOCL0cR5K8KeINWmS2NzF7eoTQtQhuAepxTQa4Osd9iDg6 6M+vbczde2zHL/7X9rB/zSYuW3O1plF50TfRjYeDJ2wL8eYSg3Q6B/2Lfr/0R7kN5t7He7dR2n0y OuDge+oAxRxj74FT8ACw5hwuinAItQfq76Y92LoAwhdgrhAWj3JAW9Eke2OEUzC3YvUoYiexpepP /C0HHB0m4LCzIUg6X9rFay4mcW2kUQUMNbBEdacKsQRwJXxZqxH+IMB0aVpVSebhwtRZfwUh/3Ur cjjpX8ZzhQwK1Cx8mA8blrLkNTXCifB2IqrAyOS5hrnYux2/nhgccdK+NK8MrABfZzaUaiHja3S7 Tx6tfKGP6dMZeVXeYB/8IoVSBGB18bcKUgZt7CMD4mX0OCxRbqI9+LA6g7Rn7Vn6LmRRCJRJpW+o JLg5wV3xEs4EzUbBGOrKSOFHrEhlYACp50lyeH9WSP1U5+QsSA9EjasudYIQhS0hbD0L2wZ3yFdj yUne7dtY57JBPs22iYDVs/pMgrVClY5eke2FDWPIPAhTv0jr+W9allehAqmLiVgZBfqF0qdWl5wg lEQalCGtXZsjGlN4eq2kTHYWbcreuwrXu6JQr8T0AQPBgfjFO3nU+ObsDqQlNkt/4QGohW2tVsli 5+LHSQssLOJC2MpoOE/3lUAxyQXu5uSOtOowdepG0LQG/fWt1wkvpOGPFjAgiy1qMLzmE2yQD67c M9fyLqcToNjImyINWB0ZrgpKVM3/zlU3kwYNSv5DygGKldafzF0aq1jYE1GlTM9brJhVLmq11Phx jpZP23qoEwuz3AfKU39aKr1Ux+kOxFqoKGOaEWjdN2c3bcn1ruboHcx96EgHudSrvci+cKM/qYeD Dp/Ximc3aDZplk0AopUJnCFbuGz3vL7A8aEAkpma3enI3yH6LWFYR0OMU/SnuJj22wxCvRmetH0x goxmqBLoQnWVbcWFmyVHB+hepC5CDVRYYxmUgCC7bTuLUqRbpAE9KCQpZ4O5C2t17i/f0g39nDYR shX/lHolEEGgioW0Xl4193fI2xoV+N+BtPqGxNseyPBAfAqf9KYZQ/AEAn682xhBmngo2mXGcMs+ 0cL9niscw/HZAuom0tmFqRO92rax9pzQkoAEwUOzqLOoN5SptZuznea57zwK3YnVdRhgPgYJmYRU KKYyZw6QGJruW1GEhgmTkWTpTwbeqTiv1EK3iyJKGxKYCS0SW1RkRBkJlUfBCmfgLD78NTb3Z5Ng Muij35Dttoq3eoOoQHnV2SwRjxEvCQ2kiKwk9R7WuNfxxmP8jnhYc247d9QbUREiSF98DXI8MEub L85COPboUl9+hokg8e3fZ7Pjt/THWLUdakdiVNn6t2yqH990aAuRtoM10d2k5aOdTlp9bhL6qN/V PKE2VsfWwxgL6m2yyaWvYHk5vt4b9sD2LvJMcjjaf+15Fo31u6I9s03zRX4758HmZM7lW5QWpD/Z 7pbsEtGa1brbl+b8y4JNV+infRrLoZuKf1dWNoFArNgjICnvbmD2pkRyNrnLnraB2qVTPtazOhA0 svHP4o5IS0GttMyRYpF6OUUKPmpree2+KAcrbn0Hl6q6afBT6xXHCv+mqY9/TWtcZ/dOY6Jf40iV 2YcH3I/buA9imHW9z2l0LaG4ynwYCIBaDDUmFKTcxj4sh0h5w6M9M3RyuyDY7qZPMWV5H9cEQYKe IlQrKN6D5Se8Ts+79ifIs3aDKoYeBh5D/U/cvUH8Xg+nHAdIP/fTQN5GxIr7KKPeLs3WmK1bg23E IaiXXIsJbxLpR5qDKC9EOL0pXnyZ0o74fKu9mn2LKRRpbVFf77K35zNnBZddblP7vFIvxn3kUpEi auVKi9K6gj3WmnNh9ZmX9lhJJBtNpFFiZkhAb9P+V+uLmHwMU4okRF7QNkq44RdHCRyWABD9C5KE awT1PMj5vqL600d4Xy+avI2Sal78QODCpAty5iXm9U7Ig7g6Tn+iGJccKFRYLQm9T/Aky73e560N bp4ULU2gh+ipg+EcwLA0icSCeP4JCMGzIK6jcLvCcjvTK7th7UW94D3ZOz1actXVusBZ9ELxkJsV o1fTbdhkNtOGbNhmnIf8gS8UEx4v1W6OqMHxaHrsQ98pP9UPNTc9oO8FvnqR2FJKw9gM7UtNKsMt aAlqHF6iF4NsLqXIuOyC/P/1loFhdXc//6eS5u7lwgTYZIrWSnxS8xODPF5ovAEHaZM22XE1DLJK pqLR3wTJHTtMXMPpozOca3cA0LktRfmNLtHYm2WCunWK/+pfSbG9NNl/2dWmgAbsKBz38FMk768I f/fhm1RQYMGPa8YXGAqc4y48BVKhzqr++SStgkhCVLbqA6i5Uzg18YFbKHNyf4R48lmihOWfLKjV sFtKKW9jGauGRsmZqG8F9efmuUXIFLjxlVEzm2YkfGl3vzg1WgDaj5xk3a5/GKlfrMQXT+nEzbWB Ax1zcW/OGsnLjt/6iagcwAYz5aZwFB3g41IS7JyTajcCVXPcxMv3L733SdbowG+bxi1t+D6dxjUr 8Y7Efn70bpyePnNDd2AoTrY+lyqKUODuEU8tOxAl1VQa1U0AfBvYT+He4JzjIhTr0/m4c8TkLqCc 0U0MrGHsmOm/7BwNro9JbC45JsrFj3uNSCm8KH/34xmPXE12QJSLZZNfnHQlHcExCOygow4Aasc7 F74Kk6VJtZY1vo6hMzEKyDR9R6Xp8WPmAyF90U1M3QZyx3TffqGV/p7p2SpCUgUcXk69gn/bVs4W mA4Ep7vNlJAe8uF4vnZlKwvQZenU1/pKELYLAcIoJoWpaqHy0ZabB2C6ctqsHF+IDnowE3q13kzP VvsWveOrPAZHJp+JAne/4NUK6NuaWopxL7Uq7FwhWeamKS3XCKBRahnaqWNhj0y1r66iKjTzawxq n8qHh1hVNHSI4YtZfRsqJhgTQg0K3Cc2fHYljwXW+qFiwWfTZq0FJTv9GbcapkX8+gmOXbUGjWuZ f2tjgHZPOhHZdRWTu/lwGiO/8lT3kyjZf/mvJJO+RuZAFTvtK6Wj3/wK2rySq+zgAHYO/Zi+gxGn Uo8g7mnuqu0dSx+2bq1/FcE0vtoMGFGeAcggSiLHCFH1zSn7eRLPET/jnjifc4Jm4f9O6p0zXCNB AUVL73jMNj11MY7L/N7sWI08W/DFotEkKT404Q9b3DyVOWWYFl3M/U52PRB7A4xE6Yt7tEneCb3P 5yAWUJNEVCHjvjPqIDdbxtyWOOdSgi3zRLvmUS+yBysIM5hVL+Txj7FAG3+2ACEX9+2PGDe/BTVd a/uLsVap8ngFN3nnoIorU/XpBtVHcDYHtW768ChvHfb3WkLNmdCx3a++urk662//oPJnisJYEIM1 gic0YMnEVIuB6wpJeWt07emnmiNuF9wRqaYhsF6FrEmOFp8t8vL4IhW+qSTBgxKdM+9tP1e2erZd 8m8F3RxrtMocVv7vuSiYe9b5lUH0ealBpcs6sW5JPwq2HiWB8MG/fpUaUHLKyUm9F1CMpePaAdX4 SUiJs+ByryLF/r0reDxZgrEpDL4BtDIoxrEcd9qtmDXvqhlylm+saiYuVGOO+JeQaxnlj0sbTsae eN5GVIsUt3ZpAjVGJO/HAs3gks3lOisWrcxOY/key07Mi2AAu0f5cxsJRxLJExY8CntRlV4WjtkM Lr/GrE7HMfjqALcJcsEMG7gKHD+MUem/oz19RygSdHD7V1olQT5rV5jdWa7lEqM24/EEcTj8EMNU xlfQpAhki2/cU45YMJs1xVeH/SZDC2jnMYI+qDM46FR/VSxSFL2EcL6W2RKJnjFFaJ8tOw47H5oK DKSM5+P1j05ZdmRvxZtl2s7JeVYEzDIGI/3d7aqoZFWAcy9wLB1yjI4B0VL0FSXWenCHGvh72Jpl Kw/ocQV0l/fzbtL8nLcudCJ55Kb8Im5BWwgowYKnq5bbSU8AgKtVce8vbGB2b47tj4yGBja2xEca +fEIJKceC9ZeMUVoqrdv0Sf6FA9viQDGRl0AB3UjM3tMwS4hKV8xKntEIjLlwLsGEkg85FxLzWFD /I9jjpaLs4KDH1WWx4l0PSi/Fzmzd9V/HSPzBLx4dknd7nMhPBtCu6hav/Q733n7h1E5OAZF3a7u Uyb6M0MJzsh02O6TK6kpheAxdh+5cgvQsSNiXrXM6jgsUbZ6Oqi6V2TRu+WYYuJ1FEerCpoD5K0f FkbR1pPaCk+kk7B04zlhj8NElLNk+nhd/uzudckUtNbkv+ChAmrM1WEfFoJQ3/eAXoWaZMcO92f4 Fzdux+exRhHp/xYdPsBLJoVshyUZtmesaZ/wsxoeXZCB9hcicx2XWo5NckWwgmzx9cWv1kYOST18 RS9Oxd8bKNXrUdyaRc69HdxcMVwLdpXXxgy3q/5BbMG4jxTjU08LeKJIOKYm8JW8HgA5tQOoWj2n UryoZsY8Y1i9Lt/dnM//aSm+XdKvLl+tR2Ord0eiQKVxvG5dABmE/vu8EccM6SkdAZWdt7ff1c1l 5VNBqu7o6H+ntXqTtERFFx14hiWG5T6OCHaCsul6mXv2S3mfmOatBkyoN7xqEkrwVgV5h3F3W0wU qHyWR5/fO4/JXn2rFfFuZGStpbdNKb2gguAUsrw4GVixg0LWlFZWcUlHVKFo5q5lJLdTv5WP6kw+ oXv4aqs/p6NkCCo7pPgil/jSVqITQ5b0d7XsoJnanci4F2kSbBMAqH0LvO591aGD1MDAJtGKvLOd +59LHFeUOVJfuDb421mnHidBB4SS5SvlqnJ3jwbhoGxbjl32dUOvztwmFz4c4EhViReMCOS5pWI8 q8RH7DoJ0GcVfzYlz3Al9LG6h94NiOPW+awRSWz0zqbbG9DMy12qGpswL7Y3NEESxmVhFuAmJrIc AJXD+n/+yw/Dp5sxVkbfP78c8DYwLT6LUQTGWks5W1LJ2CSIIiO42eDg3+hVa/e4wiwKMAgQXeUH tHLB5YanwKy4NOsGqhAGVEF/Hw/JKi3m+qVl8wKFAPUW0VlK19OWXnT6mNa83eQsrTrNu3W8523q hJ7knU5AZaRr/HKt3pQLeNC60blF4xZYR8LJqExWrlRO0d5OYhLHJWWHqla+aNalakqYw9+1Qaq9 2TN7IK95FTSeXSuhbGbYzbbimfVM8XqMkbQORvAjcoYEdb7/LL7ptq4ySrg5Djq36ztkV3sA1NHO RQDyxxbWwsdX8FsgDjZ+W8pGn1w6RKWFchTnGnXOb15iF/0NUNNEMT0KA1jE2nwYe+SzpAaJN3pg V9TZ9yauX+AeBSFLD22ZNNpqGJqQsJ16e48iWGupBYEyG/QK65gFpnIDMQYDXKCmA7/nn1cmKS77 GH1jbGuAignpokFLHBf5T8PKn3BN1MUEK8ek71Xn91ysPZgXBtvmH72n08gsgG3RYjt7GWI0XRkn uXH+Ry2GTTfO0zht+E0Vmyz+tL6o0fEEc7SGLIbgTqlGTh/awMbxR3Mjg3Go6qIzk6zDM+r99sbI bMTB9d8+ctOEGXvKm8nEH6+J4c2KlEuWos+tr0GnCj3sLlkIc6+mcN9o1lJeC1KK7tRjdaN1ee/T K2nGcXuVb7Rzo1D5EoWAJMtIhApxMREsSXv0JjEGdZporzRvyg6ljZ1/FuJjc0Cl3vZHRit3n0jn pWLUfRx6WU5m4Lf/IjZ6YlfGYN3k0HYb2g0EArsa0fkv3y+bn8b8/dV061ECEeSTKXLzMZuh4MJL mT5HeEEeHOvZubR8OhBqz4Yxo1DyPVZzP38kdlwfqKgAqDWB34k2apmXCgF9ii89O6IDnayjPFi5 Hz3UeziyHVmGxTvcBPbK5AQPSf8YWwmTiTo/D8dsbp7qE/gURoTQYt8+fT+i/F/NRimQ2QKrkhsK DZV5wLxMIQjuRzCrqIT30f+ae8GBd/gvwPBbd81zUgXG8xhTtOik/afLU1k4r7gkwZnTZ2qchoRB oILDuZm44FJ2G16e4SrJL890STcGwLkHQTfmcKgqPXePQ35SgzDCHazWNb2TCLwHCg4g5aqEu6In vrjl/jGEXc+jHeqzT59qtzG1voADNbOPO+Er5DGLKH3zh08IeGZLjGRxq5rPMslH2qqk9EHqzaTU sGtBcdByT19LvyFSWt0bd7g2qBYd3R9V15YPhc0WhVHcj6K1tzzayIfmpGuifRi/M9A/ANobqFb8 MJitQxAt60j8XI4OLFvpK90SGFiIESuC3Wv8Ekby/K4aaPon/I+SRIVgluvoE9rUl1iVYerPd2Gw wz1h+2R3hwjS6wL0eJrT6GsOlUmM0vanqR0ZPvERUrPJWIuSoTLLDfagrYRjRxClowlhW+naVpAh o24Oc2A5uOlrDeh571cUAR31CYjH/PyzWMMtD2c0skoAWFfrA9ozgdk/Z9fHtz5evWnyIuTuEr16 6tom+nSopT88v7OrjbPNJL83OJW77yGzMqny7CM1zlRTijOH+8+C1b5ghWfuV/OnS2Lj3z17RBXV uvHQly53eHIH1qz92fVWlmCrg5pUX4pqkq3/l4e5PSiHhcvze/Tv1ypXEmoG3+XuknwCFQMNaFBV +ME3FDj4V+bKR9vTmK897mT/6ivKT9OcoSNkqeZTfV/lqcvNkgZB0ZLugVjbM+YAZex1u4KWkQgP kKb1nJe6pPekS6ZTfNJ9GRWBCe/KdwKbECQwTEQqsv+mHLrwA76qECEZVtt3A/HH6TOHXG7cvQ3t v7MtNAKS7sBHNZ8QyAjfoTSunU4Pvad35dslc1vtyJrPM21JmE66kaLjbrGfZ4jjsSp7VCCCvMXu mOAygiC7+Ac0D04AanyoMQYE8CE/gpw1W6eq20s19GovGABfs4FgKIZ0Wwc1WwGjo1UxJwDrm6NF Wh7d/8MkaMLhO1T3eOUhqbE7Rk4caVnVhJza7G2HEJyt3ME2jU5U/ism6c0EJ9T1xxaVxMzmJSn6 SKaewuRyCog/gPzAPGeDtSmxGF5B2RmL8gGGajXLsptBAH5SugiF2/bFP1RBEUmPDNHWhqgEUk/g KlUORJ09Sg+sCLfeoQXPkTXpqyU07Aqm4XOskrEaHK27sT3wbRJlNa8QT+ADrAWOZ9jigg1P7LyY fgMgfygiUFWap2JhdjMPSBiVM5oyKkzNflzLT7aOcXIZQpEZAfzY5FJyGI0ub4NN1QKtiqUSs49T zgBBHWF4TAmPK+iPiEOj2n1YjXr7VB/0PvWfVL1e6qhDDJDOQe8bhGrHEWYoG2NEXrHdfsgatWIL mweVsAx5iwrICd3iq6cucpP7MdYUt+z057ah8GWtnsnrKkFG3KGsKcj/mheb/2watJWgQibgK564 dkiUo6rGdPNuuFzrszjrqhRuIL1RXkZCT67F+SypeC8LWJWP77gssgqPnc5HSzQbUlxszjp5ZBoj rmjYZX9ihtTzGbmRjrMT8r+j9xQW8+xtGZn4JEFrhAJ/UGiuFjUnWEpJwH5+3WGFECh8qsR27GZl hCLXig/X7Kc578e3ZG1IR4xYyl0dgLhYWpv1oOQWm/3H0X5f9nPjo8DoE3ma5iN2I+SJmTdqqKOw Bro5mBgKrLtuqeZy+ELiReK/g1GhjDCOP+8Zu3fcmxqzNSehw//jKMGGThzFl6ELIeM7H4T9P+hY QaYPSzIa3KT7R2mEbfWGpKsv8mN80RfFJ2b0Galarq4z4WpY6aGtbyqCgHm3PSpivbicOjIAggkX jbBO+Jyt5my2JyGHTCTdP7P6Es16uWmmzGe/TzVp824Ed+2+Rsm2R6n3Us394Ld7zWYLILVPgLjN 2rqIzuFQMUd4RLJnbM+xT0BIgnfcRlXw4EA1BlQAS2BQyTKXTtuaXnFwpeOMz6JNyt+t1OOK3N8K Tl7RWGFcLalQ4tzDHHfoz5FePZ+EB9uUY39hqKE8tKNOuk7iBOflWLgAsnCOGg9ZxcvOZv4IFkqI eTtO4K7KUOiMV6Tr/FZXfS475Ah/hxbJVvL7pZ+yKQrJpnsXT/VrdJVWYC22e5iqL99x8hN2fLan u3qW2ua12Kt0JgwO3zfDMHUM1ruA97xv+OITbPtNwSfYihIuVBS8Ly8NxCL4AV4vbuEbCBDAhi59 9s6X0OgjUbeI3/ATypBXhgd+UCx+R5Ox7wt8H8TrfMOhcfkPidxPKo4aZJxBNYWeXzNbI0l71Phk Yx0c1TcBxgz2Osyisod2hlvo64vMwDDD6vIAvGri07rwYAO21H+0oSICxHm2ingg7UPo2gliiHqq MTRxL09j9xGGevppZNL8ZCu7XjlL8yphY9YAEbKB6Gr1H25mwf5FxuWyn9cqDJN4qLuTsUTQKYdS s8YbfPhPNoWjbDQ81pIGousblo/7L02J21pBvougBIgisJS6GyrYHC77/koBbeE+YUQC/7KrKR4P bip3IeS4OwX0KQBU38gsq/Fh20LRQAQcyOh+BHbQP+wV9rDCHXUiNojFo4HN6AghoAzeUQYsQ0Tf GC4/rJsdMAi/4UmecRhIjtPgopKxqlUO08cDStJIXdmFkxvjCR+gRETmdcRUnZeewPx2JBM/+I+G ipcO0HzMRpMzFX1gSfvJoH/Cpxirj3Y0yR2Kj/b7Sr+0sL/mpdz5aTIsFYBzij2KNjGOYcA3Ebxc klxNfpMlkZyXNQEu92pswtTe226VqlveHiExUDUc2nAL8mHi3lizV7Xd/M3Y87hgsa9Qboiuqntx XjVtgsjWGm+kJiepkfYaUv+cTGhlUld2vHreMu7ykniIJh2DQuP+dyiWjHpXHOaJx0yt7aveDn84 bUAyuv/ziEqzMUf0Ij/0Z7dsHz4qyEOPFKfeDHunxOxI6fiODm3vjj57lipE0JFhPLrFxlon40No jeukJmz51WxTMZ5iDPax7sdToN1lDMyR0UElLIbEUl54K4QMB0uCULs8B/5rQL5y0jcIZVWEsfHR 7LYw7XDLOHByGFXjSUnoPll3nn2yjS9kT0gvLDFFqr4Q/414N7sLU/kjOwa3ia+UabV2DD1Cd0rz ptDheMrD6+iqTSfVATmTuKj/GSjqadrYlXkRSMzn2c2YstCYZOY9Beb4B6IB+BicMIhTY0CC7SLY 59G51kyKqBaPbW5hqYqOhc0/FEEWDTrFmF8pDv2ioyH0026fVJYt9zxSzyKy/doJfU3tYMHxR/+C 0fu/dNv5NIQ/MDMQVziSAzcvjT3o2NZ8CluQojQlNn+C6kNzZ5OSceYhcstJRqvWE1nXswjarNun EGEbj7w8V1kDsRjrsxUt5nPc5Qn5A5dJaJgHyW7N2ZVBE3P/hzi8TFw7mR5xDdPXk+2GkOEJpJkS B9HUVAMgGvUxli/egRWMnvb5cG/SL7rwFfgvw41OVGqXOn2Tsds7NWSPdaIUYr5ic0OrRYMSMrB5 OCjR99H9LZAib3BqHUR1Qj7Yz4IQ0uiOEGXKtUek+MUdeC4AdiQs3fjsXmkxVkQpX9NZgfReg+yE tAB7RN8KW6OtuUeXqWqMeWCUdeqWnJphJlcHnc15fuBJ3sw6cQlWjnLwDsR8xEvhU9BWyWPPKNVc 1YQDKX+dG9csyEvDlwEv04Wh4lz+peUUfaMUStP53vtAeueFddt0ScSCMF8/g2nNTo/yHVd1SCNH wocLBiXYPfn052pC3A/J1EONzKCehoqioY2KToEf81ATDYKxuzcw24bgCLnrYXa6Imk6JCxnT3Pk 5mJY3z/t91o16Yh/e1GbmTJvJ0FhZ4uRHADLUjCNnMJxmFvfdypf6b4Cc8zE2zwNGQI9Wxmb8ZKo MZtdjxy6gDEvq0G4sNMgfPiVbbHD+xin3bXQSvo2CH5GBTl9zzj7QSYMYkYoxi+YQTwlkyMufjlO PC/xhtchPdi2WanCIaYyPZsrPM8+CanXr6AClncueBE8BtL6bimYw/qmSrzI+rQLqVz5kcwBalsT Bu/b4BTdxuGrQPkHxxxUnvLzYs5grZiL3S5HTRGNH/Rk28Iix4Rh3a+aOwxtNYB92wIUb4yoMO/6 jb4KZFwrLW6xdNqEsW2MSriGffp4mKqUkR59Ik/ciVf4Ro9R4JOdY/GoCDMLbuOY95ZZ1eIBtAQD 2cFZ+QiVKA6lfFUbv7rGsJuIDWUwkS5A04augyN7S5arX9TynzF+Vygv8np+GGxnTRT9YwzkIEj7 Fn/lR1s3GtMhVpMdbMsLf/eCIMnLn24B/lIhzMDcCNEpRDEp6xDccdgcfVYFfI9ItPZde7O+yuWA o1ncTFXtd25x+wqH7grw64gy5VqVhHrkRfEpnReRjD6Q8zFHxFMRSHW25uj0+bRzOKnBTTaKCrc5 z+drdJAvIuTqXI2Czqr40WhoM/CZiUVTMNiUMJMmvuND04Ax0tQdOToNgDjNzSoMNRxcYuYXf8Yw mmSkJZMkgqFwdZPkrZJrP9ilFwh9ZXQ6OlFiFu0iv0UJbFAoIk+wt9eODHjzOrLLg162BCqVaXDq 9r6lcyzU3JCb0DhE6mIDnxVudnMP0BdFkXi8ocyXlOjkPI5SqJHS/Dh8AF8LnXRsu5suiQS3fLJ2 4oATOu/fU1pLsb7QM+pkR7nN5g9UN/It11FiGCSei3ikq2+bJX1i9BTHY4urk+WtGReePAQbawGm MDZckIr7i4WfJ5+3PePOj/7lzUi9uJ771lhWw2ONYG1dvQgF6WfieHxFMyN4NKlidRkyTEAhAYZ4 UAIi/R/9ucmCx7PzX/lbZDEt+P16cy/3cyWjM0Oo1fge9KgzcO1DW6wkkzcR8yqV+uOE4Jcyjqre CMOjvtnErbOxZjMQqlaPwkDO8wDOp+oNXslbAKcIa93zIBK1hgqfKAN58Zi6FJnMZfP2QEp7WaDj Lm/xLOfzpG2wTC+VZFehzHd3p2JcHupsv0jQvZBAlApylGm5LsWOoL+81g3CFIm8oMOrnmqbojTO CpH4DU8tLPZhjXhu+rtqo2zlcR958sQkJ8u5aYoFpUDD+EpzwUHJ6Deh8F4q9lwHugCEdRfXQU9j YiM6JAnXE8I6HV8oVRenJIEuKD7I51dhKOw+g99HQwFy+rwXRv6l8h1Z6wOGAxjdVusbiS35MCSd hZG3LPA01y1pRnW4yVLVTjVqenmyQ2npI2pxsKnm1dLoOGo1/WpsobhtlAMBGFgbL51H8fYrMbsh cCe0IQS3sRRBy2vvEmfI4zM+xiObRVFYjf52re/tzY3dxcji7frlRCpuDPCM+ODW7OLRPw2litw0 7VvJs3au/3G7otgW8B0aimbXzqj2+1nkfmFdj3Vg+fgK1xVd+TYbaI+tlTrC8MtATbrTds3ggxFb ECXjKffyMOOTdDflvPG2wbCK9NyUDcg4rxzTGvHnP/l78VbribKI1mdtHZ37QULIkczTyvogCNzS OXFrV08MlDk9oZJ8BG1gui3VddPtWmJhv7hVqf4W6xKflpg0BpruUpkOoivMekUNkT2nM/VKEWE8 m05STIqwWVuNj6aiVgvyVuFjAE3Iq1XXboTIgPc1s8BRSP5eEJz8gg6c0zeVuQozypOkB3Q7eTd5 yaB8Db5s0fZ7W/qUeOmUNjDANA1doIvmsIYVnZGXmrRUZwfrv/A739y5ZkccYfKSz1Erf+WmgLGv jtBSpIqHc+DYXj9vt50mBlrxORadA/wPXIw5bHC2829Dxpq88kMQ34opNAdqGwMdtTAPQmHuGzwS iQXvOUrxy0XqBC/NJzoGvjPWvFVNj4desVkjP5tXg97pw8JtphfPWbVyWYhi+CFo2DuW5PtQby2X K9pWDcaUObyhUTMiTSyNiYp2arjwtBevcQWk/BI8mrvKB0NELxm9isv38SgFrilDn9ywwDiGd2pq hZjv9qmGUalx6WWwdyxz9MwUouUcGgakTfWEHejX+gEYW07Td7Wr40ZFBzWYalMJR2q3J9yA5y1Y UgGfgQK2cNj5L/4ATxldaAmwqZC23LqBLFjSivqU/pG49+uBpn3+fTmRVB8chTzQ0ml3XFpYQYrd n5XDkr3y7rvb8yhAyQgNbJ2OenUVYKtoPSUHyc/Ilb8fRfYC+xF7WoQk4gY1nryFqlNaX7cXZS4z x2lSIWXRk7/FWhlINSydOYHgfwp8S8Z7UX4heeXxwHSrwYTSqzenlylwofz6INK3uuk2CmuoMpfI 53p+aBjBSROyyUHrA3agmrBwARKIk0LCCKqZmPMRUQpd0XzG5dd14uEdcrGxJsyJle7yzSMEVlm3 3sBkuOXCaDpA96ZiAxfff6XcVtNuUoXHSS1cIW54ra+HQVtyMN64XzzbReSCAhjJAV8qBDTaKU55 xhbbJj3c7SS7NS/TPJjIvErsIBs5pWsSxa51JkLs7MU/M8UcXFqbOB5HEF8rMq7uN2SE7Y+NVt+P 9HU6RfykCTxG4ZiT2FWfW8i30tbuOgaoaj8ykwlhyiCus7kS1T/lAkDQYy2OPO1r7Hpzv9Nxboim wPiGNOEDMkwH3xyuOUnvUHxwbEt+HpkfV89+img65C4eoY4kjY9yB3HPPrhNzfRjCs8o1+tzJmeh fv+Z0wcSbBeqwULVcakqU6ZEh5UI6eDUMs5nXj7IjcXlT1kDhVVkLsHsCAe5ZW/U8k6KBp4qG6k8 d75VdWED31QmIqALlwazbJX+3IHefOap176EdzSk8SsFjZnqOaQjhwPKlD58XhCvvw8wLp+MrozN 4gP+bQlMLSmpM/T9b9gbVn+hXkduZZ2OTJxOEBIakLWT4ysq6gtyOaqd844p/aJon0ffI/hcB3dT x2KeKHaLX3JV8MN3DQ4I2B9oYKEeJhY2sZByo4a/268FmrTpautEetSk+IYC45wj2eNAHN7XW/Ls euLOS7qKzxNol8ZUy2qx28O8WFT58wpD8zHbSyFzQnwSBWB952ZRkHKiHMmk42sa/j6yVkZdtJSw 8cFuzeokjNVj8/Gqg5akcER6yrw5qH9LGsA9Dc7Hr0zIjXicn4xTBoUQCFFazzCh2tGLCanpNiKQ +AdkqC5Y4uLP1dEFajMmMjcHh2jpThlZVEmYULmy1ryeYgWRJLhrVd41ooue/JQ30dHK9ZqNzp2V CEIEQKUXdTGk0bv1rej6GGoyREu1TG2Hr97y+gvUzXnMIFpGYcUOo0Pw073Rbu0t0QSlr/mTLOML f2K6OV+y2GHi64nqCnTSo8Gc6+mfYb3fFSJvaVaeFZZjvsDIODamrI+UIn1rIp+MM3D+HChBQSUd XxwJAsqXZT4NUsY+cVk7+nOf31nfN0+Yf7OzgJYCclPwxAE2QvANqdX2XH9F8EgcbEcOpLFUaD7o +102ar3LiQK9QnaAh6GF/H2Gj9GdQYWR5jfxnT0GDtiTul93uubaw1BtLLMhDz10trvX83w9u4WN 15IKrvSQGc+a/WMnNCfFarpxHmFvs11Ni4tisDU2SBs8za7l+oHUMl1PS4xFLuehl+Dtp6HtvX2V UvVLrKCislxXtC5zhMUxrmeA4IY8nEVU9fJ3znQkGL4u8IDEeYH335GiEY8dBeLqtL/aA4Q0iWyO yoeABoaj/kB9Z6V7wrLglE2/bma/4PTZx5E6A/XlLNv44+e2qK3XC7Vz9wqSpCiJQvMXA8RHnzSr Vmv9GnDt33E7n4du2F74qtYDxwPIDL9yUhlbyR1ZBcPfU0qGFZzBkiRxmMIXkpCkRHFOuPlFnkLt R9rWS0mVBpltWV190F+XYka2LksfylcjGGsm0BT9TNt1DQplbmRzdHJlYW0NZW5kb2JqDTMgMCBv Ymo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA0ODcvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdl TWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA0NDgvRGVjb2RlUGFybXM8PC9LIC0x L0NvbHVtbnMgNDQ4Pj4vSGVpZ2h0IDI4OC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KCMfn+hEQvKHF Yct6xINkXcpUlhn/vvAvythV+yIft0CPvdqxz9wOimk1CB8Sr//KWpDwne2FySTcrWfQVEGbF4N/ r19cydo028c+ylZVBEgh41aJ8EjFWBbZbHG1aVZHtJJ07/SjJvU7V2Jjy/zlm8vaPA4Nt5BA9cXp 0kio+/0DNNBYAres4HXHdELWxsz+peuRnsUubGgLVSVwOOhqQhFKJwUl3OEYHgicf1itN0lUw0ZO 8tc+85i0lm5VeIi8pXM/kAAWA1NoBNP+CuC4TSdYSoaHUuyJJeNhKXeQLyK3Gjc/5xHfMK98sApr wEke6sDYqLw8036wkbMejq1UCNh/30nVzgxyKWweYjktuH1kbFXqh70fWkoQMpdMLlAKTCkS0dBK Xn1Cve4kW9gGysooPg5yDvZaqWrnMC00HzeRFh9pddFQkbA4MxcDT3SqPMn6vyeuJfiOLgMHak5T tn5/S4PuniBRVZ4u6GaLZs3vLQzf6YVQy9NMmTmMuCY2g4rG1MhJ2SKZbwy8hIThrzXIYhBpAUAh TRaINl/pS3NAznGDpAufQ6LxM+KnrWQdnQrWyVcbRszL1G2G4I5qgURfesVNUHSkKS2zj+MQTddi zs7wS9vf69lpP6QcKppM2NXwuJLrBA0KZW5kc3RyZWFtDWVuZG9iag00IDAgb2JqPDwvU3VidHlw ZS9JbWFnZS9MZW5ndGggMTY1L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9C aXRzUGVyQ29tcG9uZW50IDEvV2lkdGggMTEyL0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDEx Mj4+L0hlaWdodCAyODgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCsB9dihWjClSrwYaJQDepfdqH5XA uD9/tu4MGoquHnVsiO+508xdrc5kq5iBtSCkzNIJYRyeI+i0e0rok8aFtJ87suDu+hHDHk4/WYSu BV7GrySXG+0q7pPt+/aF5HRiRxMEBwFzMvT4nIdl96nWRdFcnGRoHPbrxi8YoGwqvWPLcv/bOKhz b5t2NvffxRCdTHPQyJtBT8XXfdq4Cxlbz0GGAqTYYA0KZW5kc3RyZWFtDWVuZG9iag01IDAgb2Jq PDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNTEwL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1h c2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNDQ4L0RlY29kZVBhcm1zPDwvSyAtMS9D b2x1bW5zIDQ0OD4+L0hlaWdodCAyODgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCmJ3grO7fhC05OsB DPl3QJGI1+IBWF1eL7VK7wOl4KyqiSxPOOy40Z9eeWFXjU2gmpfI9oFrPkczzQTkmYAO55UMyAVa r/fAX1zaWTqVnHaqLG9yp/F5onxLVRHgEknq+k6lsQ195NZ7mYnL1xS9xHIpebiEFY/sgM/rQ3YM CkT6MY2FI1ZCWlBdH0cSLm/0DuojaiAffZws21zWTHUcuBXuK6jM27NP95OaQUP20SwIf2RmjChq k2PuRJQh/f/EjINZJMN1io/TEG+6Qq6SG1uAp0uwCQZwxh3CicJCjcYmSCsG/GFVlgpDH8cqNiUs QfolwjnF5P2US0Fo/Y42P0Dlppb4Si9uSSj4RYtsHFifTG0jGCNmRpVzSAVyHnTC5ixZW8EYKeZs PqZ643xODQ0VyHk4Os76/fTYhjYxlwwB6+Fi1JjEkwCEzCqOzNohAuRJNZWTSRr4CZxcQZyofn+S OYafJpSReVB4cqatZLhoxxXw4XpPfQQE27UxvqMIN9rIq51O2pSqwggarSKcVIzMurLpIFPvrPBc HWS0Mh4JhVsGGOihgf76RULyGngHGSgjhLYg+tiqR92bmawCKon85b2cF0dbQ3BFrASUvc1+lZdz 6IRawfxMf5glXNC0tlbJWUpc4ZkS4tJIoBfjXdofqPsYbrs327Ez1736mw0KZW5kc3RyZWFtDWVu ZG9iag02IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggOTEvRmlsdGVyL0ZsYXRlRGVjb2Rl L0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAxNi9IZWlnaHQgMjg4L1R5 cGUvWE9iamVjdD4+c3RyZWFtDQqL72rkttK3RLMiRPrJBy+8GlCnPQoIHRaMyGfvBoy0X9e+Vn8L a2Z375+KVZIDNBDDbb6vulHVHaZLmlJqAqMMEXocmghPIczUZZUiXdTBcFqS7t6+VIYqMTmhDQpl bmRzdHJlYW0NZW5kb2JqDTcgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAzMDUvRmlsdGVy L0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAy MzIvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMjMyPj4vSGVpZ2h0IDU2NC9UeXBlL1hPYmpl Y3Q+PnN0cmVhbQ0K5klKWLsGX+medUR+cR6dYPYCmP4t1bm1PSKr9jTyE2zzSrOkWr00bbTajFkr mHKotK4j9GAS4Y+JFNmFN6dWbp1UHPDjo2m6ykBeGlkV2ASGaX8OT28EqWJgFspqMXEzkO3LwJKn hGQ4FmR42WPraJLCbc5WVsXPq3IOCZgJnko+mYBH3bFqbG8SCCcTMZL/tKft/sNGKmpsZDsjj2tP EJm/EfpJj9m2sjGqE4TVz0RGPtSPo/U8u/qcEn5b45+9p/gQfL9t/vXl+CSXA9KzrhIJOItBvsfy By2wWxxNAsTvTGra5RG4UY2zT3hzYcuUxpV0EVARtScdBPgD63PZ/6c9ALmgdp4W8OQ6SSSJTS3r we6lF13UYjnODGlwQ2ZcFkJFDigiCqSvo0Q+gUrl+wwNCmVuZHN0cmVhbQ1lbmRvYmoNOCAwIG9i ajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDEzMi9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VN YXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDU2L0RlY29kZVBhcm1zPDwvSyAtMS9D b2x1bW5zIDU2Pj4vSGVpZ2h0IDU2NC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0K8Qn+kCfx++KY8pIS ddmx6hnHm6shQsrK/AsMWOH5OYptU9COi2nf1qsTK6PriFC+pjWXSTkN4a5hiOrCcgv6fuC8z/m0 KZpqMJ8iUHLG4H9od6Uz9Ql/1JOpmcHNu5+AjAFnL7JNvSHuLV3CgNbRitAz9ttvBS4MdmCyoXkN b0hqNWMRDQplbmRzdHJlYW0NZW5kb2JqDTkgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA0 NTQvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQg MS9XaWR0aCA0MzIvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNDMyPj4vSGVpZ2h0IDI4OC9U eXBlL1hPYmplY3Q+PnN0cmVhbQ0KFglHTNU8Zhqt32wOps26mAJSXMqSUGGJsyZQ/O5CLjy4Swca xgxLQ1V5sKqVkZ+0NOFxLVR50bBHlY8tiNGxWyhYu2yOqkuBo8LMvnBsWkYnhcOIa/xZ26ZAT4Dn l6e9MRXdLmMwTlVMoZy83YEYwWPRF40nviN0Rs2bTMOgH45iO8y6/TVml0xHhf6D0Vxl5BKeNCM8 JkhgIXNHDkZm7B0Sm+/PS2BwmoPGgQAH4nrlG9HC54+xED3QFIYaIqH+TE5uTXLIiIBM/VjNppYB Soiidq5wXedWmklkivhdg4jWMPLiPG8Pg/278Ia83oGOEF7GMdpsY3J7gkgsG/mvsrOyW15LdgGy 3OzRuFhOXmXTQRsSno1ERqND72stqyZDOIiI5sFv0UiER15FK44oK/JoAd7wYLkAbh7iRTWC6YpM /GcjImeJYvwIOKRCEtIzg+aUfsPKYlYwRaAHZ1hVoCeLO5Y4VBAic2IGuvLpVSTDUj91pZboy+jV Zjyu9AF3J2tY0DFfWlCLStCdG3Bkdj30lvgZQ6T9SlsGDG6iYODp+oll+1q2evdqTiIsE7wPKAr/ 7N4XCI62nd0GAGck52i0c4cl0A0KZW5kc3RyZWFtDWVuZG9iag0xMCAwIG9iajw8L1N1YnR5cGUv SW1hZ2UvTGVuZ3RoIDM1OS9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0 c1BlckNvbXBvbmVudCAxL1dpZHRoIDI4OC9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAyODg+ Pi9IZWlnaHQgMzY2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQqAadWwthIyRiss45c82AGk44H1hSeK NQwxSw84KY0C2Fp/at9FUHDvqoDqjjAc7J9HOUO2RMT6UJjXvWTXL+qodPhFrb6zNOEjF+tHo4rQ 9UvJnb0Yy4USkFQqKCjT3NPdANgszFq4eZd7URqU2unmkb1M6I0MTYqxssNHgRChEj4nN54S0tuN Ofl2nVrdjIKcxxFBb39FhpAFsSe4HxSsXWLov+p4nedfe35bdjRhyJSRfrdSrErVADjIlP2QCbpF r2yRNZ7xgvLJEGkKgrHqdCYTkzuwCGQsB0ncb5zQJrxVDn01vYeyITXqO7cQwDLr4rzqJYGlpJJd HA6zcLb8goA8Uthpiczd6rYghs42FAVr3E12krNxY2qdntRYCxcKxfkF4T3ZUF16KOvPoZJG627f jnrongtRW+I85+QLQDJRAWTTGRIBvLSujZUqAcxvoRI0EPxU90ths5SWU6CUTGWuV7dvDw0KZW5k c3RyZWFtDWVuZG9iag0xMSAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDUxNC9GaWx0ZXIv Q0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDQ0 MC9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA0NDA+Pi9IZWlnaHQgMjg4L1R5cGUvWE9iamVj dD4+c3RyZWFtDQox1UqJbThlVNVF7Q9fzYlHx/D4FcKclZYmJz9899weQU4IDk9t3lkueymu1hVM wyRSQ1fOurNNchsEAfSSJNe0lIqodUm/qYhDHGg6j7ucoEYXMw/eFd/OWQiQzBRLtmVA4FVNzvc9 S90P6Bhc829UlRRM4U87Vn4JOlX6bVCq9zvfO1WkU2lwe1EdT8GhrYAGNjqV7/IXwt2WlzVA/Sn1 c9h9qN81Pkje18nb309jSj7knP9/xKJS3EZuu0Vv3rTjljJJxVX1ZlVDuC35iajTBkF05FcrSJbD AXIIK/vmPPiq6jqoSuOORD8mH/Ydz+A1F706xb3ovYKlbkzXp6kj2G0MxYBP4pLset7HgaaNpSlq iDBfNYxvIqWXA8N/DUwA7ceUm2z9O91njhq28xEx93cZkONUMxtytrfv5E6O/toxNHC3ZRtsSJ+8 Mc1Oy06RxdJqSO3E5tO0GMtLZrPfoD2rwDb31UVq8tSw/Ne/wlGt5qZuHEwlFkyCq67idBt8HLCL D1BBgsnPZhcCT+6geoo4lN3HVnbTxIIMHwlIwVeHAL9zQlYL6dOu4DHnNmOn90+86QykF8xzVF1R LrirbeImBpIsftFtsuD/osNxeb0xR1LrpVE39hfwJz3uWEWHJBskYW3uN/+t8+DZHIIFpK3UATH/ 5rbQmv5hixAv0worDQplbmRzdHJlYW0NZW5kb2JqDTEyIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9M ZW5ndGggNDExL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29t cG9uZW50IDEvV2lkdGggMzYwL0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDM2MD4+L0hlaWdo dCAzNTQvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCl0mo3sBYOBYZT21i9oKE3b5s5APG6wSiif229Dk rxxqt9oJ+Lk/ywmet2Kdg07LpVJnExKDHujlQ4nt7dKo3zZPOA7nKMzwxh6b+PqLuEu63XfOKY28 IRMkqcmzjC+SPut51h9bMW3Xpialxqsbgf9mGBP0seN2jAlIb+05UkMZ7BOz2ZBpKYtTDUAPB1sN gcZdomFUCqq5Z6VO97HMO+Q31te2ITJxgLLKnsesCDmLo4WB17sjS5h7a3O6hqTKRT8aAUrOkQOB LFgisapEsXug9jIxVaFzBqY2SbNiy2LCJ0VMRraOu2fIHvXOQtRCrK1hdT5yD/A8tDARzLrFWf+m XHl9m8Bk4EFY2+fRZauj98sLvnlQrKKGy93QuRuXqzrL41G1pAU1cIiVqgHiau6ONs+t/QhIlGhZ JuPVzCQiy4jH5Vme9oQLYB7Fzwdb+S5I6bOv/qGUYJSD9IVwvy6Rj11HGIGhaAyLcaUuls9vmbWg dyJAYQt9qGJ4DWdjzZKkVhL4UpYo2tRaA+NB9/oo5tL9ytDI90XtQA0KZW5kc3RyZWFtDWVuZG9i ag0xMyAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDM0My9GaWx0ZXIvQ0NJVFRGYXhEZWNv ZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDIwOC9EZWNvZGVQYXJt czw8L0sgLTEvQ29sdW1ucyAyMDg+Pi9IZWlnaHQgNjEyL1R5cGUvWE9iamVjdD4+c3RyZWFtDQrY 5BXRXszbh35qnYuthpHdyG+KXVCCATLX1gmxWCaGz3b1UP5Tjc8gHe5MOqAE5ylbnMRJf4igJmVz KvXcKVRgiOsH8sFqSWII98vcZQTp1qRaSgFmZoxHAvxpfwrsORIdv4eDf9F1OHNpF6tzz/k+6fCM HKKnOElt8lsZkzjk+v2xwZNz6MFFP+P0NaCRp3Jn0VLGpcjb72XKqkrWxkB8WE4rGwuleJ+r2CgP Nf4nP+vHiKM8+pCF8Me1yUjuMkD9sYgUKQf0MgzhF9HrqEmEpB7pvBkWP3RhFdh+VXkK88dLgk3i u0S9rFKFKTtW/xmCVWH11sPyONooAebER88j+Jl+AUAz7y3gNBkgi+Zal1l9cZLZk1+3R9r+p83r 6hsbP3JcU/cCHZ3lLcfx3pLwOIoW5DYilpIfha28PAwlZ15WhaJ+51Lv4KmgVAynJkGl3b6v3V+8 DQplbmRzdHJlYW0NZW5kb2JqDTE0IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMjA1L0Zp bHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lk dGggNjQvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNjQ+Pi9IZWlnaHQgNjEyL1R5cGUvWE9i amVjdD4+c3RyZWFtDQqaJx8wMJmSQ6xxHjW2PccUDGZF1JvYsm1KdsTYXwZZYMGpUEEJZcBDf2so IrTSTCTR/IRm0xzpFHFsHeIlMkEPseylMyhzh4U9JIziSY7HB9eusYQCakILsCKDIeuR7whCsEun mcOSr1vH+zvHqbJG9Gi0VIuNkzqPTPQe4Ak+ojFBgUxKLD+iZSiQt9Ftp9ulILQKJ4pCIYxElVOe /Tmp4oc1LI+y/cf3N/ZnmYf93mPm6MtWmoRDRr3HKVk9+fYmZLvDF1+J7Hm2wh8dDQplbmRzdHJl YW0NZW5kb2JqDTE1IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMTI2MjAvRmlsdGVyL0RD VERlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9EZXZpY2VHcmF5L1dpZHRoIDI3 MC9IZWlnaHQgMTU5L1R5cGUvWE9iamVjdD4+c3RyZWFtDQqChN2X7Po2aUrg5OLnjq/cE4It1mu6 /Xx2KiqDIJ1G6x64wNpgWA62zY/A67ANo+RMisMZ8A/8xo0AwBoJ1WPl7MOLjP3FrJZhtrgNSIQc KbNZNAXeq8Ur5SXLHub0NX9jV8Olt1wnsTCBKfLHTctTQCQdbbNby5u5X+/kiOm8s6AiXpHGCBFF zHb7wN0lsck1y7jcf2/58eI6qrb8mGjIZWoA2zFfYJTenyKWdINYL6wAm4PzqRU3pWmTzjeLjM8h rWQj6D/vIru4C5JoBp7U9Y3WUnDikNkrUAdN+ZTBLCHR4yzWDZIP1MoQ5pR3bqTBs0s7rLpEISpn SdH9YxbW8VL2Avk9Y2mE99GGYL/rY4a4CmdiGZa6Rbs3f5vtJ+O28XWH0h+vcp3J1BGtaHCB0ieT ciGvIUWokjjooExYjRDCi9odGEIcTze6K8iKiMqYv0KxmFARCexdZ5Iz7sLETWioGRiR9vjK6pcp VGi53V4B4E5MmxCoh0b2oo8PRuJ1jxF4J7zU/Mi0KIHoCjxAkodFGh1CQoKl1Gml17XEirorhSl0 9WQTeuVsvvETKXmQIT2JV8t43r9xR9U3VKkXaKcrItynC5RnTXX157kOVKTyL/e5N+liHczfXSki sNfJLdY5ueMSH1zbMHZN5bOGm1lfhr7OeHn7dN5oWzRNtC55Ja7ys+2Ao0bNDkf2Gp2BH1xboL1/ IjoHH1SXbVhy6PmbpOq4SizKljAwYAXxPkGWd06EC0UgfZ9UXizrsxMRmGVSTn9zMKbJ+M1151b9 LDliHSb/UiceremQ9jdlVs6+ikELINEnRHA/J88J8w2646e7XHGl47SIHEx743fEwDJo0bXCmovk ohQyZJSI4V0rAqeUOcCd5YMVbE/QMRQtw4wu1jgd1gaYg4q3mO1AwQbHz7FUfRNj/UNnWrVfsAlv 8QJQGx73dPUtFQ9Q5CODrN1kIw8YxSk/Z3acGKNaYrJInzI3gHsKWT7vfuCjgf7s7pyNMBXgG8Jd f0poS8steNXB1pmi6lmMR0rxZ0+wzqDsv19vwcx5Q09ppK9s0wU9qmE7FmW7Mqb3MPG1DuaE2etO mCFJD0MA0QMrNUaGCLXOv7DlUFCgj8ElYNVKCdGKc70ALhEnMwX8inG+P2ZZsRykjoLlCjJJOzm8 kr9pIXTVmNQI2j4IaKkqsd2Uax1PKUTSVXUK/ILuwggJpIjY7dMkmh6C4MkGW9AfEkEIPeX+kfA/ q0MOwv6G/4nNqLZIostgt736tNJOq0Axu1DZLsAdPHp7Wxk1pDkO4W6jZ5UgEK4RJs9U7lwNlwVj GEM5I7NZ15FZQEvrCMI8SnPuNEReLxG5hQBzupqFzhVI4WllNgik348AhMF4J50VexA+saIUHl4D ZbT6zgSEh7ICIypqE/8ntixxjM2JLmlhxeRi+YmjZTY8CNpQ/i01kQ+XHur+AaWrSamZISFOabIP HMX7EcHK0vSfN2aFs3j/5tzkncwasDgrG0Lpkue5A0ymKBlE5BAW7kE0i4D5y3pYujf45rDsLLY6 uI3rZLLUnyGDpIpzV8HVtoI3NSNuVU18iixaOespzRcZNXJBm/7JEU5YQ8zDj0QkwU0v615hC+YF h15i6PiCPxutShlsPMhjHN9/in2r5hZlpEd6o127WYRp9qbm5mEte2MWQMmsbebB24NoWmgZIZYP lRYhUBJxAP/AkHI9svNTwO8sQ7lRwCgIbOTIrpdYT+s5Y0gxQwnHbUkYzJAwdWLXL4rpgdjI394H TBvpUDrxuXOutFh1IIuRiGaoFnAupDCKa9TPLJGyF5IcRqFXNmFApO85tTkME8H708b3PwyBeNhQ yOeaW0GYoq/MLNb3LOs4qvNX8EQRt4u4/x5z9YepZFjiak37F5k/NQZZSa+B0d1h0qvhGGzgaMNw XneRsbGjeZlNCL3MjAmR7nPtt+EYwzNAEGcPpDkx5Ca3JluZJlwmhqzJej6zxgAceiIgMXdSfiSx BTVnpVACkrMOBxZNlHKBNedHI25yQ1Y0un3QsZAu9SIyms+ok2KzzBnCPUcFxqXUJ5H6WHm6qkTn YuBw32Tq9Wbh709RI59fL5+O8CHbQvzdcImBZcY1dlgSm1KaWKktqwBjFTorAPekIMVZ2aw2Xc6t Yd1VTLG2QKvOdVlaaZzZNQPoavdkmM/a7SoV+w79ggLQea0Xz6oPzDfaeReEt6GW3M42GTrd8TWh 5MudnNvbQYfErFRE8OJej+AjRPKiNSAQSGMf5h9raNTeIQxaDlPhDx5dhcdVMm3yUM2DTpDJGB+T xzR86N6HAKosKOjr8LyLGI/yEDPZf65Rgu3CSdTe3P+bvH0KZXKtMibPMREofGWKXy01mGpEhdSv Cf40hcStoEAhXdQc4N1Sqg1BTnlm7nb/KwmNWY8VAJhPsxYkYXX+WQPWsyxmdMMJvouNu/K5Mk9X 5HEyxVJsseaQouZ/0sDrtnN0klnotamWBybwt0wh3bq/3hl/8QexwzU4KZr+rfWM5zl3hSImMOCg Ours5ByzuetZ5IQvJTM5kMjzXpzsa4K90Y+DCIcE27aR4V4X2MulAVKNwO5v2YWcUYT7isvVpGUq xrSRv6l9FPsyNIXX4oL7HhJnyW7G4ZfUazSBKJAXRX6o301qnD0xuvRo4AIL5gS7g3EVDatcre6A mq95eZd6W4pgeTVzAjyPXdlOByTzwzqlD2lu2YUjYXU/yHOys5kMF59xyYlr7RUZN7vqNmsXbA3O hGCJ53inCjMdjI+OQbAOQQSZ+f0JFHUevgnk31Fy8/pVLiM5l9CnBpcmmSJ2eXauMwzEKghxpjvQ FXytdriT6MWkbIGIu/mCXGbeH8CV+HdlvxbKV4A42Fi/kVwjcGIHD7wtOQ60lq96pN6Un9qD5rFB oUkXRphKCXCywWfRBl55uhCuiE7O2WyhqV4oslXVftQI0KDmsvubjebLKXXNYn3yyMWmtrdSut2F pn1hEoOo67iWHvUKyOuXg6RifneSLDU5itGvOjvCI4jUJQ1v9ZdwL5r1fKcQs40FjzL586k7i+G8 VPLYK84MrgHppAHiwAqUNhDV2eiOyoi6zVlpLJFubMygBUJfQsglp5huPrB1NNQxHJcY8QSdeodB cbSt0DKucSSU/p74TYGd5nLXOLz6Frvwi8zAlIZzXVy8kgnvZKFo+zEWgdCvJO2N0yp6VVJaQ2Bx vrBm8KqeWLKAw9pIKBRj4TU/2HY2DOd/S8D7YmrBWE5e7oza1GXb600kjDLfY7bLwUhfXnnllRQf IVoBaLqSEihHnM0mUH6kdR58a3idjRFSdpUJm2zRCSNa0RsEb9/EclL8v9htzKBUJYKde17sfGI+ 4WmhLmQ4yYNJoPbFoZUN4aZUZDxHKhrSY6tJY1p8CNkslwi74wXPn9aOuTL0AzyeBkK3YLe+GF6V j62Ws2Nx5Oi4Vbgx3Wl+KaRdEuY0Rywm1snWvda3N2RmbWic+YmrE/TZ4mRVco9e75+Jl+IiH1OI JAnM5pjA6G7TZT9jkrSY3M8hfgdwR8+sej04u5qftFMVonuEZvIX2+1RC0uBta2Ep5+zbqyql664 j2hloDXq0ArINjJcsnFI2GwvN6Q5tkGZXbrdSKYuUOac8pVs9h5H0zcn8W092TmxST6OiUoZSB4J fvNU4SBGubYts6OmEI5f38kwh0GcM0oxMoWPOjDKQrYvlBvhxhKVUvYrhNm/ywt3WBp4TjHSOOXM Z7pFH0leLqixMDsE6eLf4pB8+AGnMxmblsv9xm0xvScFraN4PoCpsRqVlnEIg+UwVAH2OTGRIeTO bHXfTgl4Ywf4LMq7F8pnTdk9oCTyxeAFNAB9brykDFEK5bdLeRHgv9SackU6OMwsTmcRN5mc41Yf 9xti02D7OXt6jqHdyDcljlUJL74WsVdsLzSiXYhJNrnTx5lR5DR33HCJA95QDT3nRwtDb3H0UWFt LrhQGX3e+6F4cFXfU76HgJ11vtwfyYMl1+c0UbW4+SQ5uKml7lF+CDIkivK4jlZN2D7P1VEom1Qd Ugq8BcOdkdbqfBChRKZzfB9WJgcowzDZxra+FS5E/6QmdJ9G7dSLoyt9sAqFw96vos7Ak4H6+x+T SA7FrdNwxn4DP+CJbNYtZG3vh8ENiGMaCPC66nnL5bRJfq+wNT5WDTHjiAD9OB7tgfHOy38xcuCQ OS798TJr4O39XPPSmUX8tSKG585EMBOodRaUk5FE25RBgcOc8tY4y8cElQnEqHbSUenMxf+so/7x jtSLchiwHMImyE8GIV7JUjNvaj0mhrfT2CWhi9Z8Ev7RbIqkGeiZzbE7VWTH69xPP85uLG2F797E JPMYWJ4KIt1t2yruHVB/yM8ipREwKkctB8pTzz6UG2rMjQZfJgRUZToauNnNbTwqrgVxk0Bx4h5h zahuIQgXQ/5UvcifRCF6orUzeSNoHfr/9GzzAdaPatbgphx2h+H9DeokOfC73tXk7drqTV5TQcQr JIR4keSHMg5dVUDOy0fq9RdgVA8uAErxFwFLnpTIuFFAn93OtzJt/Y4crxgpD3TkhlhqxZxeyLYS HTEOYWQRMAdQclcgJpR0m7IwgFN0EXHGHhpuX4tOjxFfJ1Xp2NrImeEOvUyuVDqXNQg1Ix4uarfC qR0bQlbkzpksyUAngrGzsp/8YhWKeGuW5Fdnt1xXtDfqGbeKXsuR2tLkntCfM3m7q/xabKZsa3E4 VLcXVWp67xXrJ9NTR/YCrV6CenVYluu3eaBBkYAsXcZWuOuIZW8UrUPZgebyEpXxr6Z2kTBGAcaB eo0lv6i1B9QT5PQohYYMOOslitiEjBhvcWMFvdWcD+VeRWmLXpsPk5RAlv5Ulpn4o86poSzBwrNM jcA9Z+BqWKzPlhMjf04fdQqunbCrLLAHFcdvTHSSstTVI8Amc3iLq7/OSoDjKrivrrREsZuHl6Di AU8Tjq8YkCMI7H6oMompU2sRezWKnwaGbxQD3ku4fZESpe/0SVXT4EAqShNpTLjlxxb4LmtvWgcg Z8T+UM/bwuDRp1EmyW470j3q7Otprtb6ALFpb/gQXl9cMMOzQ5YH7ce69oNDvC6++E3mJmwCvD/2 FEbEmXoWsgeJF3JUCBnGz77inV9FWKG95M7vznpjgnetvmNZNSpdLIHTol9Qc3PNFisb+fKlAVZK fkTvoMFi/iQSXG5pabzNSropv22rq/RHCiizCC3fIgHezpTbELIOHs0WG+KlB2EDeqCAG/Osk7sC yhQBjS8u07LfI9mOQwwbnMRjDfC9JYDsqbZXglarkgplL78aQ+/gGnu31ds+QhpyZHZuJY9aso7H yu262u1ps4GguEbxbBLZhizYSPiB0EeQNe94vCM+GK+BXLfja5TjfVab9KgG/1iZ4OJJyZWiJxSQ /rAKeS/d9eSgBxISWYx2O/ME7h+v9x25uV1C8QmlYIZ5faY8YRl0xrKks4kBivJQnfXRqqnci5RK /wIm+DzrPlZOJrbpp/H4/Bec9YSiCBTq1We4nTwiTViOEaXXKTB4qCC+H5Dfyu8gudeO2MyT57HC LSII/O7rNs6H/ZiNFeyOSEI7vQEVimIuumZ+/WtYVzQLZgCeFVgoOMzmLRwkXSFOyZHCzqmc32zk IjrjNKctoZYbQ6pZuilhSjKc5iKERmadnxLz2PwoyttaGfVDab3d323qS/Y9Wd1WqgciY4qpzqfk rQUYAzW0+q/yoJ+wFzof0rbRjt/Yip8FOoKdRRX6REXsaqV7oNEpNroljJPIuf4qOtXZIsd9f+b+ XlEn7AlCVejIPjlbOsUiD5L+h2thvDPqUAPr6GriA9Hv6g4GDAfvTus3hAuCkgaNa77S43DPz54X OD343TWP64h5qVYyz1wwYDsmGx0372XeGd1R04CYgeiJAGQOSUUqRA9yvRvj1R/VBHviJdPGXsnr MAATaFbD1TSobYZVvoHXZ/fXoIXOg1TkO6Pvyot9Mbi2OT85ZUp99Yrc2eVHDi6ZxXsMIY+/dDtO Z41smkvm2+UJ47PF780+icdM182EHTUMJU3Agyu32servpgz1bEZaUChwN8B6Ywjz2/dVfHojWHl +lFuTDPfHmhblVTNu2T0vt3Lz672dW7GJnKToFgMFcU1+FVznBOmxvGSZ8xgqT3W/ibx4wg1KLG4 tf8+btVY38HIVCjAhUrVC0BLunggRVZ2YJ5uWAo6gjRfGpRyaFp0FCjrxnrDQ3uICo1ih39A5L2s EKb7vc7Y8xM8Mok9G2ugWyL5r415k4a2i4UeFX9LLexOI9oNtzrIQJFDov6ZOjRNsqPZMGcfSF0o ZwlgyGpcpGtkuDFgffXSEpqUceg0XvaregWtOxjb5/Phrs+NLEhqyZMDDW32h03nwwvpYGfYNBlf qVKYIHNXRiABJElLjg0dJ0kL1uyVaXgpTCKYQuHsiiFGyHQtUEQlbysirDtK+CgezJ5RzLfQSICZ zz6nDHkkcbf1NL3zhPlG6gN6bskR5UPhvIYLmTAjsvL68fXQ6lfu5yg55l4AYzI9qJU8uq5jQ0aK PC2vk7SdXmD7LyNhGTOa2REwc/1iiD9WRIbfal4inqt0otakoYRmmJzZioekBcrf4CAIYybBIA3d H3fipGHApuu29yS6Mvbptk6jfT3m4WDCLo6Ml4xtHz5ETbyLdQGs4yXaHNFo/7ECPdaeaObr85yT ezHAFROeWu8LdDz4y3SwfmmeyOhJMb31Pql57DnGAnEqKE/kSP+2q8bACHuZSH5IrZuJZvFSs1to np0DQpCYfXFIFeocbCUZJiqJyC1DoMwCLMUS1Nv1RGm4rNj/chmPX6VrXlHcuS2ySGGH7YRFejbr Keu9WSsYEGV4l3Nx1lRUa2TdatEG025a/lq82B8LvHz4UHU5QIgo4SAsP8fTlHP3OxT5+K+OI7FV zE/nUqniY4uAZKYh9hSZeMm3TnnKV392Vab2kiWV+G7w+VAKISXxNI0Hlc8aFjKECtHEDGTeTuPm kfBK6a92uj6DoUJxjaXjTb24seL5ZOoGwEUQFVmyAa+w+c8z7XpmvAeBtkZ0/I5zTTIYhZYVNMit ORfJoGY4F3YYxmChCdFNYdgbj3hyRrFoKBWjPAtV2Hm1/shUOArQkwlY/y5Xb7tRzhMaKqDkLGgE pQqyYg7RxrKwNF3BFkLozYWgAjifGPfZ61ZGE37c0ElwQR/UnSkABOKDWDcQAPzPe4szWw8PSt23 rydFLnrrgxK0ebev5S/8pB3d8plVIQE3mWFUmPIrFXY4iXpOTEP9XDrBnZzVNHXcsLtaUpoiPBfw abpB7BU4OoTUUJhxsTNBHyCSGKx9ILSuwrcmAKC5204Ky5RZy/m6ldOe1D7873Yg3h3mh9+zRo0C /jdsTq1xDoxyP7F7BRRXNw/75bjTwEFI/iIjt1OwTiVnjiB0jBU0DBgsB4qxMWFryqHpbUiUqnwY wxMEgIhPmoWLnGTwG2RY/RmX23O25LK9L2qv3KJxkHnR1Ew5yaB1YCd6iBLC4yeUF8Cjh0ecORdR +bhvudeMl/c5s5JGErDKkuuT1hFrbJ8TIJ9X0K/hUnF1qT/4p20lUPGuAv+PxAssNWe/TPSzGB/P DDeRP+GRsTxF4Ebxta1/eOdfGDIJ4RphPWtKbC0rRaxG3rT2F1d7Wj7WS8G0aNkpFmOYeU8ezcy5 opzEyDekd0W2XXXGlevPX5M9htD0L8iHmqAMDHcy+ADEpJ6XfJSCugFUbwhrZHHKoEozr2Xz4x1s 0s8wcdNoVSFhCk9ObAhPZm/qjaQFWEQW/0YGYNm3ruIjpG/7svVXhm5yzzAWNTkXxPjFYOqdc90V LRC3e0y/M7XvmFq2RIqWBdn71P3iqW9WsdLcluUHqxAEc0cfcYXEijbO43s5iWK6Xbv41BPBpd7P UCjZDQnlQorgaEthpYfEOeb7dXeddZvyoZPTOmBAJT5jhBqq+8Ks+ocWozJIxDlB7BIJo/vEVMeG yfNFEriC/gcg3K9nhGGcnDDTvmzSsgq/pskLW8Pi6e0Zh7js0RYpquFe2L4ktFTP49gf3alk+YV+ o4tELqBJtnqeAuxoEpJHe+DbCvgvpb7zfvZ8vdwoJbm//ZPw81dGPsiMJtUzt8mNNTSisK+w8jQL VVYVK8KUhoAk5DLsP2SXK3xXE3OB48KCRqUBIZdyAnsOYT+Kg4UWhPGPoT6x1Fmeq1BTtnnGWcV0 pNx5+xy3quo/P2/lMipnbRYEkbPsHH+EglTJwrKAC/afMlVAmBBR3Jt9TBZR4a5UQkF8pz6bfE+c fciMkQ0R/q4jfCKm46Hx0rdBsePclz9ZKLVSC/BDMbyjGCfqN8dKOEWFe6DqWswEvmk/ZU62Ry4E MzWuK4uXGSbI7p3P68+TiMGopTqmSCLq9f5SAIwMc0OJQ6pxmV/wQF/Jy8uazlC1T8Nz0o5Yf6Ta lOItoka2f2aGJT+AM9Ow8otYoc+6OR5LCL3roLQD7na8CC/ChFdk9Dx6GA8TLhAqA85OYffHRb2v hMYkiRWgLmiOLgJPu41l6mEN6ExTkyuc/P1FefZbrOtL2cwZtz9bEsOIxC4y/6CEuqoDiM2iaqBf nzgmzbH7HTzfqUuaXYrN9uRqvsFSWW4v+QCRgEsKhktMKymNSh8+Rzq5uf26KB314Vl0Tpk+4VmG foWekEWt/aETp2wU17hyNBEXpZDChZ9m6seAZTgWRB4BhavvvyIuCwTd8154lnbuo9VfG24M6Huy GHtw44MOe3NHi//xvPgF8zvtnHJhq9APzxVW0Or4qhuSVlrD1l2s3Tngewy5E9ejvWg6d6OuomBN DVbOji3WKT9UYYUkuwz31tVWkgwNYxTKdU8p7wuI0E6USQzN6HhSncC2w9uyduCWtIUl892Z17jj cnGQmozNm60LugB8jJmpUBV2CNsnj9Qy45y0RKSDA8muTk9/1G93u++CCTHqoJQuVAp/s7jD4VOL ftjmlvbmscE2asb9ZUbYqAYCBb67sNFd+tK+wwiAmkzRRXUlROiItYZvYT9x2mFOMuJFJ0sAeqXj doRRadB+KX6ZXPEbOwGEfpqTUcUlEwx1nGvTeoOYlyXk+lj0nIQoM7PhAq8A6lsbhE4YXn+sDPjG vFiaHTCgHuxHMOZScO7PSXqBKwH8sN1ScIdEhiVWnrDOD9CPHi3ZBfJtf4j+X+0a81NLaH0HOBWj XSBbLqzS57+O5bn0jHFu7oPPZQfkYjVqIoJ8ebUG8CBX6ts64uMFt9kG8DmG7Ky/zKmVe3WkEu92 faGqMl+zTCi4PK9rSoKCGxskEKhhbJlRvTYDKfljk5bHWvGhEQEttYJWHusypCleqsuHRSvs3Nkr b/tnMe6bR3YAEQ9RuqumwpKjy3762f7JopojqlvOjR+beacBgDCewYApny9NyraaEq6i8kd+2ER2 ndA8RBa7AfdCKiBqvCfHfHEbS1km8gY/0Crc44BrFL1ZA5M5L+fW2pqq7D0O9qypax/sfk9ctHfg y8rGFjnFkudw4+56ebR/0A7daqU1ZLZo3w5JkdqIAwF6sJUb/DLTB6CtV3Wj6G2OMNXnyAj+F/JU 0weEW4/6khcSXL2VM3V0aM5nOAPvqz+TacECR34MQZYCe/Kjcti0z7UmCJxX66jEPbN60JYWF0AE 6+OTvXO/+1+Xvtyuavyh99FFrQ9D342+yJ2ZZLJI7hYomosGH3QZKYHunCs6DZaSXHM5mw+f8wWR nJb5nzbBIbyHfTcHqEnITYbHYZAcrdU5Qg9ekD6c6IPtogXCNrW5nXhh6HVWRAMtrdemHdxEHs0k Vym334AMuXUusEM9mkv55PbbuFRD+nDAolCdIt9UtTiszOurHtpPrnmEUfI/tetrz432rEx4yGYc bQiy3sdAnL02ykkK0doPODWBcIQBXBqjWfh5uKC2zp+gs45c9gw2ijWzV4SAhgQ0qvbaVQ1wkKoP +r21OsZEfTXBgRjjHybAvSAqKbiWy++brY0uZGAB66swaU3qtwdY1o1w1gjUYkMoZjyXvXzPq6U2 atFJtKm0qSaPpXMiCNPm9QUkDr+iy+1vJIteErdzaTAIrhzuyZARepE5mMcPK370IShB8Gab9DrI oNTrxw7zgtngn8VvhcEpJr1zILEdB4mAS9kMgMyWRqF91FnAkHg6EKrLyeSzIjYvlIZLFJAS0PfE jpE0C3drm7UPFan7ZDy9zy3ipuA9y8xYrHQogvE+jXVU2JjV8H3TyGLMMH6+jlvWEJsuGYNTJKI4 nGkxZ0via9sRmUloqQLCTZRHMOSPCEmXUbPTDQ8fY55kduKVUisYCbdGXDrl+//YTj2lQvOCnlKh wYxskRpEfvHKi6SuAnvd9WEgCOlDu498P3SR/TjgpVp8HMW/yyUrmNS3gWCkbfyoeEJnl249g/qG UM7Vh3Fg1BSUEsqNZprgat/sV/kFp/gy/VpyODJKYIL18X43TXYgHwjFsJPX9c1XP8BfagNltDvg T95vtVraDtbw4S+xDXPLeQGK+MkwjtvmoQIYiPu5IemLPIp+scgdgnJcehANT+tABS8YrATa1u9k 47SONdqss+j7ScayFt7RfrZ6fiZ+zZqimJiGsxnNT1MOYQyGKcdDuIiwNrIe2pCdcZS4AuSiliC5 Q4N/iHLBjvX/RSSyg215eL4GfKpnon/XOuhZ3W2BunI8MXS6GblA6wNCc4qRfWtco5KmaovQ9eDr ryj2neRR9SI4oXVKEoD0LN3P37KUrCA1V03WMclAuyZ7o4ukvnXUZHMKvO0UNQWG2B7BFiYKjDyu Q+7NoIHO5NkWXDKE9RFuEBxy+cu2DorgfHYsCbkvk9qVlTXLqcHG6KgEgqkqK3tg/T5OkhoPB0UE YceNyDSFE6FIzpm5URpRE7jX84YjKtB8zzRdtueDyLUJrLTUvrMckQYKiHc3fCImqOwPwsNUOJ+r uk4jz3MDkSwMXkJ4nMg+U737QZ0WfptBX1bMANeAl98bfgGi1TajYwDSO/a8JHU+fEz3kVvx4D9Q wbODLfXzhS+9pG9vyBZyN2nRbF7097qubbxisyztqBnWG8aZ26wiRjEAdFcIeduE0QEv20OHVtY6 1khG4X0NZP3V0e8I8gN0s4zT5c2NaLYk1hJBt84ri8t04plMpmW+5er1faFjL5vNVNB+ym7d/hKZ DSx0/uq2+LHMZy7wihVhNUtfkxIruPNQDaFaheKNIsGEYODzGDhlZruwtcgmDFBVD1yF1rNXLr3o VcHaeeVLjoZs0jLEL8LXGEd1yfpj3K/wryFwF9EBlpsmMXFc8BBBLKJDzHO+wC0/GWYmVZclV/Ag 2opAtH8TVfFxw8cIt26on3mFVaadyNZmxqreJdH/aalgmehdrpmJj4gIjJDKOcuz53L4+wopl3lj 74tvs4yMU49H4Y7FEh0tX/bQJ5+p901+l5JMjUpViZkAwXKPjIghgjXN8NNrcjmgggwjyLskWuOl BqHCkuFl/ZWnQVYLIHGRMRsHXODrIxrORM8D+2tUu3zWTIC38tUz5MAHa/VePuBhmilc9pvDgDq6 XNHKKgkat3qKXsRe5YvzoERmQL8AZlE5GBdWLUusK2/e94o6m7X2ngfdyY/gy8vmZr8bZu9eyESS p6GyWTDFRwVg5ewxuzuaaTa6h+JRCDyK9NT1qkc8RZ+QlUZrDzzjn7WRnnY6jqb49zQj3PQ9JUEA F1kEq2zKstCBMCN6LjL/sCvXtwMCakgqDlgBWqD9L2FbQgE1CZorklHQg5oOySQ/t1NmMqRlrJJR CFiRLbuY/+0Ha/lWpzf8DZYrjEOEEyrcxn3WMjckiz3G3hfHqE86UVubYwh+pGl15kXHEbbXaWzY 6CV+ySV/9ahU6/82O80CxrpFSvELvZRCrQHEYGHIXiRbuEuP1qtSqs7nxYkkhRZy3EOWZKIqNZ6l I94AZGLp1rhOZmMVKpOUKXGe4liFm9LudX+pjglyLL/cyjgrunAJQ3xnoe/wX7ri/cZtUSBWBeeE WdA23S5zrjCngSPUSkbI1CobP4W9oxJj8MOdeeQ+Z055J+50/yxdbXZkMCn3zsVFZtXVoNZHAklf G04WWDdFczq618Y2xVFJW8dhuV3CHy5Te5Gb3jUzVZ3crJnxi+i9YcDonQrG06FfPMCc15hfqhCu C588Jia1wR/isXqsR3+cjLD2CpWFutBmYIt7ByjfkhsFvKugTkSnoZPvtOKdl+7nyywgnV5RwPvx DIMj/uAcGIQ1F4/AOEM/owYewoxQQQ4QF93//zx9CYJq7qfo/zZrJPfAmJi6uUFQIsFS1i4nKyCY 1j/5569Iu9NjHgy9+Vxw+yVs+0BdI8Ov965KcIuoLIsyEQud/XS4VdL8/Iaq6VoX/e9TUPJ3DMhg emR6EZEKoNcQ5b7elGOmpoF0/DOW+4CB6y7iPOR0OD52FKSiVNdSCCODfcPk1sl6vZUtOOHFmmgu X4BMb8QxdT8I/tMH5zwIb09hk6UHcJY6ut9YN61n16Tz3E7WR1eNuCvgN1aAn3ILtH32bJiR7QmO 9J5wFMWngdJgtMpoXQ70Uq4ZUsr0chezQM3VhZEeAvQOIkoK9dm7FO+DPoSZ+FMxIzlQqDEpHnTg MMo1BPULPn14/aNEhnZqGWTHcEmHh/kqDmCf31O+jEWVqipvo9hw3eocxglzqQw0T+dlcNkdjI3d FGXJKpkwbkQ4sB81C0+rriHE0G8P56nSY60wU0562ACZIDU3SqItVv7NiNvFUHVoW6OoekPnUUsB dGaIacp3eJGy2Ei4gkERrsCVC2sOPQP1JmW7hRechspGznnYfIkla+wjj/57ZOHI8uSiEb22L/F2 5b6MLNFDlONB+hcKw1QWsKcHIxigCY77nWnb3u5TTlZoLaoE+w1UtDb3LbDx83PL6JuQLHjhp88T iLY2JWb4VpYdKIqIRuE4jmfDZjAPbHcqxvrT4hlWeRp5Z5mMDY5GXjo/KiSdag9uPxpLKJAr9pUm /10fYkdxSzLgWS+S+z0ZrU0NUu/LocmW+AjfQO5b88+SgQvFbqF8QlFQrnvsGY6NzBWRuy6851ZK A88iJPMpoS03oy7mIhjb5kC05EM0KxhgmxrNNBltEwtoCjfwlBWsenlzlat1h1jY9bl3MaxcbvSM 0lXO53KFB2bRCHIwWO/Uqzd5P0Ir0cYQkbI6WzCafHRh7/DSG9O36Nq6ECtmk/wY6lGNH4Evqhdr LHhTOwEEP+RIw4eh7uPpkeVi7GVZudu8KTHFBn9zRhGxb6tsFbgEKagvJ5G7uv2OdK+O+Jhx2r6h cI0XVEkyZ2mOdMGR0MoXU0jsGaQ9dIQrkcUHwsHUdrzxOnRIlVroVqS+o39miXz2n7sRQXrb6QDx ACYxpjaTFyd9VpDtt1Knx87BZcQXzUMEMZbcrYOfdmZAtXOt7OlnQiN1+lkK1MB0Hk57oeJ8voU/ k3w7H4KnP7CTv7dJcc67mve+mTeGQFNg7Nnx3d4DwzBnZ/e948UadduXaOzddyOcy7IBzYTDu9F5 6BDD5LJBlHGFT4Fth8hl/SNMo/ptGtQq0CjX7dFUiBVBEWDnd+shnknspiz3SlCbReiW5nEtIJTo 3Ohstr4PC1Pv/aCdW5Tu2kNSMUgcYqAl+cNSHrNX43wypsILWYxJIJUsV+sdBGX7LFebcggzd0F0 C7Uvw4cbK9lmpKokUuVu+07k+qby2sl53I6wFsuOyQUL14yHDqTO6W4wNtLAhzH1bYKw/rUbti0o GVDtEb3P5Tn9Lf/IHJJB2UmkrIx7ng6+08FD1dzyoZLoB/o0zo3mbK7Ddtuq9AwOV/567GpnPaVg +8KYsGhosQF2ExK+h44xjm7cgSBAgOK/8rJygaoyXEthdZgl6vCbMibPWfL3BO/m215HG5pwxsuM D4kenVmy+jV+BecT2HcCT/K1Ll6nlMRZZy3ysQ3h4vCPcDxm/GbJOBX4i9IO1dgIW0wUmZVcunKh fs7DTeoySN6sVX/TnVJlD6WyZKDKr/SWDLez5uXKaQamWD/jAH7RPXcjvwDvhEAWD98Ibe01/e4Q uW7BAdzoJVKdE8Oy5IZKLdYvvTZ406PHWsUUR6A/cGMRr7Hh185ucJaAWXSyHptIDChNshVPrMkG KyVL4AkC6dcDlzZj4Oeij/QdjSc+wiKqXAAEnNFAa/JFqLQmBqNb5ru4jZCUtcb+1V4y/xSPy943 n4tDmVZE2RAJz3zsU6/r70FAGHW963KUU7BIB+bVpV3gi0rWE6Lc++2NyuuZNwVSRBzZjwwDoECZ Eeuz0wB8cUoYRwTU8nU+u2pM6SIV6Ip6Iu5D3+h9IfDpzLwri7YHowPv75Jmji1UC57XItnPVqTt rA/JeuWviuajg6RYxDjyzj6Wj17FCbnrNqV9Zm9hQvAJjQXcCVNryuvr4yJbPPbXRE3pmlAW/g+4 CAeEPH1CHQ8XUlySsyxYYL935AvgNcFavIDl6ss7TZMlFybuXEvvhVUTPo6yVYUybr4XcaScaWVj 2M/bmMO8D/Gzw4+YWrOx1mwVbRgp+SSmw5Jd2E4UyvLqDGTKmu5bLHbPm9XcozAqrgFGDwL4G1CD 1z+hXnN3qDu+zkNKerNsot+BpCsMTHMj4W5s/OF6FB7Xj2T+Din4/KV961Sni73F6fLDcKIqa9Hi BFyNUQsLdJb/Up+XxoTfRC2/XX+aPJhTU+q9LhwNqMDR6p9oEkuADQqHzGzVzQuKco5hyDKdonA+ GyP9wAMKeTqP2DkRo4t0oOoGf5MG0KwnootRTM7Uznqe/yX9F37UNq6eGmyySBNPyVRy3GTA1Xd4 tMEX2SxVkEGSgl0Tpuot03z+ckul7sa4WXc38LDlWOoNbk9ANO7vbumAGm9k5Q9k4GDQYmnIlqP5 Tht1T2sfbKhjsGZ7hnw/hEzqzMRovRtoX3U2KdzpweNS7lEpj1USjY+zQ8vd+6SXLaHsP/V7Vd/D Xa4gFRIQI97cp3IJhncs2RepjFVlr/qvQqDDzsuR8Ppx92yQQ84Scua9DMWvY0eHo//bVbjrkUs/ VDstZPcZVPabjGQ9esuu/s4OdAKsd2nKYh8fVndddT0ijEs0T+NjhmnasMuz8AVJxDE5PjFVlj6M luokgYCh3CCwP3XwGNSs/ontRXrdU1pnLfkU3QKsDrnnzRgf+P+bbLJPQkhDSXnDMDNvr/8xU405 vq/D7asfyctY2zS6g2QKRti/Gxjz9XGFhppYF+RzTIbrkhCOHvG1adFRDCbWYxwVpIOJD3f9c0Bv QAkmpUBsOgrgU418mAEZzDfgNt8YZooXuv+nvIPiQ0a3tgBUM6joQX7mStfZj4HCzYTJ3becrOa0 yHWkiMkCjyotqnVwOIhPMyB0ul/57cf/xxiOmrVnLoiLm1s6THNE27ra9j4VFLnKfz4ipToGiIiY dvMlh1we2bYI66BLcgdTccZhlb4qTOer9wqzfrJb/XkE9w3J297Rrsq2eh4k/g3rNVTNiABZV148 eUJ7aXQX7jDJZKvi6qjrpuSerR612CqOcDFdaryAbpG2a7Mr2pHACQqf9Q/G0S/ApR++2JjTJ7wx 6leyEiqYOp5XJQ/SSckIqLFzxZqV54dteUiUp+l9Y/FtSrhkaV5lTe5jHU/XDmic86u/MdOPV/QR DqUEf/gl+v154JSPjXhbBEdPiRqzrsdS+ZMNwEffq1xF91l/c9VPdxvzsVmHFQtbG4gcs/nFrXMm CQqQZEupqCSdhAJj6L+aOWlBZ3BASK/C9XQKTMYXJWfGhUeT62LGnIiAM6fQPOO/sI2+bfOWcZZb cwI/4FJD/QKukmcXdERJ3MhKFFD7KnmW2C3rCJSH1z+FVkKeHr4c4v5C60Aqvr2YMy2BAP6wDcIy hMgybBwakFQtu7DLQnwQxMDXhbm1zzhOOQEGhX3SEwm/NKWIO1NpvZfz9NtT7KsoyGHYHH47BgCT +gSNyQtIeXv99J49WBjixX7qho2G9qpw7g40pwY4NaKE4xkhNlHsHBBzRj12kvC5ygQowDtWoB4j b/qxPAvdoQTlJU01hFaAD93sDqp81Zzhsuxwt8ovkPB17zyyRQh6VDKzWqKEZLl15ew6DocGLT0w 5hYbQ1hWYZ+xtE8n88kPXmryXdrzYCMQtviZ9mJRTwQuke/M4eZjP/zaEeGkz2mnY4TvN9ljbICF WHEPRcXkOezjlY4POeCgcNQNnU9Qe/8a9osnhjgmOIB8ucCEhM1Mtc9+6vZLFl1PWAhITpVxmcUa bd118W15Zv8gYB2bRk/eRhRP5nNpLQ8pVR0MdU4LaaNarAd8iC7PwfOXyOUMHiTHBa13G6K5dthT dZBTjY9ha/1OVoyrXhJb0SyJUccBmXjv0T3hhTwB01BJVMjXA3AWWV5ZrVXL3EJpIUrmsjtHy1tI MgJyV0k+C3b6Rf0xkBmDNosDHulbTtA40XGv/JmJ9QPGvgjqJ4OWgOHp5FZ1epeXIWusBasYuIPO HXDxLvntZbFdhrGiWlIibWU2EGzn53fx+WsABlO0ixIHtGG/lzyLMgIaPQmrznnLHJVHw+Jga984 HRYjRgNWSAWDzj7FBK6JAahP00zbWFlz2Od0/dIuzQ5gTP2DFNIBpCWQy3xJRupuPSnMNBlLAl2q v1vuVtDk12bEDvMVT1tZf5zlJBRkQF10FuC738v/YmuWS91Lt5UrnRuIL+8kFrEDN/cztvgXn4Ha LD/sKUIOg0zKpeCdyurhIXgWpoqJKgYKwZu26cTMrT7+nuBO3L40DJB0uMehbbcFRYYaUUnrEIhK 8w0KZW5kc3RyZWFtDWVuZG9iag0xNiAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDY0L0Zp bHRlci9GbGF0ZURlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9EZXZpY2VHcmF5 L1dpZHRoIDkvSGVpZ2h0IDgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCqWUQNLv3PUx+yEpN79iVqDL 6Jk8nq0hWlmMbwmxZzACCHEYbq0Cll6njr43mtShzJsk2EmQ1cr9ruEQ+po19KUNCmVuZHN0cmVh bQ1lbmRvYmoNMTcgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA0MDMvRmlsdGVyL0NDSVRU RmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAzNDQvRGVj b2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMzQ0Pj4vSGVpZ2h0IDI4OC9UeXBlL1hPYmplY3Q+PnN0 cmVhbQ0KNAXp4t5fYa8XQlkgf7t/WD5B/0nJdhQqWV6z54g4rn6jb6vfjbofNexp5tiyJT4WmgzQ 1FRsY6ytPdjMpHDfX/pxHFsSWVZG+8KeGtT0vVYTwzECIwbID2QsQnfeWK3m23WsFAuXNAfs7YMn uEUoTcBBPhZb7kNc2NhoHDFni+wohwHmmbKdVNGOefIGuDFo38HtqADTNyoMxLYePLMxb6z50U7I dJLC/1SICOXSfAoDRApBgLW5C7XUQRg1QZl1AZp/bsmMbe7dOu8U08IalSuCsvs3DN1rGy2+pK0e vBASNf09Nvk3hMgdzuOm81hg+aHQxUcpgvW1Oe9cGkJD4eTBCLznMvDk2Sp6j0+C0AODYYrfW3cQ Jjdd8iimwUUGGg7InkctttWZI09Cupf5XP6KJ9yziI3jJtQXCpKEOZ0gKUKmH3GYeDEbiOHxfvCW 3q2bIq6KtacQOePQmMpLui2ued823HKkpDJLIzy++kqTicuMlYubi0ayazKB+nci5HFaoUOCqCc6 sNybq1lsQ9+T7g0KZW5kc3RyZWFtDWVuZG9iag0xOCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVu Z3RoIDMxOS9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBv bmVudCAxL1dpZHRoIDIyNC9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAyMjQ+Pi9IZWlnaHQg NTc2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQradF7MpuRklKu6N0ih6D3YGvGivRdGu9PKTj/DA5V+ yOB2u1K3x3G9bDfQhbAfG9ihtm1cNwLvLX/2Fju1VhQBZUF1UyacP2dXUgeyLUHHg4zzt7YVBbbn 9eRgWa+lx6I2dxVHSyWMDuz8X7uD13w+SEhSeHcrNaBAYKhtd8zxLiDDA7BX2NwX4VMTIivJOTGD Sty3qO84tpEFjcIRu+gvZm1od/ShUw30X+wVXH2weynfB8nIefLXXrlAwlsdAP1cQ1TEDF4SsagS oGGrjv6v9CIK77b9tGGbGx0KrgcGPDDXqLQ9FoNswPaSigZI1upur8KcOl6XkCOjdBqVuwpmgBVF CZZjC0O0d4sDxt0UCJ/Rmv4z2Kt6vP1/06vyWfPwfvkw4qJvWqgn27kSVdGkHW1kgrSh3DzFbojX El9vDQplbmRzdHJlYW0NZW5kb2JqDTE5IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMTY2 L0ZpbHRlci9GbGF0ZURlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lk dGggNDgvSGVpZ2h0IDU3Ni9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KIwOTeEBrY6Vwpl0nt/YNSA6I RgAtficiDylW1OlzLhue1f2MFi+sDMaoF+sonNpDgwrU4loDsFfaAAKS2DbHILABloYitmkxSwR2 9FXThyyS4hzc/ZTEp9h/+k5gd4H8PDl2ODZuHaZkKQOc3khRuPON6lgp23O+3Na0Vzm6I8aK0LWf WzGpMMnD5nEJVyd+XT0UZrjGFsic2Vy4ZbjQQjWbOK94Lg0KZW5kc3RyZWFtDWVuZG9iag0yMCAw IG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDUxNC9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1h Z2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDQ0MC9EZWNvZGVQYXJtczw8L0sg LTEvQ29sdW1ucyA0NDA+Pi9IZWlnaHQgMjg4L1R5cGUvWE9iamVjdD4+c3RyZWFtDQrBX224W+1E PHJB1senWm8hvPK/RYRLCuD0I++jrAR2awpQBVlDBeXyMczU9KZ1QSxbKJ0SalC/A+i8vj70g+Ho 8hNACgccDHkCxjU8PTiyGkppkLG1j79XE3THFmozaYsPnwIC/6qQAUPN/hqqkFpozWDpc7S+0EKY 6bUVx6LRWnZgvUREzRPFgha0aX1t9bWeJErnYru0uF+rAqZmntHWc/dYY7anZ4LBlfHf4/3l0Wc1 Tpu/ER3Lxp++CI7CTCNy0Vix9GCxuCkdwXcaHn7XPT8g+lwiLUG2p6gId4q8OICxMOKsW5qNyFed s+CInA2wQJRK/8XyuddvHDjIyyFvB4PPf3lBqzjYxeKD4V/IWs25hSjq+RP/p3PmKNs4rEFmo2JQ bLf+iHybuMbaMnlWQNYIqNBQpZ9ZpX+UreFjcyQwpIvJO2LPXNQHEsSUdy3L5UGI5gR4DmK3etgr IOSk2AvqHBMbIZdBCxLehBGhetRqx+uveT5MHaRp2kEZ0jG3nbTjmbMZF8Il/43M/m43NU+Eqey3 GshjZLgzSBsbMRxwvQyf6nYMH96sfSRhxyIBTtWIfCGLHSSTn659f5K0TcGhFL3jZoPzrstsHLFJ cDKjCDBbdUMbPLfkd8WL+jFlLRzmGyi4xawGYtOQwpxcMxPF2KNmf6HySvGo4H1CEBxQDQplbmRz dHJlYW0NZW5kb2JqDTIxIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNDYxL0ZpbHRlci9D Q0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNDQ4 L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDQ0OD4+L0hlaWdodCAyODgvVHlwZS9YT2JqZWN0 Pj5zdHJlYW0NCnOmgKHow2pbXcycQzvzXI+qFhdiSZC1L/B3l1/1rdscQt0uddkHe2cQd8NtrTs4 UdNI6MgJ1ssH7MmDQfWoqfdhsnItUPPJNew718b1n93fV+NOanrkJX7P/TJkLxzyNggGW6Md9aKn 2Df9IllSkHOA0JuWlTAa9H5Bg/+LhF+AI/JobTtczdiAuci8xomEFkJoEJSlunePLjRKXPQrojXl yEElVNzepwp21f0QbacSalfbFeWXf/ylbEkSq64AUzLDbtEiGez0sa/O+0hFXfpvwACnOBWDTYOg oQvrXp9vaZMK2ORiKbxG5tzUs7af7THw1sw/fxaZFmijfAzgaJU9gZUvxdMRPvV9SDkELQUgvFC3 0CN3Avv6s4BlUI+HngDG1z8/aq1MPbI0VmkqCMvAwaa5imZ5q1nU9mXjFyPf6wzMXHUPN7LqWQ2C oM7pzXTFRoEknh4sCKyyy7+7muVahKcs/Yeo4sWFHBmKvuoNpzcA3wfvByhCpOPIcfpuhuXx24ef wL+FZOGz115Dv8e68StXQ8l1npWbQPFcuTQ5Klg1qrc4f/lHv16i1O7tA6n0ijkj4xOZykGMrRIu +WDzFbyyJE5kRKiyHFrtDQplbmRzdHJlYW0NZW5kb2JqDTIyIDAgb2JqPDwvU3VidHlwZS9JbWFn ZS9MZW5ndGggMjcyL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVy Q29tcG9uZW50IDEvV2lkdGggMjA4L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDIwOD4+L0hl aWdodCAyODgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NChAtylz9cTBSztkEo3PnP39o0F7stVIxSvm7 +Rj/jOKTk1/WDlBVgLcQkr7JW62Gxo2V9KLUtIH2Brh/7xCOGCwM9lxsKzS7s9ru1lZ6jGmaRCz6 4NIGLD8ogD2+PxmHTOMC4Q7pQxF2nFwq25ZFzZAMCn+B0SIkLInOjDLR8Odd0DavaSnK7O8sVwC0 RZGblZrwZgIqiWFEwdcCSfGejNUppVnUSMD5RxSvnlAWjSY/JM4gAsG54rGfSvPngVlTHZBW+ZdW hx9U2UW+0mudVW7KhpszFuLnwp/D8leLmIC0TXxN7ndqo9der01X41hwC/ZjrvdsVJF/9Zge74YB OUV1xY9BRkHKuau2szIup3hSDQplbmRzdHJlYW0NZW5kb2JqDTIzIDAgb2JqPDwvU3VidHlwZS9J bWFnZS9MZW5ndGggNjYyL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRz UGVyQ29tcG9uZW50IDEvV2lkdGggNTI4L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDUyOD4+ L0hlaWdodCAyNDYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCpdeYIzTUdYTfqZABTLcf9X3Ej5Ijfbx HlIi2SGmiadM7uMb+HSTYWpT06SnslZ3/rUh+YL4xDy9jrnX571sc2YDXR5by0z3Ie7MlXgp0bnu P9PgAUttsO58Y0LPVMXMslloMhSnPke3zkbP+W6rx2O5A9RTfSlmqt9/S/1wd78hTUslT0JfIXSS 2RmjWCpuXP6kS+nRrmfPJl0Np02SK0otLBz4FvQ0ijbsxzkgnpkGOrInnOFMATlNmJt0UUIZYGZH ESivyCw/M5eerxYxk9bNn+/eMpHX7Q4XCLARnMdvgFUvBX9s9hGuDFkdII2U2dYJmIpRQ5brsX4K cStYq13aC/zoFokQ0uawVeVRJhxIQ6PXlfTc+ZcaZG1K6Gjfti3backANbisJieAlVF88yKVORAj MvXcvlLbjnll/bVC/LZK4QZYdpZgt4rxwPIwDIcUeQWtfhF+tlkJoE74nIXBz8XrN5PSYUX62xvo TpY+lA4ChJqVjUimEDqAkcFuQgrMK4egAWMHgAeK1cwAEau0aMCUsUINmvU3gMd0rrG5G+oZP37H SU14uqt6EhlNuTdFe8OKvn3JEK45q+oHS1QBvk12UuX8Nskq2hlivNlI9+BhdbSwx6TjxOUJqw+a nMdDbhsdrqoR8udUa5ADxjFHqMW9jdvVdLkeC953RU6eRhUdzblFIOv7Ave0Zvs4fdmGEXcX0IN2 yYSKyaoRXxTStQkjJog9bmVBt4RhM2J68ITBk1XzDtgKnp3ADYWUzzNQp+b3CEc18VlBN2Oe3pee 5MyEwfZVLOSdF+fgykRwugeXogS46Eo1ZZnIFJ01apsFUfZQyuMsESc2fJEdDAkce80S1+RCslGN xAeDOpiCZwcfJ/mDDQplbmRzdHJlYW0NZW5kb2JqDTI0IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9M ZW5ndGggMjQzL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29t cG9uZW50IDEvV2lkdGggMjI0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDIyND4+L0hlaWdo dCAyNDYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCjODj7bqHU6vfgWZv55lcmtR8jKTeffAc9w6RSGO WBiQ/AKGd1KcV+aNYHxJLbkU7SQGqM8kPopy67zJelNVVMikU0FPVQ7nCyeUxhU/J5hiZhRqwsyb TC5BrFkN4lA5ov6Ag8PU1J3qaBUMr3JWn25NyjaN/g/gt6ocHm0OCb2yrVgtIL1BdjZ7qBsQg60j NtH83zgqCszcjhqcdGgN4xB8P9kgbtxla5vORC1xMRfQLaN1KC3YPNwP1a+51tkuJlo6I0I5MgUY 6DoTTihfyGCjJsV12j5DiL88f8XeGGzh8iGmh0g5oB6QvpW/FDSZaqYuHg0KZW5kc3RyZWFtDWVu ZG9iag0yNSAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDYxNi9GaWx0ZXIvQ0NJVFRGYXhE ZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDU0NC9EZWNvZGVQ YXJtczw8L0sgLTEvQ29sdW1ucyA1NDQ+Pi9IZWlnaHQgMjQwL1R5cGUvWE9iamVjdD4+c3RyZWFt DQrcUY95uXKA/BEtApNL42AkwOLVAR2MY97dfoi8+jBbJPi0/IDw3wDdNNH10WwObC2/sgsDK3nd keVqcEFnSbSeKIbwttjUBEedbNFXyqY+iSt3qpa4vyjfQVz5bd+iwXuvoqxgAg07gOPD14sqJNZK NSUyeR/mZ/1A+NwSo0Fmxzpc2yGDXUuRNwaPbVqtfR3+cnZHVkZomExdnAqS6DmV4BicDHN6Utdn FQipvI10csx1X0cbnY3c/+VJLstTE+Ua4E7Z5dTv+hRpywZ8RMDsEtoY5iTp8MtigoYih6QwlOcb sjtZwgQxDuzl2xAQKIumf/Hn85DThiL7Vc4w5l8a/wtexAnoLeX0DT+wxzxeUWUYwGVEznicNkh+ 9EqtTXUVMCYRt+HA2MCJz73JVuTE+BajYJgKzEQzrBwGo+wB7enGJV++Nt+Wt47WPGJEK3qnD6J/ 5l7xNANhItRdJ0FaJEGrwni+WqVYCY/GcO3c6Fm3kh2L7JVPeM82wcZ+Z77hpZjzbfpq7S1kKI+C A+oEfsyoP7CZHM4Dp5piBEhEYzbWfs65srZQr4AyeSko6sIt8yVOqjFg0t1uNvJMkrSV7uWG+aF5 xeWv0FMEaqijMFZ8OoPuFvzcxcGd7Fp6FN2sJiz4wK3r4hegqn2p3R9+slfOL4HWnmNpldjqlNCG 5WQfuIT77IMCKaZO5IueNKBVMZb49wtx0WHQ7f4Z1fOLYwYeagTAyY18VyxdpMJ1b2ImbvTlXFAk qZ39W6RKVMxAE7wWKfY8qszW2hD9ZGK1VOyqty5+5Ngr9AZxVCv3ujSnvxjIiXBJDQplbmRzdHJl YW0NZW5kb2JqDTI2IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNzQvRmlsdGVyL0NDSVRU RmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAyNC9EZWNv ZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAyND4+L0hlaWdodCAyNDAvVHlwZS9YT2JqZWN0Pj5zdHJl YW0NCjGrIFSCYT8WpAGencb0RzXJz/jNFqCo15QSvjPd4ZQ/KaRtlKojYqGV6ZeYwsiqucOLxYKU QcJd358dM+nZpP0eCd4Uc61+ZctaDQplbmRzdHJlYW0NZW5kb2JqDTI3IDAgb2JqPDwvU3VidHlw ZS9JbWFnZS9MZW5ndGggNTU0L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9C aXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNTM2L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDUz Nj4+L0hlaWdodCAyNDAvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCvziQ9nr1t5nnGR6S7rT0q6GXNGb i6gmskLlneI2TKEal+8ll9FhXKnx1Fg0ddEOz7yEzay5IIL0Jpn7x7xNMear3zYBbvSjyn6m5k4V nJgzCvbOeKv/EggHchOGdGcN3yFvxePBV5SvyWKhJVJvRgVII0JmYGQkgWWQHS/wSwNefoJyDqTx zrAThC6txVVY9m7p0CLfOhgIAcbO5JBM7XE+ulPsO0x0E8Tj37Vx4eR5nXjcYB/lUgxZjnqhh4Uf bm2j7QkmgJv3Y6Tj2n+0DvaVJNYaD8d/f+y5MbvkNcuynANATNpWEEtEzUbtTxa+KkHVySdxe5yb xNjrh0RU7DTQjZWJHEfzkMCKgExi8NxlLIDG2qyRjSur9tuI2SjCLysamR/wT6uO7U7rTIY8stb9 OSOdQU06yq7ozFeVby4aO7cWVX6mqlvdLIYuy26H+BhZ7wE+1ZmKc5l0n8jlkNR3lIldjLKMMzJI 0r7+imPGbJNm4AK+k6EQDEn/XdRBuMQnynYcH9QCwjXU70THpNcMW5k2/EG4KuLxDdNSUFatEm64 4kfmX3c2uOUhVE5GCy/raf0K4dx2YqLFSXsun5lQjHJ7X6k0HHSmSdWQUwV0o30wc7cz/2zeyodH sVlJWCw9rNb4CbRM00OJH8aj2robjltOfF9TAWQhwk+01q36M4shT5G5eBj3zpg4ev5L4vlhB72Y grhm2EKQzQ316J+53/mhhC9NwzjUDQplbmRzdHJlYW0NZW5kb2JqDTI4IDAgb2JqPDwvU3VidHlw ZS9JbWFnZS9MZW5ndGggMzAwL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9C aXRzUGVyQ29tcG9uZW50IDEvV2lkdGggMjI0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDIy ND4+L0hlaWdodCA1NzAvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NClTG8JyBR39GSh17hAW9ETP8Ppyw GH8UNEERWP/rgmIcf7HHNcrC2dlWkW7qvEeJgUS5iYZBQF9gs+Yf+GLT9tof6PMTtsBqkmrbnp5D 2OVRzy8ATj1Er93MvBuW0xEdgKjTPvw4OMxUUV19Nbs6jfwNBbE4R0IWpDn0iROa4nxLhXSzXdXJ LdsmuUZcSREac5DYBIcdisY8w6dxeNhZ194U9xkp3bI3jzMyh+anPNdjuZa/wuCrAECBc7Q3/pnb 8ZRd9d21//9UfDJj+1MlBkO1SHIqJQLsti5EzzWOePU/P/O4jvDCTS1ZXfQ7kHptulH3XknZA7Xj fLw79YNXDhYOuSnTXGVWO+I1DjY7I+gs0KWeut9YxLto9xbLHupZqJs/0vMh2+Jscat7Tw0KZW5k c3RyZWFtDWVuZG9iag0yOSAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDE4OC9GaWx0ZXIv Q0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDEw NC9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAxMDQ+Pi9IZWlnaHQgNTcwL1R5cGUvWE9iamVj dD4+c3RyZWFtDQoQ0hIFoQfHBtXAZMfjGMytDXsfWOqga7UoCw8EebhvP+Nn+NoZlKBPsj330xNL NCEwe0TD6ifQ4At9LV1QXCfSqc2TPnb3ItkdG3jHlIVsKEasRYb1flGJpzUCBWxbuSJKyhjQ9pXE vfWh0ZJUUIi7DmUO9sgDXexxd+c+BMPzK8Aybo0abCapEUHJr6mAIVHONft2ZkDo3mw1a/044Wyp eEKF0WgGazkDUDHOBCRLCskQvgCuZk7MwSi+Bw0KZW5kc3RyZWFtDWVuZG9iag0zMCAwIG9iajw8 L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDQ2My9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNr IHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDM3Ni9EZWNvZGVQYXJtczw8L0sgLTEvQ29s dW1ucyAzNzY+Pi9IZWlnaHQgMjg4L1R5cGUvWE9iamVjdD4+c3RyZWFtDQqqMF1PIeXtfRFpWds+ 1XdkD7fE2QMzlp1H7e2DOup1/Y/Q+UNQllT6lbsYlbfyrrnbciDnYaHr8yF+cIL4Ah6j8oiwA+HO 6OVlMD2TUVqq4R+N/qJ/QL+MkScy5417GTGoIwE9Z/fZ2f3rl+s8EZU4ZZvty+aTAQEeR3YUOuCh s5qdSPRVphYUo9EGg6PmyvdAPN/9ayvzQ+4c5DlG9x+Sy8OhoEBGwIBkYQhaz9NJ9Yv6/7jtEw9a h9srLEoXM0RKNo2BD1eWnB9VM+EEC8Jaa0WZ//JRX5IaQJK7qJM59JdeKwUlAGgNM7P0u2q7wYOE 3Pw0FulBiIv+LcDukGqVGTvZFP1DRcbns3ToIShonk6x+On+hNuPl6H/PatJ2db8+YC7O6O+QSVB IMU04mT0avp6z4NRHrtqKqiPJdJRexVnUeCQI9soSHDrfApwfC2IagTWq/xbsiOMroR+cQOnRNxV 5HlRa7x2K7/cVavyEF7W9fDkAvJtH16HbOwGIXUhQxGUTUdMqNO6qb2cZtTbv8jwh6+6IOCkhhpd ASh6GiijlfsTGvHmfuAlPDvT8hlziuG0rg7XOH9K5ETPt2bLyuVBT0nsLknHtVVwrJYQDQplbmRz dHJlYW0NZW5kb2JqDTMxIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMTg3L0ZpbHRlci9D Q0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggMTA0 L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDEwND4+L0hlaWdodCA1NzAvVHlwZS9YT2JqZWN0 Pj5zdHJlYW0NCpZdPHWRB3TDN038POZvkR1j3RLF0gRlJcY2Ihx1I55bNiGwSAT7a0LW/WgqtTod smH2zG0HmLhthmIHEVnBb9u3n9CU47fEfAuELCU2rTe8d6b9rkjUEj3rVn9tS9LZl5d+yRAnA2k0 PuqcLwkaHk07Y0+J0lsThsJAAWwgRE5V/hSUVfAalY5diqYlkUFjTUO07cUBuqpQlxQb3WzQvLJw iX4JWYOuuHstgknxkg5YUNJDpRiB1mVMctoNCmVuZHN0cmVhbQ1lbmRvYmoNMzIgMCBvYmo8PC9T dWJ0eXBlL0ltYWdlL0xlbmd0aCA0NjAvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0 cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAzNzYvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVt bnMgMzc2Pj4vSGVpZ2h0IDI4OC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0K8XtDO40qWG1EaK5Bv8vd dgbOoZ9xNwSFBgbUNWmwnhYF4GxiJEOvuIGQssKbTBfrS456qqzEBb1GRPAu9+0FFhGNUkoMfsLR LWjpExBreuX79vQfUXkqUT7eRLZd3Yn+9P9P5Srme2ywxiWH3SRxRZCwWmaHruiYVR4SmqTz3J/N Lzr2Y4/WUD/M7SMsULkZGSttcq6bdZ5iWHPLGx3sqr/QBcgY9+cPYA4Aiv00bqqxOdyHFIZqnacG mUUXix0F4crxv2tjqun5Gu7awYF6QxfK6xGx3IZqH1CPGRBRew22BfCg5RvI36aBsYwhkPz81XJJ i6exzqPx9dkKt6yg7+Flgfa9uuoUHi7BCiy73QDdo4b3Z1TxJaOOOVD7TX7pvQcEKbx6hqApA1pC adA1+z83e1Q7+tr7qYdBnVkVmTxiExJz6vOCXbiRd+d3LYrCqzN9YX86KF9HR+rf1MscnrUZatmi WVi4bWmIqtFAglFPz9vOqe08dOLFuQPmPr2Y5xR0gcVG9ACpn/zGqFDCHyUYHq+ah/vEkheAIsBf srEH04WycChQ6kZ+hSbs5pDCZ/odLI/jjlpD56sMtu0crIRN4jdxg+WGpkem4A0KZW5kc3RyZWFt DWVuZG9iag0zMyAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDQ2OS9GaWx0ZXIvQ0NJVFRG YXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDQ0OC9EZWNv ZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA0NDg+Pi9IZWlnaHQgMjg4L1R5cGUvWE9iamVjdD4+c3Ry ZWFtDQpDCmCKKgwRB8cVM5++KECpJxl8iRIhPs28xn+iZ0Q6BV/skY2snemtFKICsPJ4OcJez5yb egiPw1Yj7RQoVQP96oCMuRmN3X9HYCtrYX4lErNf/8gQmjHSD4MeHexTnNigFuJOyCjRD4VUZy6x seR4Po4KY1729VPf0wZzNKoM0dKz+dYgyFzDRjrKiFlK9rqJruyT4fiKar8lwGTYNIkzgvFdbz3/ RjYMvTGkZGgxxIX7ZoujwpErs+gjQbZ/r8+PVelyjeEG7RTzukP8Y06IQPMLMQ5YNOdI3NsXrS8x XLVSHX1E9PA0yf+USy0ksVpwKGNo7hb+K2Wk65pwxgnDR+6yoXILYl/6tuGdkIBl8gDG7m6Ie03S n/CL9rxlegst0MYXC8MiFcYoRkyWadhTS3Dx10CgGAtqPDg4h3ZPqcHl1ih0qKh5JSFafj+MDY4/ sZjUH/gCkI9jF9K6e+XT5WEMaqW0S0B9T3enEKJSQ4sAoMZyWYKxW134/H+/B8OMeNpmHoELPpsu OxclB3BGxQQ3uYYSiUoZyqfQUp/bJTVAkNfiWBs6otABLHRSgbNq26ejlyFclJW4JPey8sjElZ1p 0QVNBiQTCbPkugov2//He952DQplbmRzdHJlYW0NZW5kb2JqDTM0IDAgb2JqPDwvU3VidHlwZS9J bWFnZS9MZW5ndGggMTI2L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRz UGVyQ29tcG9uZW50IDEvV2lkdGggNDgvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNDg+Pi9I ZWlnaHQgMjg4L1R5cGUvWE9iamVjdD4+c3RyZWFtDQotetfYkdL5WiXzXjv/Kx0lF7JVUMbUIlqg IcDmRlmQ/pjN/+Bir5xInN7ITcmdJFw0NSMZFII00rJzxZ4gAerh6sHimrrJ1t52ChkXMFuURAGZ movsc0hCLiyPw3+qCV5BRBY/rq0OUjTjulTdRQVln0r2eQL8XGwKadpL4AYNCmVuZHN0cmVhbQ1l bmRvYmoNMzUgMCBvYmo8PC9MZW5ndGggODgvRmlsdGVyL0ZsYXRlRGVjb2RlPj5zdHJlYW0NCvqm wiLzluwlkFqcnGVzUuGP+GBMXQSAVabFeW/ECP1bDpYDuHXJzhOz5LtdOwRje/EFd7ynLekhBVRH 2NLdJqzaeYKmeUiRhoYRS3jAI+njO/UFa9jYZrgNCmVuZHN0cmVhbQ1lbmRvYmoNMzYgMCBvYmpb L0luZGV4ZWQvRGV2aWNlUkdCIDMxIDM1IDAgUl0NZW5kb2JqDTM3IDAgb2JqPDwvU3VidHlwZS9J bWFnZS9MZW5ndGggNzAvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xv clNwYWNlIDM2IDAgUi9XaWR0aCAzMi9IZWlnaHQgMTI4L1R5cGUvWE9iamVjdD4+c3RyZWFtDQqX pjULjD27/GIi0GwW6JyWCULmDPcKVrQ8opKTwuxczB/sHI3VbAleFJYhqDxSvuPgi68khXrpJS9r TroCVl/fWo/8XI0YDQplbmRzdHJlYW0NZW5kb2JqDTM4IDAgb2JqPDwvWE9iamVjdDw8L0ltMTgg MzcgMCBSPj4vQ29sb3JTcGFjZTw8L0NzOCAzNiAwIFI+Pi9Qcm9jU2V0Wy9QREYvSW1hZ2VDL0lt YWdlSV0vRXh0R1N0YXRlPDwvR1MxIDM4MSAwIFI+Pj4+DWVuZG9iag0zOSAwIG9iajw8L0xlbmd0 aCA1Mi9GaWx0ZXIvRmxhdGVEZWNvZGUvUGFpbnRUeXBlIDEvTWF0cml4WzAgMC4xMDk3MDQgLTAu MDQ0NjgyIDAgMTE2LjE1NCA1NjQuNjldL1BhdHRlcm5UeXBlIDEvUmVzb3VyY2VzIDM4IDAgUi9Y U3RlcCAzMi9UeXBlL1BhdHRlcm4vVGlsaW5nVHlwZSAxL1lTdGVwIDEyOC9CQm94WzAgMCAzMiAx MjhdPj5zdHJlYW0NCs1a8kUvce6L0+8AS8P6W5YBs/5fhPteRKwwzKF864bA39mSi+iWh9LElPKa CZow8a89/RENCmVuZHN0cmVhbQ1lbmRvYmoNNDAgMCBvYmo8PC9MZW5ndGggMTExL0ZpbHRlci9G bGF0ZURlY29kZT4+c3RyZWFtDQpo4znWRijsg6gVPzClwLCVbEZlDso9vGJSI6CMUvHhH9jUeF/J vMAwvw3VRlqICx/a7c8jgi/8Xm8AAurFsMG8tozaYF9+6it4CrIQU0I0n2QZ57gqSNuRYpHd6vPJ ddTDbgxzWsj9Hp7IB6HSTdENCmVuZHN0cmVhbQ1lbmRvYmoNNDEgMCBvYmpbL0luZGV4ZWQvRGV2 aWNlUkdCIDMxIDQwIDAgUl0NZW5kb2JqDTQyIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGgg NzAvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNwYWNlIDQxIDAg Ui9XaWR0aCAzMi9IZWlnaHQgMTI4L1R5cGUvWE9iamVjdD4+c3RyZWFtDQpXA4hzlnFO9jQp8whb 3eRQD1rsnQmIBEZw+hqoo/1iZSomXCBX3LyAKteve4VbXZLMT+n8sbCxumxrfTHFXTQ7x7rd87SZ DQplbmRzdHJlYW0NZW5kb2JqDTQzIDAgb2JqPDwvWE9iamVjdDw8L0ltMTkgNDIgMCBSPj4vQ29s b3JTcGFjZTw8L0NzOSA0MSAwIFI+Pi9Qcm9jU2V0Wy9QREYvSW1hZ2VDL0ltYWdlSV0vRXh0R1N0 YXRlPDwvR1MxIDM4MSAwIFI+Pj4+DWVuZG9iag00NCAwIG9iajw8L0xlbmd0aCA2Mi9GaWx0ZXIv RmxhdGVEZWNvZGUvUGFpbnRUeXBlIDEvTWF0cml4WzAuMzExMzgyIDAgMCAwLjEyMjg4NiAxMTAu MjUxIDU1MS44MzNdL1BhdHRlcm5UeXBlIDEvUmVzb3VyY2VzIDQzIDAgUi9YU3RlcCAzMi9UeXBl L1BhdHRlcm4vVGlsaW5nVHlwZSAxL1lTdGVwIDEyOC9CQm94WzAgMCAzMiAxMjhdPj5zdHJlYW0N Clw/u7incu/wj68fGF5DGVXlJTXqrn4TRJf2IfFRpgskJKypAmJXsHbMr73lqhUIi41Qfy9Fw9LO cfmemrmRDQplbmRzdHJlYW0NZW5kb2JqDTQ1IDAgb2JqPDwvQ3JvcEJveFswIDAgNjEyIDc5Ml0v UGFyZW50IDE0OCAwIFIvQ29udGVudHMgNDYgMCBSL1JvdGF0ZSAwL01lZGlhQm94WzAgMCA2MTIg NzkyXS9SZXNvdXJjZXM8PC9Gb250PDwvVFQwIDM3NiAwIFIvVFQxIDM3NSAwIFIvVFQyIDM3MyAw IFI+Pi9Qcm9jU2V0Wy9QREYvVGV4dF0vUHJvcGVydGllczw8L01DMCAzODkgMCBSPj4vRXh0R1N0 YXRlPDwvR1MwIDM4MSAwIFI+Pj4+L1R5cGUvUGFnZT4+DWVuZG9iag00NiAwIG9iajw8L0xlbmd0 aCA1NzI4L0ZpbHRlci9GbGF0ZURlY29kZT4+c3RyZWFtDQqdzbbrqicUspriztzDD+PDu1dN6JBw ASFXHcMTomUwYoR5j6wHDF5Vi/zFx8H64OR3bE6gjLyl9+t9HK2t8PkeJBWqi/mM+LwtRIaoM/NK TcojK96JEbZ+SBIAFsJ6n4/r3BR9/7sm5RonxYmiSQcgn2DOXjrO7oJPhVv/dmYONMPfPRndwESP UFr4ABSJBDMmkZVBj4++/TH/d0rZlBwwU/NBb4ETpu2CEQqa5qAIQ3IfaJvFfglZSxLfwhernKco Xro4mylPH/RCoxesMhoIbIAAYzaCJf4/AwpDtoOqb2UO/HvmHzfsDJ8R/tPPU39OFyrLtmQ0tr54 s98saS9MhGGF4i2EXlP/b5lBvVAyJtoz2z0US+16kENP9JT88ztnQ+28deyP9ToMeauaBDTqpi7I /K2EVzvGC92yUkSXhkhMMlRMcTNJ5aG5XnPIZz0nJGoQHraws5AAvNBy/kBH18LZMxhuprNOSse+ zVw2j3EWzPqtp9tytJ+PBALJBVd/yHghlBjmdwzTMwZHbbWzkchQnv+KrDwiFBgikIpcmAi1K8qq JpeVZXcecXR7wak6bBgqYd2w3Axjn2q0oQHMVFzDQ6PCUnWHoFZjjGOsc6QvQciIQgirue6DB+eM KuFtjEK4FPmdJdz+gLILG+uNjqd6RDU1V/FkMEa0FRu98hljqVJnQorriGICLQFrRRszall8opKh 1SGCOj2wPad36PhwNjcIDri3MML5Fw2tW5f1Uz/ekqRjiaaCeZ/E4AjgjUV6VUlCpEFqSogrfDGJ 6vaJMeor/vVIYhIq+SJMDhsIN2RtDRLVw+e2/yWdn+xCkY2ISe609sNvKGotEBT7/6ipQNF5j8iC ilFPUQC/KulMSA5g+wtaKwV7wcDWeMfkKZPQ+kJQt4QRpX7+zV5/ASFdkST7DLQe10TZ3NL5nLSR Pt/NKaFoGYsfJ6bhNG8pdv2dAvHHy6KtukoU2RSTDTK1pQLdH8UzvxPNrAwwMdgmwr1MIdbVmy1I rfQVrD5fWXG0KOGDbhFpCoHWQv8Q3XnKr/sez3yTuwKQO/nWVDcgDP+Y4b9hBHFZqgztVrrcE0WB 5lHH4RIni7c0fcijeThoQOGbZBXNMulDdeZSYw6Nbd8/hPP5LxByyOoi9hodVoBh3bB+Xr8mJEZj Fa1GUbUUKP7XEyGJ9AZoNAEX0k0C9FjuDqMBUogOJTdvk6NZjqt7oIbYkp65vY7HZw0kvJ2xkbeM 7ZPjrr2+liOtqMbP6eWr+XXzusPqf5nOiHZB5cajlkyUgzPdvGQvSy0VuZn4BRpeMq4CgkKqISLN BV7ghvcXrn/+6HFYdUP9xo4FT6RpTl7tVR7ZQ6cSUyuMsScJWtGdaLwX/pJXRQLCXxQle9nCaZ4s j/g/oROStHb6uGHH8x/2VfqTrARYdjR9XlYTPRd14xEBmR9CEcbczf6jMkTUEWegKXhjFvQNO4nL 66iLrmRYHiPjHPMLM00V7QMatadiBa0h1MYZowjBfuCafUER/hHzhdkKNSLNMU7TKZmcpV2CC2gu LM38ASgHjqpdLA7QceWHGr1yWEzg/lD5FLPGdS59FRP0M5Z6RGstho0gzs3Ggzm0t8gw8GMgwGpu N1NIKvSi+n7ZJZzHEK9l1ZYfLfFTBWpquFNnBUcdGE0ynwX16xMMjPSjRXz7cAhpO1j+QSUa3LSi 19imPnSN2bb06enP6KPwM26xNCYYVOgCOW1j+vrm4BNZhZdW2upqVcDflNY6vHGbLOWLk7dLllLZ 8SrUfS0wMDOvV1IMbdevHReS0iF1xS7et7uDJc4hrI5Mx+xjOLzSn0EaswK2YKkZIZAapDgXS4kk wGr7STY289Z9oMcily/J81N9df9NDaems2Pts7EnOQtpkNCYtpF8fwsvgkOunCrAIDx3AvMFWtr6 yBum7mIiN3JLaxiABFNUQg0gBCFkptCSdLArVLspIAIzgQyWhde4EeG0qdkIokfx6DEmGUTOR/JN 0L2cvIVtM+f/tEbVTcumK2LdcTyIqCfAMI0AoWvGYmIHhZw51lPCS10K024Eut7zd16jXf821J+U +7nrC/hauN8yTGanAU79VB98CNUDR8sHPEypgVGBgZaWuEkxskVzxGaYknnbXW+rkTjU0cqGo5rN I/igqfxhwQDT301xVvWvc5sE7P1mCMYldVS5rBY+Up7FsunwAwgAGDhABHimBprd19VT88Kt7sOu 0+BFHuDbAY8ZMKCpiDdPPmTAFQckW2fpXO/9lLQdr6OnZvP8ryHnlybvRJHwv6Hzfw21L7koRRDx eweODDCMiZXGJz0xNrt4hhLLFfpwJNMC3IV3xWxD7CfU5euaHT3ilk9tD6dg5P5Y/XWhYVRZisdD rPaBN32iM3ndMFpmVq5+b7mYDOgMGwoCCg8OaZFBsDRVyjFwqp9AJALnInSoyw/ztPiT/HxcWzDU ruoHuZ8EI9fU4pROwSfLBJaJO7oWLtaT6tuI11wptdwP6ILcm+Gfu9Nu4Dt7wkLwGJAvs0epXacj b8Sb/S/tsAE43WkRjZsWPO8zF65hEivySZ459ZIkvurw2I9esHq+GkWagzjB9xYJWTU+gJNNGNmB 95XiUY5WbvjrzFJe4BdTmklFFv/10MT8m86aHnp8VwK9mNBJQ5R4vBmMDbL5QasvznQ9cCTd8CBx boOyH6QTYqtHr4WKEvFt7ErlcRzyYD+yHby/dP/tecDv6QcbSZ71qBuPRsavAbGtWTfuLJrLt3fy pcnw4qKwbZA6Vulhqf8wLPoPOMWaTa7XVp2+lBCnf776Om9lRMSNkkDEHFs1QgXhVhfwRq0GSln3 2Q00rs3yeNOerPeSrKRRU0m6rub5Qia83VtVvqrKHjXDeinGKoyKZDtjKIQz+I51LbRpSJRLmgXF gYCyow6kuPoo+LM2IypPDnieCdZtxsJwtnejnAK71KYIxoxMaWTC025ePkdHEY6L/nlJadYT0abz nRCy+skrQge7bL7qdbHozLjM+M3eTJSA4ORafUgvnGkApclx2Li5AWdqBHELR/Dra5ijgAX8rVwk gljuo3SGrYH3XmTWpHjqqInEGdpBmmmeKRf8NR4cxKf8YcYb45x+KGPJAw5/bCoEi8u/yrZ3ojbm 8kztTTcCzq2H5cAdHbftj/9n7vr7l4ECYoG/CY5DcC+IWyCBOru7aN+WefbvxOtSUa3/Q/JEHgky 260SZ/VgRU+B1jwsoJoRWkLFGdn/IcMZOuKGNSlB9ZUHpiy/PJZThvsqCPZexPBKHjR1kolK+Jd9 4nlxO449zy3rfiv92UhsyDjDw5oIxEIybc5P85D/216W1/4aPBBAKDQQDCJyrTG6lGdbHVjRHEtf LxRBKQ40pZEleE9h3nWREfnP6pdajSh3T0FtXkEm/dNFkflhQdVw/bzKs+6OB8nXnvZMyyUyBS/1 93SHjS63YFy0kSLAlylXDGqItvZIh5kPtA1JqyAYedOwTFfY3x1Q/2Znr8JiQau5eIfkI+VJFnU8 8rRDd0tTKVMNf5iRu/UrTwfNr5XVUqt/YSa1m9zFonDqMivudK5qwa0PlpyK5dFFPn28hkLAYTjR 3sBeKGSWFZ3mo8hSj3y7WuWj4YaZYZevNUjZKGyhilxrXhT3be8kpErwYakhKzJr6xSLqr42dvy8 AEh1sCGXGt49C+0wcjAlmB69yLfh6ihGby0VqEHaOGIZsOUpRiyG2DINeRQXGxl97eX4+8BEcXh3 egXHFlisNgJpTz2WAPGFVhcj8aRAQR90SM+sZxd7vvLVpXsJkzqIHNhBzPo35rORR8z1zoKskSkZ i/65Bql1wCOhRLSzZANR9sQVwnXYGzoRmuENnnDdGX3wxFTqWhWzGZcPfoIiNUtX5tg++HjzojMk d5mYYtSvP6/L8qbuOVrle9M03pMf9Jd9VU89NRD3pD57rUNwzUiGTkbDcOgyyOW5KsosppyHGtOr zOzmDiN+gHCy0olOuER5sOyt2mF8Yd0GBbc+lTvQwOa4CXdQMsU66IlVn4z8AxhfgUbsWqnhg6IC LJves3wYGZ8T+OEYMUa7gjrH0PJ9SbA1rhRIluBq8xJJEhz6f0ftEBqTmu2mBxEduBP6jRLfYmaa YU/6nTpZ2tg6NyAgmJv0a+1Hdsyzv6U1Q04Fq4qoNgXv7QIpedCtGRuHw75BrYrDKl7gj0AUwkmM EKxVNdMxDOXoMqZqZrhwvlvGDv6F13qEIBrimbQqS+nt+je+QMbbZTY1UNdHTnkBRPHkgH+cRQKK V/MZBI1JN4B2gcgMjWBocnQPWcmW4rTrlJtouwdEBLtlQ5LfMj1HnqtEUZhuImOBoj1Zj3nGkcga 2uDBXyts/CQ09pvlbJuFYyTb4l4WugX/kejluMEeIiKdOhzUGKG0Tqxf7ZY2h81g6pP0hPH1pvAq Fd3RApcbzqABErCIo9n3tiQxexEpYYbgVGgNjqSDyHv0L0FqPIzdEfUZSAwOX1/NYWM/kk0pK2FJ niRBkaHxi9K0DJIOyuW2lk0TJ8pCt5wnGO0M0j9CM80EjTC/vpCTZa8vm99CzU4xr8NgGK542Q66 7ISVpJ05DpPm1oI0j3bbe2RcIEMSwsYtgPK/iV5ZNAZ6LA03alsIjaXx3aAcNJjXlClJmIy6K3hP M0DeXYfLjHJV34wSfCKcNzfkiJ4pUazm0bPzestNjUI9GQFAWak0lG9Ael0O3rbKhYAHecYP54qE xUrFlhPibcN/JZ1+RkcMTYaiLSuGWTo95mYebq5IwMHk4OjKKUSFr9m0aPJvjeqS9Mnl6LLTTdju 2KKI4C+3Z6fxwLx2naSjt7c3E22+LPUWyTNEw7sLJNWLhYvQ3xlOzeZJjo4gkiaCz2F/Z+AZPX0A nT7ztqr7jBxSp2lU/+YI42EYvMzjyvnzvIWUZ9m70TanXguGIsxSuzPiZIpDwW1BxEWMIzrLikkQ +/BKaNp+AYOXLFdhLCjULv8mPeCxmgQxlOLzxBHCv7I6yvGrH7ioi9BKMuD3NYsr/M2QKDlJX+wO mZpfAapwQE3hwTkGSi9dopc42orRQD9uCXXuFT5mgr4y7G4aKPHAqhzUya6YtaDLTcBGpqSpTHCL SgPBZT+N+cDOGzomlzcPN+cKPy9P1maIsvwgsSQrGHhzqU7s/oQ2lEBifXPrkU3FmW4XJ4N01LfJ 9dZlaekyxg2x+9f5K7MPZzIbdNYAlQflVTdF8dZP5kG798orTQzKdGx9TSiA/zM8xGwfPB2ptFkf Qdw9RNjc23KxhyMATa6dLxHR3crqxDIxI8pwZ6suGUJEiGkBgv1wh4M3R8zvSqEX31MuznPVaF0C 5B3IKow9hWhBjRLk3DfSTzqN2f2Ob4DX4xxZz9cpNs6nICHTjYLXo8uZcq9K4TAyXibEyZU5qaiB BFqr1r/VkGa/hkuWtJQbnpsElTtRoJjkqCpVH+dLutYKOaplCAkitSqW3jsRRgRn1xQPbrGY5/4R G/tUcn3SLm5AFbK81Sd4gE5rIOsc/pdmS03xvuLcFKokEz+3rPxtpH9qPOKy//enI/xqsdzu3gQS rJCcifAB2OB35R66D7EWFb0AOQkwb5YSTmrsBjfzJ92IpQXjgg1Q09fEkncUnOf3F24tNYaMPuFC 6qgF+1jvUh1IiZbVCiLASRpJa4M5RP3/2SGU9750tFxQMxy0WaIzkn1sVugnAI7IxXAOJ9GiCev1 W6JTqb2sm7zvaYMiftvzS/L/UJ27BDCkKUt0H//rIFzGDUl8QmZIfMOxc/QJggV14Pfl78lk5sQC DfZrVF+KVSw3kd8BtiBk0VshwlCYOh5g0cl+4Owkw3BX4BiC8c0GqEd1PtY8hGvhyEdTXgkOX4f0 7m8xl1iEUp2Esh4p0OacRIxuEQNerb12HR3i3wOjtvYipya7YOfzGZNhOebY+HLL0U3psQ6wUWc7 FK8oCIVANUe1lp7am5Xe810ctmOpTmZnOd3dDFr3aYjjT/1cV+HUIQi3CIn1VdRb3HA+IyEggLs8 +XKQ/tpRkzzwPs2sS85vUUlOGnovi5sDIEaP6mIq1ORmxNDkzqcWmMMu9M2MJm1zcmJb5r3yWJlt VU4FSmQxpQ+fkZvdtexBltAmFX2vzHAr0naFbIksIvOVQ8PWXlE8B7R8O9yxzjlDBGLMR/vb2uW2 LyxXvhuhZdGU1LFc96FMneFJ/YpmDr59luGKUYrtExcH35ZnKuh6vYIPT3LB1KOwOctwOwYt51fe ZjDQVaoHBLNdCtGNz4k6umSjezsu6aHsBiN5h27p5z5VPOmvZdpufqOMz14F+BFj8KKOS6jPIajM RZED75NeNidcW2mSrn/N0h6HQNWZVboC5q2lTEuqMmDylzneikxIdquhEoCz7HfPraJVZtAf4T5B JuLQgQuXMOxDwFtVedOtJo2K/e7x4YjFomzoR5MIT4hamWXhoGuNDhxGEozH6Oq9BVlsaa56p9fU vyY03zlH46uqR27RRWvRBGGZ/Dzh8VdN8DbMfPFkvAPb750BVEIXsby2rZr8AzWy/xD47cAL2ODX jSo1WLkTnJGp6tsG0ibtZG9LcfXkt0HcmSdc0nwHgI/jfLS7TdNh8SuB2fVPq9M1DK3bKVxBjHCx HBCVUtBa8InkoPjiUr9YHk6KQo8tJeoPZ5Ks2xV7VcoWuLUoFLMFYfwShdC3FH8uL1GtPiQc5+pz hj4IXkoDyo4OxTUgvkGdlX+QhawbBMJLzuYReFVnY5JmA03tUtPYUeCNf6eLSB4QM2JCaHBTt3lk 92jVxPrxe3GoXQn059JmUqI6VpkcCFLsP2SYh+it1wN2EAzZJijDsW4jGE0L3/mhAvqs4sf5ummP 1E4taF+RfHUZeNkw2vZoUQpTKLgkw+Qze3PrbEx2Q/Uw97+6mWqH2j3asi1gPauVrqMX7NyyVSfO EzR9cAHyxxh91IGrBmAkBSytbnRb+xHXH857+SuFtwfjgbsj4snbxKc2/ZLt4GKiR3JtWVxI4BBb wdXtjsxJnN3uIOqdZ/T1yky1712EVEWZQ5nL6CnueulI1pKcTW6CF6GzsD9ZelKHpuvZFJ+b4FiP v2T9XdohnZPHOJjbtDAzAP9kOLnKTAX0RtQ4icFASnohgeSO7+zD0eIATyYb2rlJB/PGtBPRyojE It7Z//tgZ/i5apww1lgfCkYyLbXK0xt/eQ4TV7Jo3q2P0CyOPJO1DtevxtJ5umNvPy7ku9bWs5T5 An8GOlhi9dmlm/JX9B5ZgBDf+AQ3GzJXKa8wAsZVJ4Msyt0McLs92tMZ8/LpvsjtHqZy9UWTtzCy ZplUq4b5693mdiC5euPKpaNil4lkHo2WLu0AqXti/a/ECDqSH6M9ISj318+YlFGBQRwJtCHGbvFN M3+sXC0g2hYTClbjnALxJmBy/G63ggi6FeAxOB1iePssAGBEo99ADzbdXfcQzF9eTcz5tJHaD88W ZtmKXLR1fAcnO0wWuhKfqK13MthmHmlNVD6rOyeKSC1jGBAXcJy++emvky9PePMs27r4uJ2/NJFX vI+ASF50pEIGnfGsQcmUang1VJBqQ2hP5ySLs8ZpfdcDvlvfleGCHs4YlYSfYNrevifyjeqlqmls G0SeF6z8DQplbmRzdHJlYW0NZW5kb2JqDTQ3IDAgb2JqPDwvQ3JvcEJveFswIDAgNjEyIDc5Ml0v UGFyZW50IDE0OCAwIFIvU3RydWN0UGFyZW50cyAyL0NvbnRlbnRzIDQ4IDAgUi9Sb3RhdGUgMC9N ZWRpYUJveFswIDAgNjEyIDc5Ml0vUmVzb3VyY2VzPDwvRm9udDw8L1RUMCAzNzYgMCBSL1RUMSAz NzQgMCBSL1RUMiAzNzUgMCBSL1RUMyAzNzMgMCBSPj4vUHJvY1NldFsvUERGL1RleHRdL1Byb3Bl cnRpZXM8PC9NQzAgMzg5IDAgUj4+L0V4dEdTdGF0ZTw8L0dTMCAzODEgMCBSPj4+Pi9UeXBlL1Bh Z2U+Pg1lbmRvYmoNNDggMCBvYmo8PC9MZW5ndGggMTM0OTgvRmlsdGVyL0ZsYXRlRGVjb2RlPj5z dHJlYW0NCkEGrOe9QFx8klX3OpkOuMe4R+CHuizsjVnZKvXr/xf+D/Rq3D6e6NfUAyZL4x3Kt4t7 wwC6IVWTQvMOIWfq0IgGimNivmWcqIIHXiG96+MQFD/3pvN9ztb4m4sZRGGbFrFmJSvxmZMsbVe4 gs4ybSvb9NCbYDJ5Ivlq/6j696/DIrHNHHM+pOHeWYx5uHiZ+A8nizibhXdnNJjI6h1ddQqZhoTb lQ/mceT7F5tGuoOmVVINGN8uHuSrtkt98aLQtTF8mCULU6Er7mMU55c2nND7qF4IP5WY58MoYCC1 7JSA00a2XPhFu8xFMcIuDCNQvcPlQ6ftP77oOBSclO4Gwsa/pKx1svqU9g9To1rs2u+ZC22m7Snu 57oSGHkYEt1+DdR1wYZap6f2F1Uq9SMQIu2PbOH8ecy/SZe6asXEF9d75PpSF0D7KhDuh3joWF47 WgG0aAFkKYDeI0RyjKS//0AakibPw7ZMdHXXbhQCsTN1JF6iYnvAs4FMhaxFDy6BoTpYE5S0kzTr kQAf2qFEtHd3Va3c//XMngKqGN8Vt15ZCC6OvsnuVFDjB63dQVjT3ShBGutX+v5Q2hGjLBynxHep DzNusk+G0OPeqa6IG87tLvuYCkOpXPA1un+NYZjrdxCatwi5F04ATjwpszKUx3CYxl1j555Krajp 8+JjSM4469H39MyqU7VbeUx9ToWBiqXi2IT9SZjoTd8jMXhvo4ZdibmLTdgOzoBs9ellMfOcS6nJ 5XSQgpEKiVjoIH+6WTQrWfRFNw300touIY84YQeFAAXVJYOrolNqqNmnxYhFD17PtrmnThKcgXeZ K1nXfk1Cy6p0zrij5oY2k5frGMVp63awB6jNmBAWqvlVeHn7eVT2O/7wPEefBYa6gLVJOnuYQWLJ EVNCtidycrgubDtPHjVKsyeY1LjIFnuECuJpzyci7TI7sz+srsQ4c3Xk4brBvhk4aqhR6CNfRcDp LfPNC+aRpnQXOdJZjUNIxIIirMbQa4lNe8cuhuC7VH1baCftmblLkZJoq8HU7rS0JZV+TwXy6n6a rh3E+hVhhmXt/ILpDBEf39t3gt0qEUtBhaOmB6hLBOMHsJ682veFj28KpLYoXS3h/7YKNDlQs+N7 Xdd9vINwGM0BhEzyFhW0kneYCZDI6dueqGN3CP4U23c0q2wBCXJl09sJCQtwpNzesfnseRHW9MhL meOvzNo1hCwLri/x3WU0lFJQ1LLNBGMCb3+mgx5VGHb0GaDr4Luw/QhcLcmec44nqNoaphJXodft tIeSdP55oquLYIabNxmG7fxzjqQ98eBk+6lN7Qixn5qZE9mjp28UBHEnJ7iGAMK8OWZyiCv8S9s3 CS/rlGBKLaWK/or0hE+r4bKW88THpbovaG0MHr65gk1jVrG5/t9X/3NO38Cw7L1h05N6ds1USHjv lH7ErXqqzErCM4jPp9E653znLVbO335EFoKXLwB/u9C9lgE9SANNhz3AZ1ft9FVQKcYwNyryzDOT aaJQgCDrudP0lUHjnm5YAx/E+wBGAMnM6k4NiRq9trFT/2y4H0BihLKPLIvrCO+ROQhy0vw7M+k6 lTkUip6UjelRoyLL/ynthdsIqRRDgzAENJkRjt7u50CSoqb0DUob8E36NDx0WSZNyZwmXh6GVCWk K0qd3x6tdPmAgdbRS4fNQ/+4xEJNQhH0v0/3TwypeeXD3WKPw/Pya/tsp6gvBN8V6CcBneTWDtfM ZY2zkp5ulrZ3m7BkaBepzImN3L0CCpmHZ+9ym+PrFYM2C8DiSDxejuuOWbXwV0mZlqNgc6CcAFuS vDPeHuA62vjp4zgGjcRVoeyb6l2p5HePAy+UM3/Jmp0iERKpAVUzaAM/oclVM5UFmIsxvkEvrXsP IeNgUSOGUrGyQ5vluXdpDK5cknHqtk/os9bb6yEyqkq18i6fWRpab1VYNRlqHXk6GZsQBx8M/j0P DYfnDGwRasonGTbxwatG7+6UAo54HsAEv684MxuhxXjnWVMJR24mH9RKhNL9XByqG88s23ya5Tvj GGG1btjqzGRv/vjeudQBc7Z/msBs98bGtA7NBwtIrvBpoXRM7dggAxMve5zosSSQhv0q4YIiYCSm GAhu8lE+5I3Yo9BbsQeLkgPKz639zWaXoO35wxBSTlc1clhMdJj67oinFoPpWPbD5ni+bZD/iAzm fS6BJv6xEI6PP/j3cOo2Bl1/nsDG7kRcTWU+zPSwNG8FEwgzImrm/Pksikjby0lYhXF8CB98EOfj 8mP0233VP/xXUl8PoyStC9MleyIQm9GxPwAq2TssSNE6winkNTaTXx7kDLKSRJ2cfs4EiT8N49yH /+eSWBcLCZHrzRozVb4b3uJ/qbrVWn7GlM3bCnFtqTg/9hg8TsmyGhnCsVDSKl41bcG1C40UZ1be yf5/Shew2NqpX0Em+LCXEuvD+BUWjzh97ib0N3QfX4QjN4c44trO19tXgc2qI7/MAQxxkdfAEzIG 9dau/UFIsmvTzLD+CBIZqUpPVrfMf9gYR76fFFzquOxAoCua0Hi9bX8AUfA6gH1eLAWIvyETCqgN Pf98ZLzdX8BljynaS4fvDtbIVnNXqjwiV5w+SMRfsFsvRJDBjKYJvitMtyt1NWdIurjNnZ5h/Pex 2MnBibYGWKu4xuLdNm4EYdUSbDmbb4wKi3BH3L4kWxneRvJnsibIg52xVET61bGHPNCgPW8B9VQI vad2RJ9ygtlNzM5neKJN5QiH/2iiIXwQl0vvMwbns+kK10lnDs9Yj7eGPJJfY38OI9LGKKdPKaMJ kFctvrmx6Z2N9Juzp3zlumorL3R3BFBGSFRjCfMTQbvOWEU1qe2GcVLC2agD8s7y1cFLrooEYZfz iEam22ugI7DOpvtOYPCxkRUwcr+Cn6RvrqCdBVZezffVrHHOBY+tuoEvSrGTWkZoD67y+NLNIK/4 1lO3v9RFzzr47uA7twMV9BIcvFiJL1nNmVmiZUBstMhBgXMuC4sGBZHVise/zdVG9vYJ+ChkAxhk xndroShGFJ3HrbshtkxOQMWXC1zmwjUF4tYHNSL2cChsMjjvC4rdAlKtDm5lLaScGJEo5imGT4Md FEIfyemYcIIEdM/No/cMHjtDeErL6if6md9bIjMpwiXjAKOwcAnHKeScpRboDDomY96g3lH7nNNc 7uskmfsJx1tR8BEOu7Tg1LIVVJTI/5DGkllOMcbqzNwD8uV6+kMUFzOkA3PR8Nm9AKJ6ePPty7BD csa625wcGo/WsJmumZjMLNfW0k/Jt0PyzObU35T832DlwO3AJWZmv5ZSkke5NF/RomHo1o+fFD4/ 6pnnAFTueM+LIOFLWOPc/JcVXYG+7fpN90F25CfHeZr/I3v7DJVASTLbW7WATQSnfJqzuLgZ1kjW FzP+dUuQWS4M1YjmUjMYViBTFYwLXwlUSV4if0jzebjGXpme4wSRMCbnUjJkFDYgs2/ZIpqRnSZE LLo4OdskNUZGlzTrezZkhdF4JKd2zRRgBzkmyeIaddQWqz8shCAdTeeWUpuNtk2RhmsNyN3nIGAh icVd7K2+fnBe43FKi0sndKfodCEUF3UwEEpXuEOB/XHYHsYPw6/rdMG16kWppAcCg+sM2Hka2p1Z w9VcYK0zq+/e+tI9AQqKFGtUSD+UFCXLp4FCKIN3ySMTrhYqhF+tmKhKgKbkBLLD+XkwOIzuFmCr RXhKpniGmUuViRMscu0LxNIvrd+dnzFZVDnxbSz69/Q/TNEF01WYUD885W5StJraoo+5tM6VnTNj w0mD75izvhNEtst2NQZ3VEf5yr6HCcA+DY+Bzxg679/9ZQOVBggEVtPeT9TiMcqmQvMFicHQ38g1 XBytwqNA/tZAgFViDQanRHQff3UPodHhrnLfEcTLA85eUvdvpvSR6iIEgbTTSNq3cNJCmo8W/ddf 2vZVrgjD+h/bNG2YI66Q7WlU8cQVn5uRno9yW0EHhqDbNFuddGXnxO4H4VvZF0tyUnycn4oe9WBW rWtY5WpSpmy7ojrPQ1VnEEOuGY37Nnlna7pM9cD9kwlYV5SBY6sBxhRnJQEvXFvZhf+Tf4e9qoBv 1dj+u2osccGNZHDJy8zmRwDoiF6RWM1qKCBySRkgK7XvKyW5nN8iUeHvPCiu5jn9s8EOPybpNYT/ +p9OFGvXwQH0tfsmUMNNgORQK15YKl1GMQzYovm0YrauR8aMcVwLTpc4ZMpT7jR8ZKDIgmH+fwIi Qs0/L8866+hDkBCAs/FWcVofCTIYOJmUs3sYH1yQuFYsqO95WfEA1sgsTw9RSMr7Tqxy07QQcTxj ytzhI/0+fAapIZN8aK53AkHG5ZaIfp4pQGhIcj7tMPPWWQJGmRT5WzRGbGdfyM+fFMGhcO44ctro 7uhGdVG3TAfvYKfjFpl4mA38eJFRou4218ivqasPZp8CvXRcZLzAMFsUifaNAngrZfuNQ6f2Q4ZW o8ztP1uCKPBE4mmwgjaP4QKWI8GYb+WBTgxuxEYy3ZHUnWVFxlqfP7i9ZiFG+LgbPmBS2mie7QVu kNdcKRxjpS35YCOReQCMGuGryH3sXOWfRm4gOsJtkdhdLQo/bHAQn5wfZ+cl4922/KWfOj/W/I0J 28KKVCw/SZYXKwrCds047/sDHuxSbFTqJTvvS8Fmn2GHG8ngI6y6jnTVs760HSCck0jCphWhJIkj TU3qotT3YwgMNJdk+Udw65dx0r6F4CTs1CduoA8q9+Eg22SvTARDNsBM3MaSrxnd3k8XKpItK4Ti Ao3jz6vAsrz6e3YG3uA2pqLnUkwsO+aLIyjG7Bt4F3nZ4uyCOqJkG4vNEfV0kLZq6v4ux2v0uMr/ EA5swKePI9megm0laSH60GMA54AvzuLvtXXT95BfcTXb5iJhL1Hc1Ve1OtAOm+5aRX5DQ/QL2iyr VqIfg3Pq5wetFf1hAodF8ksXHXcc+MfZDxSzas18uKNSu0xaMKczAzW/9tIG9ueFXVgM8xGMShiE soEC7E77DF3rU836DtuQd6KBc6hBucQ4uGl4PZ/N5idowzFoMMrgajb0cb/X7dY0N0KUpNdNYTwD z5aCY9yHvvsQ35usXfOjYZi7Iq7piXBM4lPlQuqTb9G6IXBgygpDgin3Lo3rQF7qL7s86B7JYF6s i5ZzMungp558WX5COA012XGj7hMFXlkqCh/CCDE5SjhCoutuJxrdLNWZ8A1y5KvItoTywCU5MlJX PSnBCOS6zhIEO5m+n/eY8Hdrbo1FwKcleJWJMVrncwdGYoUtGCy9Z7VlQtENIK46qd+D2z6AerIE 6fMZFEKCiPsKInAUmSy+eTXfDEgi2LVfmnFad8qlEcUPSznMPQ+5qNSun+RLwaU/FYHMHT541Il4 OODzTA7IQ7h+wKitSgAVyrglGMH1CfFvEOpCwhpqHXix8/U18uqHINd2SVZaKhV2+kyQpwoYfH4P tCUBTmPtV9oHrxGp9aUdO7A1HgfeZaribmhlzmGM8auMuteaYOa8QgY5rdg83+2wGCvFYG9Q/HXW ZotsnZXoNkaGTVzlzCQV41p9sT2b1GOrAIgmpANcgU3mfZVcSW9bGrMEjDPEXg+raeb/yPMTKiX5 6cVJ1DfCU1z3LdagK3fNLWxoHaMYdXc81odgCoGv1sYPyWtQ2vWtefANCNwtg+MYCsywMsDy3JvO UEqyg6CAyfUXyKH8wfNqjvCHAa4aslAd3UMIBnr3LpXEi9xsQTe7xQj3eaPRSFkbuVZAX1IAlOme 76SQisOHRRJsEAB7vyj1UN1xtGmflJ+Xi9ErEm1FfcZ/uZAF+7Tfm4mXvrLxS6F4k8pEG9WeZQFc d+N6U8bIepebjG2gA+CEruaZhrvhk/iboV6cxkFBjJgMluP1PIayhu9YPprmMJDD6KkOnkcaHQum NeeVam6y/1EBQD8SgoWDWKa0DPBphVfz8zfJNv3zcHH+H4MkybTPq5uHzYACBTUQWkOTeTFiDhxp rp7hfzhjlgshRYLLLIqm7aMG1SAAczs7mTa4B/2I6ON7na2NjUQ5c0zcnqt1BFe8X5S3Ct0dfi3C THxmfs+mMepljUoKMvRRAEUy+sHXqR0MkJ/UPMjiqJoOnb0zHWvlHzPllnLxuJovAkXAOo6uvVEu 9mF4Ek2EAlef3o/ayJU75jV8b0DrvLBT7sI7+x0NbJ/AJLuk5ExMH3RX4QMuVdVIABGMgtm3SykB JsCv5lA4imN++lFI3Ycx2ku4oG5CIokyvGdKYZfYihu60EweMXG8d7MtopYyNXZoFA2m5Fa13wWc 5lUg/evjl9MbRd0Ycz99OWbZkybSvYzPNOnUZn8TKHAFA7uERP5Q2uf+Uz5hKOv06ihVIK1XsETg uXSn0d7mY6Y9C5pwzrAPND8HzSUf3djyR4sCc2Qunhl0oGgBClSP9jQviXd/xBfg8yHg7RV0odDI qEJTZtbCiOf5UDOIPYhF2k9qVRtLtc7D3kpipkcNG31bICd9e2u0Nvvf/+u2C7uHXHNbZmXZMdwg GVUT7Ls1P8ELGX7q7mM0qwys3ewUV5WIPpYci17tUwspW+j7SDU6HPQd0OgW0xzEaIc98oZYBxPb yHyKi24xDkLY3k5QhfypDrN86pfeuc/tdlGmEPvij7kYB3eS56R6qVuz7hTeavMJrxu5iOUak510 RxynkkGJosynG21K9yjtAbF5nwRm5s8DkdUW98rmppj2y9efpV/ZYn7eLVXaztLPzfEbeyql/6Nd 3DcGJRfvueLjmR54nMVcz9nm1ESjevKv6sSxigOn1QfkzVJBMEWveAmpDfcz4r7IAghkVGMRQQvB rfwurBe99CWuUkHoQY8PmY6nj/lq9C75cIrMCjLe3FVipdGtK7XfOZn4hdslv6fXNQtNN4KQz9Ge jzi8f4Hm3Hb7BxDVbbUIKViK2wSU/8qPfYkp2NtrZZeEFo0vGy6FnrrqmwrLvkiKbxYy/MDHY0Qk FGkUhVD7YHRxcSDBogeR/Rxc2epSmQdic51TdnvFMWgx7v89xrQcYOEFG8DfSsJR+8HL5tL2Ulkc mZ5vdHYvcX3WC9iuFzfLVYtmYKRHY29Atfn3EsAT4USepG6gBdC0nq949AZ2Y6PwLKSVWBeFhUSo kt13DmaGT2DjjFUck5uLN3MNCc0los37+KEgw8iM+AIN/fJh+AsFsh0ljy4BNVfm120toWgjL45j SUz/6Kucg0MmkTojcorGZyigkhskW+bLQbveGCORganFNlCzWVX1WGguOZWbbi5iC4hSkYI4/KZR mI6+y8QKd1rrmkw45lK/C9X19sHKRWMeuVktXvfdAL8yeCwFdqPLB1MxrRTfSay1mAPN3Cb6GB88 mdvYBmomhD+SJdElS0pIBBm4uqi+bArjEvx5CpufVuqBB3VeGc+p6S2GSskniFVD+meZJAv5Kx/U 4v0cdEKNgyatWLw5ucWmdunYcITX/NVoZry7CO8Ysz5s6oLfHlDk2X08csPnorGsuJV3vQafb2Xz 0TKwezXkZVUItWZAzLWXn43AQQpY5DQ32NpHvNCLysP+smWvxXtE5l/klr4SmVjpEPhL8f6SlAur 4oOq1b31gDfSQA5bXzbx6uRefj/INO2Sy3iz7yo004ttkQtGfZ9dbwiWoMt5EYrf8kM1oSaQ1tKy NVECAuXmQT2QmJsUoMyaF6FWCi75B5/bOy8BjoaymlXdMSOY8YqNn5anuUO28MW13gPjHomdGjIi 2nAHnTzPNY/ghDoJS1R0LIZWU4ufOKcR0KKhvDzR4C+5HeP+grmBRU420n/27kYXcVCigHNwIofA HREhHvcVlp3hcjtp27sFp3Rp/dNrCRZ0mmaxb3JpzPIYuRHXLpq+CybyrMd/VPVo2Zx6BUnvjMTs fZNVsKt5a2DcdLMQQqA5pVupyzN5Xn1NVzagwax78E2gDHUMbAkUWkPpI8srr76NDW9orsdPuzw4 DcP96I2Hkk0SCx/dDiIGcUtAs6cs5gB+2Yh3DwgyFqz2aoINgujSyw8PLLeZ9ToC7HU2b7jJnptn T2sOjy16Mz3HkceGaMj64EwjRC/RyZg4mfhgG/Wltnk2LszXMRDGOVrrH7ceAV7sGThtQuHnWtSV Gug06i6cwWYgsEz/D8kmTCydH0bYrcCSdEmkWlDFcHzF3eFeWn0SRunS1BYO51mTSB0dRHbRoqmg 8LG1+wdE+TYFGoo/bZCl39b4NfUv1gNNZCDiwaYHErhupfG828BSopUo6rRpurbjnaWpVQcF+QXa xPI6fTadvWe2M5IvCh2J9ZF9eaM0bnz1pDBZVXImBbxm1XxlGIAvCOC+Ps68brlEoiIGJlMGdjtW HZyRzEEzEIyfwKoQzrnuDe7AkV9qRhM36OYJkLu9BCcogNKoSA9mfevZTLFTYmrd7lY90AXFVG4W 4w8I5ei9sNn7nVdo8OyQNHg4O8fZeYxO4I0ILTAu/o36XttusZi2ZJDC7kq9AO9OPMBozmbb4y5X ZtJHSV7JPmgYH1M3SHZIOArIcxYds6YSFKkbmdZiXbmOvIrsCLNzMRbeB/TACm6uk8urwR6Zhorr ZpTTmELNH0ofLU3JgotC6kVKxccZZAsYBYChhTBHrZ39rCyMGw92LMeOvkvXps63LlEIuKcNnhZp f7H4efPV1TBdenkToZ0I/Z3XLo/JJuysq2gNfPGs1p5FZdbkzVYOxRQNn7IrrBAf28IDeIN/7PSD gOUYt+QH2eQV314506p26x7lrjSNmLEeALL3nArY3cY2NSExVxNFkb5WPfsvCMmtQRUeqyKyEeS1 PlaDr2aid2k7tdeMoShhn4Fr86ynLM5tN5eWSGFsCHdjN5qdgxAYcPfdoB32GihivDDJp1dcyKBv HBS8JcknEiHkd5a6x2Nj8bjKf7vKAubY6GzUaC5rYoiWbNLyLRoMWDIlnb9ngs6lhOOPrnzmIsPq dIC+JlkNK/d9DWsSjjrYoJKuPZa4YBJkdbfGcx65V2QAiUPjz0RHblogzAsAwjkKD4qtVpouXthu yW0J+E6XbZng25KrOZhkroKCyk/DTQkKRlqT2YpsxqLLCDwhoAqf9LwEQRpdUo9zTN01+L2onesg kMMa1b4BqRYrQYnbZFow/Ec07srYf2NVunFOc99R8WLaDRsLAJDYQ5/3O37wIOJLVB9l7zXIIsod F79XXJuofUEzQf+OgDZEn104i/h6+HLzj0fyLf9xoyGP4dZq14+4rBVQCj/DJWTl7uBDXtWYFa3E 7vYtFqoJY4mD6hb8+I3l4ja7gVTSKzyDN7X6UBLDtZThleZsZ4FaiUaGmsqLIEeAFLFlsAIyhUUM 9/gH8S9C7Ar0d2e7M7qH9z8ujyZ56y8+LMgamChEIfrX/FLWJO5CGbTrZxUslENWaGijHvZqyixs uHOESnwCTknOLWifMSXGdCfQ663M/m162f4JzrS6GZpztRfu9tSvhmAwRCXEXWXW6mPaQxxVOt/C o5caTTwzWshY7J1Am22b7XKbdmtNAh6GsrT5ntqdM46/Izj+RYBIhwxuml93CDTKMnH3fblP0B9p ONPZt/yEZ2NaaWRYYFz9b+4iAMXj8q7SR4UFCXxOPlj/9PitBIHHxfQeceCKS62/+JaQNEi3fPIp kjn6rSgDP9kePcnXSp4vSgE3KPo9woyhvAX2B/ArAdtC40Byg4aa4rrKWgmPa1hzcV8+Oxjbg2/r 7dcWUoDkgQxzrO0wP2XUKS05KaIJ/eQrgyiklMirjPXLSZWUZDZENK4LnM9jYdRE7cQiXO/dgdGb otPJw3unjcHm5ZmAzGSrMzFqqlHdJTyOAcQfaJ9abNgO+FTpWLHUEMuFUtHilaYYEoXwkRvU7dLv JCTVuDHxUXZ3lLZnFnZThXtwYFcONupXeUXLRh3MuCskk4m3MIfxFR5/WSTLJcrdvOsVQJmjTO45 +d3VJ91THzySV8UG1k2y938dna6iAgnYEhLAElSUtzcMJIomuKvW2fzWvOw7h/KLK5TgAnlKPd1h ZrlNI+y4KTwvClgnGaudy/+j/WO+zTX1LMuwsuIlZ5h4kTpz95PCc1mdUJ1LU6Shqz4NALxAl7EJ 4RqWxJC/ZvX96slCSNV/Y085tvPzgB4a2ZX9H1AbIq1cU5lk/QQIuduQNLNz4+Vh8QnTOYsbZpHO ucQpT3tsVepFgQv/qQw8J4lqvBPyAFPKAVxOjskPoYSAcmDQykgpaZXwEXY9/uuSaIEqsY3XsBvS +eypXtVwGlvoKbV0HQDgWQiV4JG86C8CV0Vv4ZEHCh/NaxRXIoFXCKb8CeI3x77jG90UV82CAIot C24SYcizuE+VB/z09ueDGptMxwLm7h/gSW6fYFhcawWzaK9MSIAWjsrw5Zuqkm8RAPQvGqRsFA93 yG74XaHF2BIjww3aD49QT9UiKKzS5F+I02n9TUCnfbfd5X93BQ/7B7mBbjkKSuU06JBwbqSzcFLr EDUKOSKF9I1DDfkaKqLBIOKPjjUNhA9AQWnGNxtUNINILq3S8vzVp7YmvFGHkeGufNcwwyWoYNg0 fp7rbXH4IJYC3eYVWjhjZsX8jWFXE6WJUE2hy3th4AV6fisFPE7v5z3ljtypWGPT56zpec+vQVhc Mv8eH7svHweNfBVOUMEJ+SMlDCgAJssrYliFnJmYWlEaIqPcQwAEQGFZOwpUWn7/4zwyvQHeu6zr WmzRov34eDnmy0Ydv3m1hPBUBweHXXVGg9VVXRGx4ymTYUDOis4dTeHZdNEOoMaWk7ydnGtqZ+Vv ZAYWzsr1emGjNiAjKGeLeXqR+Cdq9pAJGnlqiQJD8ilL5EhS/ri3KTXl+dPCaU+EtZ74k0bmBpeG 1zYOhOQk2XxUmNA2CTzWUGId18acG3xNxU+o7ydtCwTMRixwW4PvjWi+afXtlFtk+RaKqXOTqGVj Nc4xKR1O7dzd9KZ+26dbeGPbuKjuaJK6fqM+hMybpfHIIP/z3faPnly5oITtaMUPTXBZz/9tuSS1 H3GGZab4+cPqolH/PLhx+I5u1sXYqASd1f89lLQaJxq4M3XXFJdASzBrKNmtmyA/Kle1aC8f0VCW Mu+H4EScDxCdKvmbh686Sbwic8+mxQhcluv8abLcW032lDOWLRGGgo8YW+jmfUat/ke0StWt4bOl zhV52mOqim1I9u5NGZeGGufs6ZvlPGE60VBE86x1cyS/GD9CuSIYW/SZnJP70VTi7bmJoMfTf4xj PJ0hc/6aW2UHAIPs5TBaacALdlMSch86rqY0oyLXymQMlF0cnp/Zj+aJNiwe8ARCmdrEJ2uBi5DT qQzRtHOB0mcTLJ6x/leY4R5X+6ZMB32hOvnS4lfSehA0XsAuSbAoyAWw1gQffbe7vDCwjlLE/Xxo N26BA3zEOjoPQmnGWqzPlv4tKNYEKGoHioH7TOjSlsypqwUc3F7sZyGuWSwxdIxo5vP2p3MRFItn B93CqlTX5xwjmcG4kPOpQO3N7zig9yMGH+4riDgGYpWEmEfHCXz1W+SkO5u5UuUazEKmeXqGlqaN OfJregp9MJpYpw5mJeu7NO24gQ29PnmAHEEu+eoe6rPlIJa+MMUxhOPxaMkT0GoHCSzcIeHv73dj GhTsDgFo074VZMm1/QE6uAXnxvWBg4d4bn15YubXyx+eMLrLycCodX7cJhPgUNaqJsx6IXr0ReFt pWNEzViJWGLxI3aA0YYlkO+/jtS0ELtE0hZ6MexBbYThSAG773voFMOm9Iuho72zUCWAA5QidHwO Vho+UL/wgk/3QY+vThGqyV3aNjIp8pJHcmnsh8cTE62ywnhDu/8VbnLvd3wXK59KqN7f0+K+6JuZ D15fPAIMezx+XmoFB18VlA5iJtXMMDfTLyyv386QL/sTb+UF8/SFtr4Xb7K5de/ul8Mkw4UY3COD PvPBR7dvmY1Gb6QLJjxyGWPiSVy8RB7w5Iucj1hBQzR+PnzR30qawQwKTc/6/K68SXVJKn06p5jQ 9cuoSj9TM5ccXbyPDHHcUbg1ilRWxvMP8n3otnXD5QcHVJMEitj1AxVr15va4BxEG53UM4cWbQD3 YJK2saM7vGW07rSTh1PKV7wp3thHMYGdLDUMTWreKv2CsGfVHBhTW12icUltsVwo/cnxSKK515t5 Hvgi76ByP2C6aqDLgjPFvo93guLd5ShRDKMEhUKyUyuom6uYxtR3TZduwaRClNG3rRXKZHzTU5IQ K+tY5sXYrP56QPvjXPDcfKNPBs8QYSoB+eECFOxx5JRa1i69R7dVT+SFKDUFmhyQcY780daD7zV/ lQf+YxnTcgxtIMH9skKzGmVWomrD1dNT64jIhT0atzdm4ryVBIkVQ3NptMCl1BNErqcZGmZzhwvZ G5MtIviNVDxcpFmQF8MUCYBZnP1TfK+JFErw5WkFAfTZPyMmjNCH1H2C7fuK/3ijw24adL9bkiKP vFSJcClIwO6XvaBPjU7xxs16Tqg4OqfDe66VyszgsSQgG+5FQGwFcXFGnKnUJHRrYR1RCzoXJG3U PXM6bJ38y4hJ/JGzKhLBPYeStby1PX3IP+dlXx6O+D6ehQiybGVYuwqgumzTx31zEDuFzJPA68Bj q1Ljofn8xWreY20coJoKSej+lcCcV5hdVm+WcE97rJSrL+P6ZIlG83NtWByIC7evK3mA2EjP7ePy bLE6d0DeIhviGrBgTnX95F0guU3NMetpj+sRTBQqmO9flYX7WevWy3r9dASghBhFhYuNqcdpGrUa TWVoV8qA/cZJXWPpcdEcr2RsBk1EFFkdZt8ys/bVOE4LDQI3lUdkB+SsbA/nYxjpa4wHgM4OJgTI 9/NnBr0tl/9qjfTi6D4Qrp6IUExpMChEizI3tHJJrWvG/lZ8R7giBC53KuLK05AYmGr36dVeo3vH 7ynbuFsvVO5JELB2AKWMdizJ2OKd9muo4cV3lC4gA5bcN8WUttHVveEZkq1zsmpVJJlgdoNFOGew TxOzkaruqamjEvxwuCrZzAakkZY3oWCsPlqbNzAP2aTrjzvN53WMjZFLUMkMIup5urPXM7b4ckLS a5WfbJO80+8BTwF02I+6JFt8F98HZS9CQvOF9Xdk7TSIOeVbELacnNSM0St9UDNUeyIwN6EVdFg9 ajcyfmS64c3YQZLFkqZEi5Tx0lO+NqA2Gy1P4aFEj1u5VNTnnxpH15KE+qo12F4B/AhrShOuBER6 N5DX1f/gHwG2uAqsxuFtCnkrPm//R3/+F5X5iCDdXojwC15pPjFB8FwGqMdmj9kcMNqa1UScrna8 fDH23R57d05mNiROY7orndZHgHj8+30iZjB5a6+lmQJChHmAiEnpc0SU3HQvVvMEFdFAKCdccsXg i96U7JMI2dyROjflUtoKffwU5BAYj45ObFacrHdOjkh6RrleKpjlJ7yBDv5e0wHN8EIboCPSmWTk pnWKHmoWARuJS1lEstBQZS8RMDfwxDfkF8ZgPAO5kvqRJf4b1K/iqUU3yIi2Z6+ERIpQTRz0S81j OiriMwWXf1rJbrDwoH2knzttN6IHTE8TaD7eQm5HllTVMBwwUHF1D+02AgPIG9fz8HdespmKyJBQ n2rN0I9cd150Of6cEvYoYHEd/VQ8KtyWmDTSZgFNSuE96Lw1JqMybPzLQGaDILFue/GJQ1AUA/hv 09b37+3/Iu/nysEgYssNxmr/+zvriemln+hs037W07MJdK9Qj0f2HCsC9A3NkGaSM+ivDHyMkfnj MQ+Q3vTAolKi1VaglsP2izz57hqCJ+5WIEL5Z4c0lL738yhBSXgr7tU61zSgTH0WmauIfyZ6ysGW fjW6E2m94kBhUq0DkanaprFE3Xj0pVkA37j/s50CbsNl4jc1NSigR0YS2/ZLRmNDro1+NToA1ipF RXafmiydSZOPrvnv3aBROf3/5JSxsiPJm2etC30tYb1mjhl87auO6akjkHjsBfPYZoF7gxTu3kq+ iNu3Tif1iWRiZdItDCl6nkEhQo0frUAHdmoSr04DmUGCZJ0x2uo8XtdEjacvPlRGLYHxxIUW9i5/ F3F9ZsTeCjpoj5gitip0Tkn4Vp2pgPdbdGqGxEnxrGnssxqeBysNMETv833UFYmpiSoEHcYjWkEL pBWFjPnQRJuN0M4Gb/eRb1o500llCRQAuXdLxVfcso69LsYDfLuzdIAjx2xfghNpaz0K10O6zcRX ulF2vI+fRxA/hqBI00mIuSxhwNQZBerO9S6B7W+n2IUvI4hUDCDyLgl3/ZpZaeHtQn4BJBsn14l/ oVhSi6G5CGh2bqB65dJHBQXGfGBsvM0hbHE6Ul/11ixmDs/Hx75ia+/QGHRvVfspR/4mCNRIHLNQ vLMAQ5vBkItH7Ecx6g8Yu7I8mcn7K7QXO2kuAZUYuA/4z8xT7L5/eTrKnA7eLl3otmpupMxw2CnU dxnu+7bDaD5l19mvA/kbOei+6XKWtfEYc/ICbZGTtYjBNaCdv2bakQMWQINEl1ceSZkx7csYe+X6 gVecJp4dqe0w2lq6sJbMxar6Kse2PlYFNTFf/JSrv4RDZhfXVRXOBjA0cdVqhh7TRYXyl4B7aqdT /P0/CXueOKW9UrdFU4fP/t7hpsZxqkgHiB8HELlBjbZks7v1R0irn1Ba7tmrw7RXnH2ztzfElTot YjEWzBDvMt6JwIe9Qld0TPpY7c+JHCtlRlID8lYIM5j6TD9+Ga+WZBYE+zQBMuoGPjpBkEnbPzbd 1liGXT2Hg0ZTtoC1atlAMscHazkrhLuX94JxXSljVLfVLahe/oJY75Qn+NsOj6Ej4vEFEbkTdL2o Ki3BARl3g/n4EK3PyEVsPnBRnuZ39sPtrDkPHRv/JSQ8QluE8EU36IGscsNjT62g23akldpcZ/Bw AQt880e1vFJWURSYeHkcDJl3OTJMryKGRbomAmpSqdfMrFoxjyloJ2MVcnTc21RkUT4sjbKJikpi W3n556w5p7c+RXuVJVF13bhf4cFYWedqPOsvslReAzuX/notDx1p48mGR4y0JiuR4IRhsM5wPLg+ fnu3r1bqYUJhSDqVeQady4LibLtLUik9nnYHsV9OaeRzpkqV6ONAJ6MwEC22rfsgd+46MwyGKPhJ BQInSQ/r3L70kjoIVYJmLQ6mm9yosau+WM41I2LKv8Gfl1YWX4HcXCT6EF+uD+L5S0eYmQq4rgMC h3tjMH3MKiX0M99qrLQIAtzg3eXEB04GBaRPWk6ymBONbMxI/WnzxTm466GCpW4TQ6B+u+DnYeMz xDzJ0477ykrlUQfO58dPbVIMhpj8EYE1fzy6aAOk5yt335iS3iXlV3lsfjKdoVls4YHkXUWNe3Dm Mm4/Ac1g1ie3wJ/B+8CkVPCuYKml5XBdqH0+SAxh8jXqE7d/8Ct0OIgNatJvyqU8hKmGUtgF7K5D FTysepC9OlF68nYlGCnFgZlwfXYCFb5u1Wffg4fy9BxHnrCBkW8oRfP0BG8LR720ZXYl+jcpAbj0 iIjYlE7/RD9zycWIeYbDvdG9SvYkSF2YpDKfv0bkvTQ16QcafZ+u++TsFTumvSKzDT/U1gvSra9y 7Ze+KDBgRZY7y7atsIqeHmI1yLx03tVMa1e1/ru9KCY/Amqj6IixBziOJKfgzqtBrF0CU0oGD7kD kyVw4xE5ec4qsfOKs6dWSfoQTRDW3PRh4BgLW2VDGiTE8/bnjzggcZXGa6KhI135gRBwwCCx+U5d OMNm+aR/5QmirV2lu1xzn30XC5GSBJaX9/7RNM73Rzqx7q/xgz6UpQj6DySybcqe+LhiP2v30Ux1 kMQT1JJQfpJ8/X0zNW18slPoj15x39aeNwge2oSsGj5IZAGt87TnVQeJUWtUYEGhqULJqMnOVxTS d8dmasyPMhdShWnOBF/Ut0tcV6Cr6Q+LegiFweKOhpHsiYf8+VzIS/pvLPyxEa3udHQNRk0fgzXi MM/LCHZo5IUd8E2FNoxEvIiE/bXTtDHyDtpcy9nb5HxUxv9LSj2i71lLiscs+vNmoKc4upqkvHs+ rubEDl3dJyGCYdze2hqy+ObVzW9HbIARfCiVNwc1skR1zw739Ag3la6YpXaut2UtBzIqdl9Oj8Ry rsDBWSgZExIsdby7m+gNFVq/sYoxtbsr4CHmgfgGpjN3Ej38QryZpq+WCFUM4DsB7hgSmP+S5fsv ks9suuARAuqOK3hfdlJYlQ20/wlML74nE0hMr82h95eK//k2xKbHB+LKMpQqG6M+BdfWx3h/Mn/B otyCyaMRwn+P3Yw+EO8OtQB+8UiHI97kfXWXXVzlLj70gH868Lzfa99eX0bEq89zKJaBhPNBkfj8 9/+5uNxf7d0+5bHlh8oWEeAdapy0fyhAFB/F9ZZ8J7SepCIqg2CqbM2ULjptKRyichWoa/ix9gvG zYnWlodKM3g+estbFdLTYZdwzixcpyP9Jd+WanQ9EW1tEGkmbAoQ5vDgy8ao/2wGU4kkhNYWBc3I zKQlct+w1qqxLE0xEoToXzr/twcfEovQLEd6t0lFR7t8rQ5QbsKr++1dsRbWqTtMXBP/pbZiH6Cw oGw3RtPmc4MQdz0YV/TQ1ISJwu+15R46DwnKsTiAvEHKurHvHJBAGXNf7hW3J7fwDUEkHwwcnFWB fAVbBEoFm09jANe4y8lavn11ielNrivUF6niY+OYwwnpJt7dQrjYMJ3DxgcKNX6EI/N9B7wof0xp Ff0CWJ3aWbNHtJ9BBGr9yfnmoAzmpOLDBR2XborB+mr+15lX/ASyknWGl/+4kshfRxLJH3Jl0il2 j6Uc3/12qfrdEM3LJ3fIQdhyn+YcgvRaJISZyU6kDBIcSNqV3xbiO2D8SUeeW+/favv9ILqbeFOu CQRNTMTid1fVuR3Zo7htkDB5oxieGCQyE93Xi76BZfO4QHrgMAwKbvSaBIOOUe6lXFbCLpXsKTPO aNRSb75A0tmwzmiJ7u8Uqm9f296LrWXo3l7YaOWAcFX5+hc7Z0xEEe/bH684ipkACS4bK50wdQ8y jJv020ukiK5uxX8lMpChTDUwxXh64oXEBPA+I030I6YH+xymWcGAtkqJXEeOc3oHsH8rgLLWVIZx /tKImtFht5UehHTVKMpzxj2fsZzOK1C5hNbsK3in2SNIVOeuO18CDkju5CQFTnOFgkb4v78c51+l BZjtmBvEwSmpg+nc+/59+qA/8bqAg0qm6AwA8SQmvD4VQjMZQQqkrni06rTco7pCbwsdI+tVlAgB GbMM3fhtEAr7/FVAiOd35IwAPLlIGc41HkQz/PLQAFJ+HzIYqv1o/V/dqSqgRS4rXMVjlNvEBo7b S9F4v+XUknqLBtsr/BhNYgDsot28Do3FjXZ9avW63pfm2HuuCwNdfFXae42HJFKUognEPDX+rZJp UedFEHoefTUI4DF4V5aoQMjIXNNagcKN1mZ/TcEHYU7+K4mCT1PjoQiJ1PgULZiDCSRDo5wKVwsN YTGi5/mQDKJRI0fV2rLd7KvMjIeMrGnZn1Sp42hwpwM9n+jL39pUo7dDH/YXOvsOkUk0m2TNStz9 o3vP+WPUXPL4Tf6rav3Sdre/U+4QYJ4dJc8aq5IS6UZ6uumXnWbCpR22escZtGCel89/J/6xoOBh 76VK6KG9On+t8soSdIY0hTtpby/fe0tAk5/4/lxrx/Jp1m/VYyfYl7uoJDal0O5CR9z/fb3AUjOf iE+3yl+gP3H+9TFUj7Qw0zuG0UT1nJ/lbmJWLPSKeWvcwPplYigFJqkH7iUVNG1uO0l5ydzHDNHp 0HOWy9w0b3fANa/E2bp1qLyS6GEfrIfVAOIW8/MZ94REEX32RyvMhE3zjnUTfySsJRJjXplYBBC3 Y+KO0MDSgvU/N5zEKAbrT7WlLBCeIrercbBXBTZGBbWwn6A25AvmXyWC0+qZFb+ZfOAh/2+hULHX N04f7iG6ezwEOT3kJFyMRZ+5BNZonTLPxmoZCZe5m/D35BogifBDaSIvP7bJBV2aMCFSHcsNCmVu ZHN0cmVhbQ1lbmRvYmoNNDkgMCBvYmo8PC9Dcm9wQm94WzAgMCA2MTIgNzkyXS9QYXJlbnQgMTQ4 IDAgUi9Db250ZW50cyA1MCAwIFIvUm90YXRlIDAvTWVkaWFCb3hbMCAwIDYxMiA3OTJdL1Jlc291 cmNlczw8L0ZvbnQ8PC9UVDAgMzc2IDAgUi9UVDEgMzc1IDAgUi9UVDIgMzczIDAgUj4+L1Byb2NT ZXRbL1BERi9UZXh0XS9Qcm9wZXJ0aWVzPDwvTUMwIDM4OSAwIFI+Pi9FeHRHU3RhdGU8PC9HUzAg MzgxIDAgUj4+Pj4vVHlwZS9QYWdlPj4NZW5kb2JqDTUwIDAgb2JqPDwvTGVuZ3RoIDY0MjgvRmls dGVyL0ZsYXRlRGVjb2RlPj5zdHJlYW0NCl5O5iFSLvd2yakiVrX54qqYLXi44FWl3ArP4gXtB15L YWH93C60RMyYPX577y0iSTjBMdEWf87tuNG5SrTpTLwVpLaKH7gGD8YxVyZfF2skQwhTL6R/trn/ 7Frlgwrm4EmPx6iDNjFrDP+Iuubbbyh7wV6/Q/lGMZC20MEBD3+5S7iG31tZFonuYXRX8oTx0UJr Q6XNs6W9SBWnBDUylBwmed11T1kjf2a7NP/P7vFwThIklrNDxu4fPTblIP5e4RhA5vDka3jLCtt9 hFWS84L64sM8qAh93Ze7pp1Mha2FGdv6v3CLLrA4F82QJX1HWe3a7KHzgwK+m1+sqTwqJqmH28rL J/XL+7pER3Cx3n/I2EFoGNfnbkF3qIOVSxuO6ESM4Y5auhwhikpjVqn9ymwK7afqaz6b0GH4tZ39 RbFbA3xzz5NQbLshhoFAkXeZs1at+05RRVl7dhAC/jvnqs5YLNuHdjHumhA1rFeM4GSBAPJJWxwH YqkBH+0WfQxh2h7mP7LuJEJhm9IAAnZDH3xBrNwahVgwCCM79pMUhCp+NQZyYyjZdlwQdgoP3NbC 02X70tMvZTkmOwHgnf0P75DVpUMKfrNeSwVD3dnK7nl+hyE72cLPpB49I7hIV/PmrnlOYjmxs3zh ZLznatwTJzm5xMoeIR0YrAj+OqxDuMzo86EZMuMOPujltRXHB490ihfgwbgU1UtLgRuo0WohGZFr CpFDkJwvcNRg4HrZLsogygzGJzMLK6O3VkQY6amppQIdVjyicq04mVmCxaCX3i1Eyakgp25272pv R+MbdUhNEpSd7BmCHjRdy+rXLxKOt6yUy/9wgQak4D6ypvCS5gfs4dptUuGTapz4jN4SNZuod1CU sp4m0hQy1jMWHeGOxwIvNdSPNsGxbvVMN8O1eTP8/OwGQFtX+D9yW344ZwHIKHBfhvC3rMwn/U+q RWnci7jIUFc6YI837HJBX5y+RvjAWo9dvBqLsbyUentF8g4qLdRfqnhdqnLL0FmXMBcRsM8Vh1fK uxMxV9Sqsiq+zo+Abvy7IgsWuKiZlzVO+ZD8SmrcPc+qDPFKYNAzZ9EhDjv7fAa7wpSSB5tRhYO/ YvdfoKs3eGc8obsBzRYGhIpRcX/UU3SWDYL+gakcbkOTXRLqsKvIhLnrPg5XoRrpQm/TqPpA1V0Z 6z4bMhUWM9zQuWO5k/0wRKnMO902SuccnD8qZmJQVwQ7xhl8Owr+80JTj51m5Ywpej5EAaJ8NKeO HS424RFanEKHu+K0G/ifcxPuSRDaluij4WlLtPR8941z64Ak4rCVIthExO+8T2b/Gh2m0Lnc+xia 8JqxDOKMyIJVAp4YnhARriyIB4oPLbgHqUpiI5EkUN9ElVVnig1zmdykqSqJ2kAkNl+oatMx6ygU VtKVgBUDH1Dbpl70GajbJboF10U+SDQ+ereUrGKK0OnK15MFE7LEN38b/1t2c6BOlEyL9AdOsMVm xfglHD67cvoUvG+BLYtuoi04shBdHh56HpGYrmmwMUYzccjTlC8X8X5Nwp2j+0e54aHCb2mSXoCA sF+o1iBfRs9wAELS9HrFG8NdxZmLo2zbrMh8mW0NdWF1w3HLmery2SnW/Z9iOPKzhRgzGJ+N1k0X V1x4xnrGJ4YeTEWotEbnqlygZxDOfZeaWAkn89RaeC5SrCloAwNh7XA3P6HVDpbM0cqOGiNWnoxG 4xnJqf/gvYpKSZAJJLUGiDnQ/HMHVASQHcvZTJlZy3AaKZRUk/qMnze9QKnTG81VXd+N0Enb45Pc jAmaODOipsKYDOikJY72S3ccV9cv5lV3M+QSp2S47hFygDHEJMOK/RhmJMyoaM20bMugp3Yf0KEP bSUqM2oGFZytvFFXc5ZwTnh3tbOPWPv7t61KJsJEWExgaE4wCkBtAM8MdoCccq0+KPX3fbGfLvGZ 5TKUYstZai4QB9bEiXwDqT4cN8DEilW8bfVPbIHYA49w2rPv/3Z1KT+vCPMImRF1RWC6Eo66sam5 zp9Kwd+L6XKeq4ihJNWGQ+oTTkc66tu7PgcH/iWpQf7lbOsYDdlV4Vti/Jl37Ox3pcPicUmjZNLy V6HcXdd8jqLMtEW8PF3880goZJSFoS8EBOx3ys0iVIIpLTibXfOstvwV0SuF4U+eiX4WJwNjbV9e YkQWwwsAPKXKt21JLSEXCqtNuo4Ubo9+4F+wcrA4UkqtrkQgozNp3ceANtb7NVbnBFpiAr7J9sks ly2cSJwXioOe93sLiKkBkDJESAxNt7jRbzhM5vUVYBoP2GYzD7Go4hhdtHKrSoax9Uq76lBIUTDs uBsxn7+pjB/6BbRm+6zgAJ8c+APFdx1hewtEt8OQ5XUvKdvykZate+aC1CjYBz0awxrNJ7oXOy8T GERUZgv1acOxhMhrlWw/DEz2RqC91BF9wlswafNReS4mQpCpLYMVe6NqHXxYVpVn+cpza376zsr4 Y2E9AZt5tXLn8ulpfcffjXqIa8qScWzv8mM0XlcAEMrJmZX/20G1UZFBnO+Pi/PCH/3cQ9I/MPBL tKiTtwJoL7RTZb2vcFG12Y78vOia8ThccaM5Ui1469WJ9ENBU0RLHdSDfJ5EpW+uvlzwffjiOk1U squ8guAwztYLZuiDmM8mnMdaaGrPYUvAA1Yb3QIU967VeyGFkXuwmC7JjPBSvBuiQ+QlSwWRscbf C0fMTud+LPzbuWqgFSoe84OUk1+9Gq0KnDKTYGvcv7OYaG+sjAiOy+g7NSa1yJPhbcYRJ9dldHHV GzRJxMCzN+jeApChGbUVBL2/0APYRzAO2OT+c2/uexaobWOgu/jk+qxR2wFrZTish01Q3TjLsceg gwipguvmXjs34JWOi6ostsjIYbpUXYI4DyxJvfkrpye7hTX/94m4dVdHS0fx5RVY8blFThodwAiB kLcH6IM8mc7cLCQnovhqNvIGG45NKRpyvExfVKgGJMAixXvjFNAbJOKYD1d2TK49SVx8vL4Ei7WH lvzJWicjnhUp3fG8m1gj6+2CN/95rok7Hd3Q587b/35bHFDwzY+7aDXZKe1/ZSuX2iQsXnQgtBCU U/txicsagLOLUp23IYrYvB+GBwu7YzUIP5lS5Q78TAUAx3LQRLxPdB510PAUzbaurh9PZ0gSafzs LnKn6vNw44rcxhN8nHXHd01Zs067tz29FXU0T/R5CfsUqvVTqBmoYCSbAVuIoxd9C1suhqV65G3K gDSGjCcoFC9el9Y4n9/3cQEwst/O6pgAT0E6AKo4ELsZ2/AEKHy8n8E+F8mi+K5UA2wyj8zvGCrJ zLRB80e0JzvHoC8IT0Bir3GfkmBFVsf/b4PGxwK5BdzZijVdYlyr+21YVDU0WlNuirX9Hav6Gl3+ nIwb3FTw8NCi7O/ZDNGaGKwQYJTFUBN1kIfpM08wd1ST/Iu8iI3bgPUxslVAjR7yQFT7KxD0em3+ LFuIRkDkpWxVaTA0JgFnfx9r0EcetUhGF6wJlADFTBB7Kf2bwuakwiWU/mkeZMQTv8eu+YF5k8Zz Z5ArZkl7J+zVsrup09SoC5J7zTNJENUnot5eczyZdP+Gb/j+Pg2P1IBQO1GHMD3cpkdtLTY/gx/a 1vv8zclLgizCmXhJKopviY2sg5m7KkBTmEMHpVoqdFDYv1O1/oSMGfeFjRY5O9riYUtzP365n6HT Lwb/Mzba5PsvlpbjTzRbXKs1R6R1eAXlsb5kqFzmHZEziA5aeRmXkHiYSNR/I5ZQt8DXJInnNs+e 6ZQXiTvSD8wXWhRg14hjis3VPkOH9w+yd2/xyBjqHn0pMM0uhlAZGuwoY5havZ0jL7Xwx5VO2Esd 15u8c4clpPGaWtRRVHWl9ORN+ZxMP5rqsHeSCC21DBbeneG5Mn2DETIQrq1Fz5dOKtFG+l6xK9Gc xRSEUbj7lT/lm5mysAT7VovteFb47dShDBOB8Nv8yv5IumqjQG5AzTScnDjIAdcL6i8GHDm6hw8r oqDG3bRi6ubMHk1EJtZY6ioFPqp1fysp5YD9FiS64ttitrVCsAaGfthAG6d4Vpv6jlQSl20cc7f1 hHrZRZEK8cJc2jKWCUtPezu7HeRXTa8yBTs2j9SnkfRQw9vYD/Pcj6rpBoD8jiwy6Vb2XaCgUusL NyG411IVUOO6vfE5Jx4xoba7Secu//zCNPE3bGwK7XmQo6NQh1taJ1kAxZxi4lflFAJhkhg6OBrq I3L5TtBBu5XJBppsdos+81mqOUqj3o6EeEqRPxWAbn3N9OBPsaSQqxp7IBQ3Gpe0Y5Ps5xA2Fu81 z8TjA7rddutQwlU8omP06IqkjNZr1YUJaIoHkYenuJoWEXcHBuGIzgQEfC74No/0jRfhCmuu/WcJ 0yTc1PeOWy6/RMxK0tzoOC4Fo7ybkCPxCsmQf0WiScDggrVclkgpEMya164dekmMRph6YCJB9/s5 Lv6Z0aWh9PWlU4qGHOjdziqp9iblXC4odQGh8pIx94ZhHYC3cZUvTAOOg/QgalWRgTx+iXF32sid H9ngC5VFtfMKRzOjxpJqaZLFM+WPkWJbSROEKpwttob1B7Sb/xfMnqzbk+QnIiBoMJtQy1uLWqni 0w+RvKUR4ktcohTQVHILE5aa/52qPugNyEBubs0liIIwudofQ8BV28jYjmwAsB3aEIWy0QZeB/d7 s+MHsj7pnzsjAC6L2aJyZ9qoUggj+TZMwjX47qN4PHpG8jt/73xzE6TBiWETamTor1ZiOjJNC5Xc fl0wBQkOGYwqeQlN60iMY1fJl6/uX1hI91e83t7PtOtVOepWwlvE+4YeUOJaPgJgr6TbXaa9qcPw vOdGuXucRPdkSIE5ISigvpOzpqZVowl8FDjle4fKoAVrjjwqfVIw2Qv1n/G2T1ctWnLY4AdV7Pl/ 3rfqGM+dWsF4uXJ6Xax30nImXIO6U0zX9YonQ0/HBNwF3pfCjbGHTua6XBuo8jXISlrGzZa5i1Fs 8YfIWpPGkh/WQuwq6kwyUVHss+xlXHdxy4aMVTGUXTgtlBv01sdzM83D9Pt4wJ5bIHnxlsWdwI29 O7ZrS7TqFx7QxgVs9FpO0v6sf5RNUkHgWNi4Zu4Ij1r8yCjBgZK8TSfIu2vlDQUaLZay2N1IguZe i/noXtFjQZOlNOxhi9PW2cvdMkermQ68RWM46U8ymzra7WhI4EspXv1AQOZLDk4tgXihhejukHFB K3fO8OyPN0/yzO/RGGHQEq56On7dRmqw9F0JXTPcmzgs3aponfxzSLQ2lartGJX91Ootx5wVWxwf /0U7k8TjhymBnhHTsKonw7og10iRqjj4APF2+PVwA5XVDH4p+B66GoC8B720bgDaRaQlF6/h1sHo A/2xEBWUhWIF2nWD5cJimxb6+aLlOyUmhp7MVJsvny5GcZ7m1Q7OhTWIJx3OQJczjA+bVDCw7c+b Dhk9EPXwe0wVJpj7foQi/Z6LCina/UY7g+dasjbE5sdVqqaHiWDpkknDeCntDsF+xkrsxwQQqlp1 ue1KceFwRagfsXC2lAix3IdlTciL2Z3cVUOAS184yAg8bnBczfFpGZlqNoTyvtwjgme2Y9sg/Cdu /LjTpkIzxNtUMdgDjmlPilZ80T7SC1s73N96aBqmvDe7XMuNdUFX54D7EuEeFPnnKoiZmdbcBhg/ +r0KfV/6DClLKpRmw7XNAD/NXSM9ceLecOmnSzWVGf6jOtJEF9HHmNkN1TV9BMG0qMvUeDRSFRPs /xc1oH4bvgtu3eg4DoENXSnWawbyACLyoCG5Pngyyqqpho6x+/qwHBaokJHFk70Ocm32FgW6lsVx 5ofdc1xHK+O7VKIzCwt9SK1zlqgZw7BzuEm1WuHCvUj2MoJn94P3gwNLNek0W1xLgrfP+ARVlzs/ OqnsGfQVFY4AGt+pc+Rx4hsVg04Ss3jQPklkIDNmdArMuL0f4fXwOKV5Q9Yy4Yak8DWJupMTSFOa KuBcmYScDuJh5jmJI312JstP2OKNfudDc4c2scIcQBGzNDCqL0sB8O2UMkKP5tGYQJKpmm03CGbc 5nXhtC0B8UAUh/oyVaunw75bQmRire8Rvn3AtblJTiUIaTluz8ogtxAK4U7TlToHrzQSVXyOGF0f 9PB/cgVJq/CCP6vnOZE/oL1JYn3JCKWkKdl/rK9JF3T9FYVIhNr/YXkIHBPk1+U92xixUIVWUgG9 2J8ELA153D8iBjHVLO2UcRGNP3QF5GRKDNZoVifw0VKOHHYOW5s1vykDKPyWzFknjGs1dAEk6E7f w0eP87P2gW+Cx6zN+4HcjoDV66YFTRk5XRbCq/yVcJ1Tz3y0qfl28/ObWUp0Zrh90K2x1z4Ey3+H rVUuf8yitBIWQtDSn9OZEEFfLW2IF/iFfL6TtzMtVl3fwEtypdW6JBsuoR7hqUqlOj0tI4xUMUw0 aIUmXZJYJCm7ZraOmkfcYG6TJa2vsBZ8iwz5pDK5VMi48hXyk4A8nBqMJqHPZBj4NM1HKUuazd3L G/yoeCOaYsj+FVLKFHkA8Rfibnig/z6wX6mhIywGe0i4VzcdBBS8NUPhxmnehZVdfe4uvdmyb5+g IqTzdrxn9ktpX+o/n/l6dIn7KDpLivP0CNdWWDL4LeueurALoIMITAgWkHSpWXAb91euYdreqIBP qi3C0gpcjWxJa6Ou4+xXTCWlQ7N2K95GHQpBvMeDQtGw1cMInnPESJENnvP+H0qIJgZmtahqfXey JCtMUcqUprUM+zRsw7f2ApKwaikuTMP76Cxm2vqar2oFSxKmnSMLgZIv1rD82mMHwqQbzvHNzif5 lHbdRxGTytEpO+KsG/XtXTSBt7DlBPCx7VZK0qu100+ENVPqP5Drx7QgE3ZZ18g5E9q/DWaDQw5m dHGkNCf1AjtPtPtTNM54OrvKh8CL8LdESVxLrBXNTb7Aa+RSH/ivS+59wbtH8puMzi5qvU2UQm8z 2Wg8QxdhXKDO1iqvJKboaikPfxAj2h6O+6NRy6EtVKJkqWxQ+OXIGQDbpY8+/dkfxjzeS8VVUIgw /pK9hMTvAEKhot3+n3rUDxrD2fiLN+6HbCMpI3lVHh66d+/v9ZxJQiAr4jHxqH/qwuiPsFxvgRI1 sIqFJffxZ8ZCU6xFftyAJKPryK725VFlPslAZgNKiCSx4bdWKbQVB0YpSZZlVfpzVIMPzQNK+paC 4BRgobe+iGqt5h9YCVRorqt5rBTWCEhe3lf6qHTAfcflhsHBiVd6lhjkqs+CW1U/Q38UXrRoSJXU ooY9JFMhWfy28O7IAdRsO8JOMAahNfV0X8EnnHovbOseI7SgwSpmJXmiYVa4LMbW1gFiLbpREzMC I5wUvgbiVtVWLghBVy+nEsXTsvsQnRJv1fIhayu9uvyiRKFtWy/L02jLl/kT21Xf2rx3iR4UUnGH i6pz+lNzsAZrmaoITp04FbtyHclFlVeDA+hOlRRVv7yRiTkSD4he1OyVs+yVOjVP/ksT1hHPdJpK p7m8nFvWr6f5NuAVSvTUc9vEmS9vf1uerFcfEIRwA9kNs7TqKhbTZDoIp0VYPvgXFd5vkFgHNAfR 390eK+NyJUkPQjf1T/hnByVvcEXDEjlrW7sbCgWeMY20Aw0+pwW4Vl+pCm2b4o+xYNyIWIt/m3fL kuwduH943HSK2XKtcvpbqr+OiTn7isFaHMRnMmn2wLcw5TZI3p0v8PfQSvGFjYFtHoPNyqRB6LX2 EVJS4pokCOlsrcJKUSkVft6CHPtsQ1rLv146kpZio64PvgZ3p2k2XY0zoZHjcbfwvJWkX3XgRFs6 vU9/DCtDTxODeIJoJFziwSa5+/737xQxOUbOXdQ4caum/QMSuombn1+Q/e2E5KzodenXSiOQnBrn nSxnYr3Gvln23Vg2tzgAmxS0DDA4vHHUWvd3uEkGQ68V+XAoX6trq0aiEbV850KfB7/zS3+t9JkW YJ7fnyQeKZvzS5zPQd9fcSqwMPvDau4TarFfVDlcTPW2ZkpGzleosi59RWmYe3qZcNjDgMrHu/GK UkVKN2iI0pCyFExL48cVxqBO0iX06Er2ieQb/x365soSD0/VTE5tuh0zmWnNLIbz4PXWtNPaJYpd FQHqoC1BhZdtdNLKTWGvAUs0wH8apvurRgVlXmcimY2RjsgErprYIFcYcFQnrwZDi5/hElFxrAZB hv6Pec8hFJsFTGdQyyNG1w+6mVCO3txJrfVSg4YSyHELNX43aszOMxcRhUSzEDxvrGX4laMsHKE7 p8QJ0VTY203uwla13D1tnCl4dX8H8XJtzsu0C7KP9UnC4TBZCbIxhxd3Chh7egp5HzCcbq3mScuL PIcWdVS2NSEcXJkr2XSO6F2dK83GQf8DJbvoA8SrNaL+YAh+GGajnMkadView+zfI38FV47BKd2K yNoGyK9xJICviFX7H/ppQ9e/mLxRYr1gO5SXkPj5VvkurH3sol3i/PhbnATsaLpUe2LcfZbdqDxT bS2Op+7bm7aGLTZH6YhHoiA15Cf6vVZiCRwP+kIGhRi9rP0JPXObIz3D9T7zbAMNmy/7XCseU2dT EpD4YIlp+OvBb7HRDQplbmRzdHJlYW0NZW5kb2JqDTUxIDAgb2JqPDwvQ3JvcEJveFswIDAgNjEy IDc5Ml0vUGFyZW50IDE0OCAwIFIvQ29udGVudHMgNTIgMCBSL1JvdGF0ZSAwL01lZGlhQm94WzAg MCA2MTIgNzkyXS9SZXNvdXJjZXM8PC9Gb250PDwvVFQwIDM3NiAwIFIvVFQxIDM3NSAwIFIvVFQy IDM3MyAwIFIvVFQzIDM3NCAwIFI+Pi9Qcm9jU2V0Wy9QREYvVGV4dF0vUHJvcGVydGllczw8L01D MCAzODkgMCBSPj4vRXh0R1N0YXRlPDwvR1MwIDM4MSAwIFI+Pj4+L1R5cGUvUGFnZT4+DWVuZG9i ag01MiAwIG9iajw8L0xlbmd0aCA2MDE2L0ZpbHRlci9GbGF0ZURlY29kZT4+c3RyZWFtDQpgec7h ud1/4xhpvZdp+/ptoddu2baAM0N4yQO8T2iBHFdoGzzdc3ZYiWIHeX4QezEG6B/ser+D4ykgl2tY JWrulefS0lxEHfKJPchamoIpqcHdEePN+ahgcIw2xE5sfh4jQdN6DsxZwvcYRYlH3zF7IrrKCMnG CMYAXO51KGq+TF25p95S0aIIwbYFrsrip/nk90jEfxTiTNX3UUfbPk5AAJRj88rleAnG+jZ+b1GP EOh7NMULF6kzNq56y1ix9NLou+KLK7FvlTw2v0AxPUB3T3Hobc/EY5RphgfCYvNSyW4hF0oA/SF7 pwK7EoGTKPnW3HGaIarZz3fz8oxNFOWCo16M5L/qz3xEghPC17XwL+jtJXNhpSe0/ctqyeiSeUra 0JOBbR1xHPsazn0Apla25hc83KiI25jhs6Zhr6X28iR26fh9L/pK0Tp8CgmKYhEQ7Nzb+570PnZr dtzH2XYtvh/xvE3lVGNGIWcEtUNQEqfaV9EMQimf+ut1C2Xe+x0tuojH5g5uN57V3l6IPYD3UZm2 PxWECrYRf2w7ZuwUudgLuY0Ae1GqU9n1M1WMLEFO5X7YvMTNlXzr6EwLR8RYQdXwASkQ+KC9xFFs 7t1JGE14G7YFMZirmf6++PGHJ9+6Ha88GVKbNSe7KH0vdLlCDri975W2faTn++1ukBfthUZjgZ+5 jc1JnGMv3IzeV4O4+strLA/kgnbIX6uiqjoAXconng9vTNcwDgFmA7wIz0hIo7/zE29hDAbBVam7 /RhjCWhF1lXQxN65c4O9b5s4y3CF6PuJdlBIp3nPSn+U5Ab3t1wp1MW9KekF5vVqosSDhxAWh8db 9t0+IyedqHTc+WAGSKDl3cbiBGNsr5CWm5DGVI+HygUpdsGmqqTb6yM2Iblk0rc8bo8R5KpvZhmD aqLRRD52siSGDru7oAC0ANsfmRG0xpk7vYIrjX94z4fkXj86UQcBPLjuxdD0pHBKGLVF/y4/+LyJ 4HJsHekj70YNbaVKBEEHAYeA10BOoIq+VO8Q89W00q8VEtOqXzZja4TYrWqmSfeQ/GPDiAvW9nWK 1+OhDG0KTV0Lvg1ury6zlnijFtHDyhxedpEZSQxGif09TPQ5X9Qvhy4nYaRY6WYOtVZgPNu5pHqD +9/sDWK4/1fQH0tomOavbHy+jaKBlBGLQfUdUo/Ex1Sb1ZgTLtqIycrBYK4h8AarxsZ2GfN5PhoW QBq/3MIDH/GhU13Udvp0tzKPchr1xhPIudc7ak2Y/Aj2Vmqgt/j7g24k3+xzoBD//EFlxZgyegvQ xryLVRO9kPbclRHSAOVsHLGt6R68ssBRpOLDOOBJHX7EtzWg8fJ0WHnRndDRfQcaZ6xVu0ofceWi 9WB+X20G5I+Hpq+iOY9Zqu2/wckZrd7ylCRXaD2IqAuLD+VZ+dax88n3DC86GPIW56blPbRW/VWI sMEAO9kL7dkUvHsZuZE7p6my2Z6WHdMgU6BKpKZ2iDojIKK91L3CiuLZNbqEOWfbUx3EiHmsRERH MwH4k07Z6vdqvP/W0lFQYDvm+p5sd8qS23/gsehQ/MLFxZnf7mrj7ow8qElSubBydG20MkmHg2J0 zK6bKk46q4qbLhRKdddkihr/ISLWmRFAYpy0n3Mkd9+ywzK+HQ81Ls08yoHpb2TpjRGdKhbo632s uZhNa06kEMtX6nvZMZDMRdhUuNFyUPSKhbsHR0V1P9zEyttHDuS/I/mJt11qjdAvHzcjUEFVDUDe TjU8HzvnVKNAGo/RJ022cMhtH4u8biQhXkThd+8p9UwAYo27LCA8DKNvkxZc2hl8F+gi8TKERLuL 9kVfqj2tholP2zPsXimDDmSY7lBMpN2dOJf0pgU0GKv7Y+rGdXfOKj7zYsYAGK7sy1GfXa5GrhMN UdHhwTdP05DfZZWwFUIXM9XE5AR2V0JuFeNFDhftWYd3UNSm3G72aA+F2P7+uX0RhOGGqmkzC+fa 2v55jrQAU9iPXmWHB5mas+iG6qLeMeMQUvZaxHP3BmC7ERqHTwhbUH+HWiOCeAw6oCyQQXF6tT4p 4ahb8FKhqRvCJm0YQbmLQAwizTRyBSzds7M7WqlYVhrEhUQ8iAcINttSIqOu9Pzpgk8h283sNrN8 IO74LrKzA4dCtBqnM6Ac4Vze70HLG1H5YsSWtcCvXNVV7cGEjFg7S7jrlDFlAxJNxrxva8UK2CWQ XZYCY2yw+3NhnpTqM85R8Z8Lbbr7iba7R7u5vbwQhHKWwYXrjmXoc3KyktvbTnbdK4ff3JvBhJFA fsMYFdWFe/Ni9tL7XBa+dBnlmSCsoiBm0uQ1agDFtFcq0VSEjWenog7BJC3MfQvi95s9xpWjjUd6 l8xQ26vJ7s32R+RglHfKirofuY2MkMBdSm9hI+E31yGcMxJj2fG+JmMfME+QTypy+gxrZdn/2+Bb VULZjN9HpxND/U8hfaZwI3ftcl+stS9Bt0FbGDkIUQtxc3K5H1VE+ZSlIBlFd49cAaFR2d3sTy2I hdMf2OzDs61M90x23ZRo7VFC2o/UAWqWxPWd3iah7VKYYiXv1TCaGvxArz6jsDfGD59xCLmW3qWz uAcVu6Anoni7H4iuNbXFIKK8/7tsUlGC5zqM5V6B3hQCenMxT0DAA0gPhJLu/2Mo3En8esm0Uodh +jPktSLmwQvg8esdcjdDESYW1VjlYgtlBg7pIrvBYEQF1cCPXABpQwMFLYJtwoFFbtFGLlBsSjpA VQz8K5tauhW0PpJNLenRRVi5WeZF7mXoSpgr4XJ0qLzjXBGVXtB2skVFRTXN6/aSqi9AtwdtXbrF ASLj4ZRjgxvYBxvNUbFBUwX02c4FIyMhg8h6trRA5TF8aQNvvqAaPIs92gBCExwmKRxS+Prr31Kr bOeOit78KSXffj5C5LZKnvnVqlobbg5llHVn3sBZ3ZLWgctJLlSEjd7RyC/CKSk0siZ/mjuM8+VV qQrbKThvu+EyTHwguRQZ2Gkq6RRRyDg+XN1bXnw2gdLog7dVz2LueZ0ETFf4XlZAGL/VIKhMHmFp HhYysrejDMmMg85Gq7nkIQAoBc0u3msNTGzGNFzIm1wp5MwjXblSF25zlyIaWC5rbe81EgRjJyI/ B7FAgjMvZAjYLGxGoMvI6D3mdcNidId1Nf3RssGU0hSOk1TpcoZr0ybspnKTGkr1cCPCgo3fGIAM LUIGr3fGRV6WEw59akcoAKAsXZBtUzE4mz0Ah8tZiUmvqXKmCp1WGuK37UiocoSAQWKyz3Ctt2Sf gJW914legBKT+XMIj3FgM1aTAfksaTkA6oXyEy8tHavV0JxzURaC4qKCCpSzw/5OkQJNTNH9dD0q /fDzng0pBRU2XqdGx34rzY6RJxne58GwBUybeCfj8BmRQrTpKbUlP1w0rxMfSLP749M7tga6aI/1 6KiqQUbDaMoMiKF+zPPBi03gVa46lteS7Q8AMOno6RkmWnKoNGA1U21LulRF3z9TfpWlnv/wBC38 Zs1L+oVSrdbMTlLxq9cFzdD7JUgVP4MHqWLSSfdgYuvwBY6FtKFnmtMAr1apz6Zaefaijfqpwaxe A7fyHH6XOGfsf9m/AOwZTm8mD8VmfXwEmX1KosTdr+pWSuk6jAqp++e49eGM7EgJNAM2EsKXxIIX 4TFqayZMTpaj96cAQf1cXADU0vMMdi6C4Lt6ROwZ1fWa74O6REkE6qgozikqbpvG1mKJ3K6wmw7g JrbaxaRAvdCR0gHnbgxqPC4afQiDYzjd13aiFpIbUQ1CDxW3bkgJdzMC3/Ezib+ClAeas6PeRtQX pourhiyeZxkjeMjYV+BnAN04YEL9omuw5b2QDaFup9OMJ8QQq6ChYOW7zxM5cPDNO9VZAaxKIH8e IUM1IClkCplKDsZERI2KMW4AWCpuopHd7zyOgkL6bzR17Fl3PKdHVnGDX0kzYyBVJehj7akYljqL fW0x7Y4Nr7DvpF8FGCMNdxOuh/NNGBrodojuv3hfdJ2JwqBjUH3/yD70BSHMz7Gn+eGLOJrE7Owa lgWRMZ6+2PIx/pX+XaJvR0NGw9cVYb82u0uy9gPXPSiR774HAHu27ENp72/yc/x+0Sj50c3eQfqT EzK90EbLFpknSkJUHDZIi/JNPq/QfRb9KGn1UATAVR158PiXBxQ7q9KTXjiPlj3M+tQJ0W3X5RLN z5PD0OweuuqGoL2ayae/Lcdtf7hEhtvIa/2QPP8hlbxTM0Yu6ghHyf0gey00dtQyyjuHRdHqjnJt 8HRb2Y4XiHuI6gzbPmdt3mw9lqEev3LnBucCwyUelqsFLlQVtrjeHmTE+VrChEBB+fErDL1Dxw5N r99IBPEGJ+/6A05CuBK508IAwArs1y56K+fo3Z5m+YUu8WCYLAHpD/PAn8hJ5N2ApULjWulq/np4 R6YnpRfFzOJ82zeJGrFTCGOfhk6Q51tAGQOztqQ9hG5yavRDKRwBU1BL14Y09diBzkhVYU6ev421 db3dJzNWYkT4sLyMazadWAJrbIM4ojB0QYB9hqMwd6GWj2S3CgZYpBqs3G9FQZI2olh4jN8M7Ezw WyN3Few69lHjeQWpbG8Lj4/bu3AtHjsZMges7wUusFQ0EsAwDEfoAS90ThDPmkqISkzYwe78gIcw k+B2+o8r49ic4Gw0pyc5aosWunC+VTAZGBkzojJVqQ+TfqT8flHDP1JczHwIJ6i5FsHwLbJR8XqN AucFml+Xu4vvoz7BCnZLdTOZ8ToUdsSZGtMRhqz1lHdkAi/hT764o0MhyahWxN1I0NC8xAtER6Jl lquHMn8HRocOa4Qn/evnGq8AGwQuKi49e0AOELnNRdC/J1Mt2ahJIz9G1hUxGeec4Sc8gFJ7CyU0 fbtdSblLLi+VQf9YJoigdfdH0EotuMgdDu8UpcRqCabqlKakXLjFf4uPxgBiRCzWJEb9DCP70/Ry eNwDoaklFrFM22koHUQ2lLeUH7T7nIv58A9f19LcY5X6Mi4nJSDgIPDn1W7v4ZrsX39r3OjO/NFX 5ZCjdvki2qf1j3jJdyAq9dOuFbfx1m6KUWd6MxKIiOagj0ztutsV0fiEpDOm8Ahtjm8E4cKmYuvd SJ/Ew2TboQ449Yngo744wQ4Lumk7eGEAp+czLrSe/dOPLk8PPLyUKWR/Aq1GLHizzFobyfztpi68 RKBs6BcG4Lv/Lo8ismlNN2MkVpTlgTYKKX6ypU0ximL7hS2nls5++mPdyVJ6l39QkyhytrbjiOgH fdARck35e4GDfLuzxsBN3hEDFvfu9wU3ldyBTqmLPjFMR1Sc4J/qSOpKBE3F/bWCqs7I+y1HgJMF vSa/YUtV2hgzQr1uk7XqIDKqiQvEmBH49q61NY30ELuln8VFR9UYT2TIP9UrsoAIoKMlVYqencaq n6JyLGyCiHEWMrchm7skmckaVmwlWmvjXAhEGowGSRON4AloFNANNaSgZlVlcCxv8jA9TtogB6gr 3I1v3nmoBGr2J9+T63qUoefGrpnNDONjq9YUiFx8hsyaWJZBGdN+nsU1hW1urOeXyU0osdeNweBj fBlQBpoocG3Uw5TgG2ZT0gm3OSkjxTcnKdqtc8SUvzviFW9SrSqSflFF7cNbigJJrq/kJeMsF9Ye 1NGR6MHrtRBLRhdRXK96Ba8qsfmZhSmIB9L6PPOzCDzsYksY7fXHthHk8fhchuJ8uzAE0cvdAkHT IhJYEqYbhdFVw//F4CFEQIeOy+rNd6ssB1y0S1bEyww9h4EjnidmA/aTkQc2WxJ4ClMttufIORJQ JEaUs7dxDmHQSj4w5Ehl4P01UMEUFbjPArhv4NSD2h6BhRxG8QwAOj7I93JdVKkpAbM5x/P+dStx nOlzoyn99lYHMhibGXfCK2D3wm2ED9cWzuCllarsfeh+x8ood7gg4yzjtzqeCQC4y0bl1g1d5Psd L3xd81Ax8IKG6O9hBDjaWJaVsVr+i3MbBCbLyaK2DJF1tzJMKr+s6xHMoyUQtUyzgd9OZmDk/UBb 8PTn/jzjdmMDV/MOhStoZk4Uu+YvMbEbJVHuCLGJlu6BQxT/dc7DBxVZ/1yW2OU6ytk0jrHjGbI6 u0pyJIw0PqKMZJ5nScVKsamtb1CNxRcgs2AnRoxQqzdga2iWEgJrKbGFqNckSdI6y3eVxb4vkCrR vUni6ppp/GdWGWjfNX3ymWnoMXMmVFtL2U3wd2x9p2T4hL4OEa1vtlhynsSQqViQMzu0hm8+l5Ty nO7cWahOlkAefIoMuhatZztP33CvGM4sbzSnjKADQKyA6gDXW+3vMzgqKAPrJ5yT1GOHjv7ySqDW C/EtR3ErZuEqrjKZ+sBwa56k3Dys8qJ/y61tESngiiLhHtoGkbEO7Ykex2OQdsTnsqiseap24OS5 mCNJoXP74E4D3Tdf+fqSXrzx6ICqs8lJBgtfWnH/WPZSY8zVKtJtiE+xvuDs+AnUz/rryKEygviW Zp1nWdVTH6EJj6wq/MgCmWi9UGK2l5e6GDn2gjG5D7k0R8r9lLWFWq9sUvRvGVrWIW8mU+k4CeYO +BdRLo6g3G0ra2Xn6hV3hrgTsG3HIwp7pg5JAbgEEXeaQF4AuaOxXWwGdVO1lYzAiSVV71JIZwlw A5opcBYE/qjVulnFlOuFqkSwfbCftwDtg2rIf1jE2qQQdC+GsNL3/QHPfZVwxf7Tkk7Y4fnHMtme 3rmRUNAHjybzAuN9LNC6I9GEfcgxcKRdI+wqU0zi0T4FYT/2BzvVGO+KSLFrUd4xjUo5ppc6Kcaz vWQ+s7SW+K5Ul5vzRJJdyei4cheq6/zpswok9uBuNsA2kMpb40aLdwY+X4gVz/U+OilS09D8IpU+ hYH39qlC0MX1FdQjRmrTtpkOZjkXmP2VgRmE2Z0Hk3jkMyHKV48GgHdTIuWZfoFunSx97a/rLmVH L3F1eaIdURuxqn4KCdnL5fwUyLjs+MI1M3WHIdUdEqjZOKi65S343wcurpLult0nFH5TREL9s3Vu cCtacB+dcuFojC1o6FoJPoCziijLfjAVmO6YBSwe9dxYDv6PgB0kQVmTxo6ejzBWfc4JDF0D+qVA sCeOYrAJqxBwR50PZm0/wz4j/OXhmZPPvWXHrDzPvtMh0vj867F1Kw2Os8YCUjDzwUwDZRCx80hW PobJKsvAlknYxVhxvsV49MfevRnM15Xh4e6TJWC7hshqNHF+FOVXaWSrP3A7cbhW5bWU0H/fDSFT 4zMaPbVeRBigMu2PsX38kn0GvbPyMHYv4jdQLWxepJ+QrxFuZDHss09foG+ZISlQ0MG5sz1VxIUO 2/NjDxw2yxR1s/np0n+C97E1HiQCdcdFmoTnga+uAiq5T3ghfdWxDsgMqDAB8tyZ65afRgid88ho KnHR42gTIoJJo6G9wS2OByY+c0/8WSypJWNeurWmMkp2/gSbzJhYQRfkYlwXgGQ9N2r5hQxKNm+b zbBx5THvc/vv5GG27vpV31Y7wugriVrCatVK1WH0ba6XT1l5r2A3+/0TvbNn9fpIFHI5GB8/5x2u 8n1BUUTgfDWzCF6tzAYhcvuNJkxr1vOZEWaJy6a07WhOZ7BaQEGbvxFxzKLtVsVxmvsMK+0ZTXyD BRDZQsEABEAHkXo5lgAVjVR2qZuduVwwVc5FYizelEjslBY2fEwIPyYjYYiCaT/BppZQ3GPk2Bu6 EMZZaO+oFVABlFdPQ9YFIaU9VXJ8/EHBiU4+RYd47ORPPuraZHKuLFA4fs4fu2ylP6K6jRf7Ip96 sSjjFFnwwVpBhJObQkYfTVqdf4hr0DPDRpDUMszgK5A+e5oKnkWzGTXCVn2kVjoWMrhX8lwIseEr +xDWIaVA4awqHjU7CuMVIeYzzst0/+C2brwdt9U20ooP3h+Gs3Hsr1/h7a+YdRsGtGqCfuKZxigU QMwZMxsuiEYwgoK2eB8Pk20fGo9bCWhc+lexKQujPu6BGdsp90iPni4A+gdacK1yc8YKgUFKanab 2zVl1hxIYe70vp0qjjEMzDmY2xib6/mFUrM9DQplbmRzdHJlYW0NZW5kb2JqDTUzIDAgb2JqPDwv Q3JvcEJveFswIDAgNjEyIDc5Ml0vUGFyZW50IDE0OCAwIFIvU3RydWN0UGFyZW50cyAzL0NvbnRl bnRzIDU0IDAgUi9Sb3RhdGUgMC9NZWRpYUJveFswIDAgNjEyIDc5Ml0vUmVzb3VyY2VzPDwvWE9i amVjdDw8L0ltMzAgNTUgMCBSL0ltMzEgNTYgMCBSL0ltMzIgNTcgMCBSL0ltMzMgNTggMCBSL0lt MzQgNTkgMCBSL0ltMzUgNjAgMCBSL0ltMzYgNjEgMCBSL0ltMzcgNjIgMCBSL0ltMzggNjMgMCBS L0ltMzkgNjQgMCBSL0ltNDAgNjUgMCBSL0ltNDEgNjYgMCBSL0ltNDIgNjcgMCBSL0ltNDMgNjgg MCBSL0ltNDQgNjkgMCBSL0ltNDUgNzAgMCBSL0ltNDYgNzEgMCBSL0ltNDcgNzIgMCBSL0ltNDgg NzMgMCBSL0ltNDkgNzQgMCBSL0ltNTAgNzUgMCBSL0ltNTEgNzYgMCBSL0ltNTIgNzcgMCBSL0lt NTMgNzggMCBSL0ltNTQgNzkgMCBSL0ltNTUgODAgMCBSL0ltNTYgODEgMCBSL0ltNTcgODIgMCBS L0ltNTggODMgMCBSL0ltNTkgODQgMCBSL0ltNjAgODUgMCBSL0ltNjEgODYgMCBSL0ltNjIgODcg MCBSL0ltNjMgODggMCBSL0ltNjQgODkgMCBSL0ltNjUgOTAgMCBSL0ltNjYgOTEgMCBSL0ltNjcg OTIgMCBSL0ltNjggOTMgMCBSL0ltNjkgOTQgMCBSL0ltNzAgOTUgMCBSL0ltNzEgOTYgMCBSL0lt MTAgOTcgMCBSL0ltNzIgOTggMCBSL0ltMTEgOTkgMCBSL0ltNzMgMTAwIDAgUi9JbTEyIDEwMSAw IFIvSW03NCAxMDIgMCBSL0ltMTMgMTAzIDAgUi9JbTc1IDEwNCAwIFIvSW0xNCAxMDUgMCBSL0lt NzYgMTA2IDAgUi9JbTE1IDEwNyAwIFIvSW03NyAxMDggMCBSL0ltMTYgMTA5IDAgUi9JbTc4IDEx MCAwIFIvSW0xNyAxMTEgMCBSL0ltNzkgMTEyIDAgUi9JbTE4IDExMyAwIFIvSW0xOSAxMTQgMCBS L0ltMCAxMTUgMCBSL0ltMSAxMTYgMCBSL0ltMiAxMTcgMCBSL0ltMyAxMTggMCBSL0ltNCAxMTkg MCBSL0ltNSAxMjAgMCBSL0ltNiAxMjEgMCBSL0ltNyAxMjIgMCBSL0ltOCAxMjMgMCBSL0ltOSAx MjQgMCBSL0ltODAgMTI1IDAgUi9JbTgxIDEyNiAwIFIvSW0yMCAxMjcgMCBSL0ltMjEgMTI4IDAg Ui9JbTIyIDEyOSAwIFIvSW0yMyAxMzAgMCBSL0ltMjQgMTMxIDAgUi9JbTI1IDEzMiAwIFIvSW0y NiAxMzMgMCBSL0ltMjcgMTM0IDAgUi9JbTI4IDEzNSAwIFIvSW0yOSAxMzYgMCBSPj4vRm9udDw8 L1RUMCAzNzYgMCBSL1RUMSAzNzUgMCBSL1RUMiAzNzMgMCBSPj4vUHJvY1NldFsvUERGL1RleHQv SW1hZ2VCXS9Qcm9wZXJ0aWVzPDwvTUMwIDM4OSAwIFI+Pi9FeHRHU3RhdGU8PC9HUzAgMzgxIDAg Uj4+Pj4vVHlwZS9QYWdlPj4NZW5kb2JqDTU0IDAgb2JqPDwvTGVuZ3RoIDE3MTIyL0ZpbHRlci9G bGF0ZURlY29kZT4+c3RyZWFtDQoKFP7Am8ECIoMwtg1LagfpX0Ak/vFBHaIjE5SzMSRsPsx8guDD by1aVxZg2OtZyjmyZbWaYeP078mhaILL+MQ2suitDBABldRnJtLZ0AUoTc8850hdXbhRu9nsHnwH yf1cwh1DkWr1XU+86Hw/GZ3LFHLgz/nvtXAx9rb1gW6PX/KpkwYbRPuY3H28rNwQp8dXvdxK0NnA B1HU6LdiWbJmj2GiR6IMKGPZMU39Pwuq2OcFabzH8bBtncaXNJtGUjtG4whd7Bn0Maf5nTzanL8d gROpMNChPLORXTTOXM9aNrNCrt9T4CLYZJalP7eHsFPQ7qFkQ+Owp2i6NDfV85YSZCpPX24M6Ctq /co9xnjiSa5ljiiM65/ur70i8CySQsOdCM08MbxA4w0r0eeqpOQbHOi/bSB1DEudw8+QB1lW9zdl iMWmc+v/VuqYZiyyGOIJuBHVwG3pg7NfW6X7FPWb8z49xZ3eJ0ry2SvtYXMzACi2+FwNGJAucJ0g /aV/v/4IGroXVrQcpMfxVSR2mQKyEPxKqYE3WJZvGEZ8rebqXnHJZ9PQF5FfWyaC8g4AG9SGRdod TBqPaxXBkdW05yIA871aE2Se5/+mHVOGJzNdAVZdkRuS375i+7qV+4FI80B368GfNF0mIdOLSFNY lo/wus9CtSUeYGorG6/gWfvTUQPZ1ecVzOQudnBjuN/YSGR2/oQR+G0YxnhU1afNi916X3zulnsZ PjF0Zf/JhSCfrsYPzK46x0mF7bGD8nS1Novvj5Xo8P8kypFl7BRCwZgzfKP1dZ93H7raqB5GIy4t 7ndMtdSxt7pyH6LNqAQgpiaZ09JGCrS57qvwCtIvP/6iaSNoqrC0tzFLgyMChV1ewrD6OKBEKdDU iSkPjxH/ifql07xhiUVHAPeTXKpSty0pGSK0N3vPhJ/WEXEnSvSMATSbRepEmozUQjWlAATOc9mL M/SRTbAxFvhgB3ab/XnriSHxiKdjcDi2NtkyjM7xVP6MalvTuKQXJV7CS5MKbfFtrAilYhBBbS55 q3ILPtcQm66LVachEvbUstdnqgi0f9+Qp6SNvR+2v9u+PlZjJnJBW45+pQOFx6IpNHMxsPH7/iLq ehGW+l/NW9okdYF7IHgSH7XQsqEHI3EqS1WnlGfSK2JdpX+j07Cd7TYz9DpcQi90jzm6xtd6BwlU Grrg/Z/KvTQCVQwI/CM8Od9/umW9KXOOdy9U07/NcpJ5vTfeNKGfKxck1gtKM0rbITkwOCJQXmJX cHcxtg3qlHlDnZHs/QkgaUecSWg3ht8kUsJckbF58sSjOinwmbQY+qdF2aFH4X6l9EVtqtWLDeqE 6VJcl1SNI8AUjGGc3+pjGXUhmSxeHsC/eKuiBPoqfyFlqJ9worYLTGEvmFj3hkgDMiZfPWoxddcb uwFC8TCtj4gODjXKrcFTRCfiElzvgeMTu8D8cYz5lA2E/yyeOmOdbAaYXcImLU+RKzdYm4TB7btp E74M49IZc1fAJH2InXPSZymmiBt+YbKce0wp0HVAib4oXGKZbRrvIm+76z3jaLmZA3F9K8b+Kqak xgX69QhqQwrOoENvUf6r0mlZCuf8IWHrUJy2ZSLMvlGxc8+rSOIRdEAJIXsm2nxEDiUlpdhVdj6f DliX5q4tjaTVBwBoCH/4kuXp+IUNq1+Wv/11y0gVmGTLI8aXStvxh8jOlmpDxfoHdg9NmiHnIoZs K3vDblk7VbOtN3Ku/Qt9jCn5YMzHsLMeVtqK7gKuFZie5mHIF+JHnENRDFCmDPbyCq0psVOMbvZX gMIFh2Pz1u1XyiLUkZXl/Am+WKAc8F/VFDfyBiv8AYXdBiuwNXqQrI9hDw+nVdPuMuWdokiCwevz mCrlOsPU3vbQrbaj+TxDoPatbiaQhkobQj7S8npY6FRkI13gBdnJJnuDHfkLhW1ZNlZf8RRLNOaf 0v2NlDb7jq4N0XWnpUyZu42Q/sZ+7ppZfNEea+3NroFA8CP3Yid0e3UqViaSL8Jh+1cpVz3fUBRI eMDIuWR9w88WjXe1F54AGdqCCRNeF2qwJCitxdauZzkVlsj+kpxnDP/PMrmMixwiS9/P6fw/3whk eGkz6vzMaFxBNNh4ejie9qjDvWBqPSf9Ic3tFdDvSQFsOw1KCpRgAbS5i/R3SdfIMD4lEXWXVMYm WB+Ut5g4aI2+rutAXvDKltzQaWl5hlmWAXqlQTV1ndRbdVsfBB5ie+bPxhp/AOYgN/RBRFzm4pEr kDgVLPpLLcs06X8hXnyizeeb8RRrhk33uP98gQ+CauwHGMBF8czDzhzFJp2qNmDKoHFr9zhkZa70 VZdYdUp8KMWb2hKeHh5Op/tKQMkSjVQQEweTQ5ddjqlqmY6EaMdRIoN0DpUFth9fgR9dhIHSyrZW UUaggEdBkvhZEqeyVc7us+XtpBnWuqYyua8q29RGMMpQiugLLnUO+bnWjr9yGFLeT70tUbh30+bf Ga47N57caDDdnsFLn1A17ikC9DAaIeWYeBg1HAOWG49zBa1ku+/ISe1hUqcDcoQ21vKhn1k/GmTx 3Z+O/y7SuqKqCmVlRNSy61vwED2qV1mhBFCGyigUOetL0e13r8wgL0a1MhBz+SGlRzqNfWRGVVXr B5YdPHdgCMseNK8inOgpRetCKDojQitVPs7v6wdfYjEt5+e8w6RrqQTfqe3aFABoh4CorbofvnTL i3KEi/v3ZsqTZAs27fTTnI7Sk6XYPW3n2o/PB0nEydWBV32GZGqX4czUfj8n5jBXXr9+55VSQVO6 3+49IOo/ggHTKRSZznAyKVDYTG1Nu9lwXryetEJ3dVfoBPaiCyQACdH9bWEiCIId8O3HV19iAK8Y WmAKw3/bfleJU+5Phl3GHHftnYFTLQC2Y5lRSY7F8X11nhndGDSEkMf6slPtIL7/Q79mtprshfio vLYCJps2FuQQYDiL4lnx3nYs1ee+tD3gWr1ewKTs2NqnjehhddrGY5MdfXD7tlS9VDDk7/9R3xUM ISe+mDLCGjFkvMV7nal3rTMqeswZPY/wSzfvmMl0A8AhPnjV48oGI0esDkS2blTUm0bDJq5JQStJ LLO/3AwYEGm0Uj4Jfg0+gAr1DKNkYKhKDPDkfN0JXrLUIdnuFUX1Ua0PRDyujBG8aQzfL6DmxyX5 XTJ/bDVISdy3OejuzIRk6IVvwKp9wEvTFfUZHnLc1MI5JqRafHSmyMNNy0xgT6ASW21RxZerOUDb rKTODduqPp9OZRrU3TsHwyRWymYyNGwGQfwnnDA7SPIkJI54uJpc1EV+dLC32ampcGohGoGf6pG0 xREA6hncgvoVwRUAcXF7leKbVDBI+S6KyG5ArefzspWOhRF+tV/py7tit3J5Kcv9SSDsG++6Cp2G 225o3A4nzfZDxCmGZHxoBbRhw/7Jn74h2ZnnIUDviTdGuAi2xVIsj2mgcbLRfo8I75VpRbWhVea/ f8CE/XP3QMdc0LgtaBDg6ZNyYbcts5a/lzACO49iL/D855l4k9BHyBCTEyfsdYA1zGSUetk22lKr ihXb8t89sMJZMD7fVNAFeXjh4NlIDRpJtq8epMbbVHvmoCeV6QHvVL4ze+vhYH/dVdxrtepVjgBu fSIUjVqg8uv6/3G1GPM3r6AGYyn/S+4ez0JciIJ87jaIrFNRks6lGv+7oVG5ECzUVKYfCvQV7e4j tT8vcdEJwVcl5voGuqS0T6/xC5Gbgz00a108lHOc+1YfF20T8kEz2KbNR1YQE7Vdn8rAUai2CFsz eQc5j5IhWQqvoPYkMtvlBz+8DYI3spHpDusffrCHceGCpEdZueXwth95FEg8tToTzCw5/MRozOZt PjDpQBeAtKrx9zJWyCwWJrD54qJ5T4npN/GLV9MN+8RIfroYdFhLXlIpAPauGStnf5WMb4u3XZyb Uv1SLzm/Do8VNRNEI/tOiSoW80haQIFn8kEyiV1gkydIBDU1kWB0B6SkcUqcuTk0zXHcsgmwXkVL XKypJLyxN5uGtFVILYouVSyPAyewYGu0Y7stSr08nSBeklqPqGJfv/w1k1YqBJmMddhAgquMawLU MNsS21lpTwGNpQIVu5DNRCdIamdb1XJbkcrkKtQc/L3jhfhNnDXwuFhRXbjHDylDjexx1WVOW64f 0Z+Bny4wKN1Cpj50R27Ag41PzrSmqKc57PAR/ECyyf1AcBIz4BNiVbOLDT++eNaOsIHhS7R83Qy/ A/pZo2tBvsiSIMh82Kc8NsHq2vo/5+UdbVqortH5CxDe0UeNn3xAG1uhkh5RglC0So2MF+f1wxnM o6ZrFOronPURpG+gKHdVUG25sqDZAaQU3egn7DlPbEPzDxFqq9JDRrtqqvxMuesdS/es1irfCmvl 3hLql3xYZwyKxiqZHF12vXvEHZq2WQGieImuGLnO4HpwXfONve6JUy+QYlTYH6euXhqQRcq75PJF w6LQFBABVDCkFDZgN5+XpDc1wqUQ3H+6hCJgPnDNGHAxvA9w6E2mOUyskkmKibvF2nurtSOgvbiY SGH9bgcQq6dgutANehwefz8xYl0fMr3gEZ4I6yTk3RcmbLc7iK9k9M7nX7QdPwoZQ9dsIvE+3XXt SFaHQzdMMFbYyJCKfeWsBrFqLMqU/D0QrZGzy2sv4QYA3Mv0lj3fWlz+NTXwYHNeZGvM0JYGG943 f8f2dqpIIT0L5hLgGX6y3zHeBwzc93Yc534/mW674ggkWCHMeTRd95vvBVFeU97lSWxF/jG3ymUi +JwxiX8e7mGNe+dRFfI1WbpEQKvC/376LpTeXh8AC0BAzG1OVWn6/cftGW00VF1IivllvKlEMlL+ OEz8agocTAInVVPcWhn3faOWsZCZ0bBNVuMCV4ewIV3Or5CMj1gnzafsVU1CmMW/vwykmmwk4MJt PQO4FWvTS3ZLIYWMuIWZvtBCvUzfmN7PDeCNX3lC0NbSY09c8rUQedVDPMG70OK+or1u6qp6x8XR oHRri+AwiqSX/lkuXz1p0TRt0OHtfdMJRiuZSHfFKj+upfLkQvabJ2Fs63mEtu9YpRMTS9EOgQrs leXRW0RYHCmeVgkrPfLidQAauas9vF67gjCfBFOCKOtstxmKBh91PMBBBf3nYHTD9wMpKEMOxheo g1Ky0XsHTDPFAZb6LouaYlpYfg1SOW57SWdeDfUkllTUxj/y9QNKzKLUNUAVE705yEGCzx3bZrSR 43hfKWB8d4ZTX+k822UIXmiLlqKfPJmfd5WtMD4lwPPvUsTDQUWr8o+LD2nG1OL/ef8MHIBrkrA5 U/9AUhy6fsg0jDOBbmM11wS+L/i+NxMeEV6WlDteczQvlAm8uObpGAbVMUsEJjZaeidD1W/giVwz 1J/InS7S5vke+UtkK6Og/eVCma/mNiFgg/Kys+veRHNZyZ4AOcjVVj6eIWj06HF1thU6vDkPzp1g MlL+xNEOMlwhF+mhsxTWqziy1TFRIeq8GLoM1muzgUHNOEYL+qqlLXqlnndB7OSK1gXqInCU0Mx5 saIfL8lkweBsv+trehMf4ARt3bGX7wLhFSCF8qsI2iugTek/oQljmj79PQOQcaMK33xU0fbI3Bgk LKpaPJJfbJlA/IRC3vQAnqrropY3cpBa9rqoc3La6fYepdlBucdd0GRRlfLpidQoawTOCROnKWpc M+zfbg2c2kuw63uj/2vwDej4CC5wXU6yHMrWAnAe9IFIhdZ06abwWgrkx5dIGv5HfHqASyXnRYlg R/fXBoNyTViCRpZeOr+QKbDcSmDx5KFddOUg5R/sbWUS4mbGQHXQgpS2KHD5lePfqanAYTsCRecV dEAHm7j9TpEvPnccCq8Z6+I9Qf7MJL+fYIbHCbV47gXNqOML2o9jqwkudv+ukQBH7N4n1WMKwXre DzCABTqid4BNQYdiR16poHVSjH/pZYStV8KD1EWjXjwJOSsx8byoljUVhUUoU7zCxudGTUEIdIvw WvSomjxQbR4c0M185wQSd4TAU6R25aC1NjkjUQwImPO7Wws1LumyyubHfFGXBwkYtDLtzkYL+c36 rQAWsX/Fa4vPNsIqTAuGOosTjctmbXmkCsitvToOq6sohyG7DMf2Tn/GrutXBy09x1+enHxX3URy vB1vZcUk0gdImPlbtWFo9iVLj8lpCgKsDSKceo4tjssyrX0T/WPv0jhNvz45yEYxSOfYmI5hBF91 kqWzgGvKlLMT18sUluKGVOn8rTj1Sp1rrinaoJ7n0Ocjpc3T7qrOaFFAy9PISAnceo7v/BdKqf00 QzEF7aoY06zD0FQQKPO25sRAt8ttF51xYRvrnXR8cSx+hv0Q+RTJX580xW2+NzBm3FEzaqDAZ6Dd 6Qa9tKEFyzvFz9I/yXw3ne1CFkJbpiXnC2XVnrSQ6G8wEQRn6S7TaVhZ77/d1bZvyoK8XyxksB+p 5ku4DYfErFumaYQiakQORM0zRcmU4AtsC6dRThrPJ2VRcY4qZSfhlVUZxnfVQk4zxDsV4HlWARxr gEiy0jQCk+PnprjvpeOJ0TWyTMWk3JYhbP8FkkxhrL4ohtIT5TSocaqUBHgzMqpbbqq06BObKMeW mEc9YS8Rk2pWm84ycmUj+iTwsS6EDI9oAbF2B3HLfF94OMksY8dsgftCPX/iB7DtX+fuihxrMDRE 4LkDSuJibf7EQDb55PbhaNL7J+jLUuSagY6e5/Pa6qOp7tQX1hwi2c0mi9rA0L1P8CjGlXOVuRWE dXJvgX7kzZDjiYdTFRehChjYtTSrc7u9I4Vb1H7BjHxHWuAGe8u6jx9CkpejKlXiZLUriVm+H0w+ VDYuRFur1aAjJ4hZJ+nyvW6xMeDe2mkrGT1L0cfsy57ecCFu+hTU4fFlgSSgj1OnU4wVUF31Yh4C 7hvrgKQqDsuBOj2AP4jznAGhvAXZRR3U8RJ6ypjEpXQ0lgFa90ZyYoaWWfzIZPY2Tvc9x/eBT8+o XN5qBNU7crJ5NWqc7YEKBG9gfkDFsSmfH2EIcI/o3OtsuGu+eqjYyC8AAZUDtovEsVUUMXjy6HL1 G3MUplYBIl1ztI/rJuW2gkkC1n9AmsUYTyRcjpqLCyQMOOT2Sr2FaBRj98i0puTQS3iiwMWOs/7P NRgvTcRuTBtTAXScSjIsIZWWivJzvjGam1t5EBrxcveRnIhepYjGc+c45sqMPwrHWOK4Sm7SxciC oT45IDrVuPvox+45MFump9x+2Hx+dsCji+qQYBOcT5lQWpGv1qKO/dkkVtEhsOuu9Whm9YIEViZw BdsQjR5z+7oYcox+JnJ4nmSe76iMeZJfmtkPfjL8g17hNWzwlo6NITp83YG98oH+6m1yUJVzgmiS G2Y4eYJKHFM/BwHKspgU4PXj93WFmhSvhplfn+2tU0fhQKwoKYjo8uU99A/fFmLM4gMLzzBkj7Md yNS/aqkMrR//nAdZEQizMQg6Bhc+XIJTd/UFCTy6w3wl/xQU5mjClIXvagQzcSiK/XlQIr1uHgG1 F5uVIOie5Kbv27KPX2StvOJCXAnh4diOSsk9R68Rp9Z0N8hA7YE3YpCOt3PtUJTzZKFEtfZiDxAV JZ4vkYcUjGGbJFoNQGmtvQwHkNDFavXx+e8QJrihVTymqiQpKJ0JggfXxDLUSbMnzN9vzhKv+xw/ 42xJkyjanlWbV1OvWmFnvdhuTek2YcRxphZRacW7fO/3xUQ4AJ2yALZ0qrl3Wn8iK9jsdGuQ1Wck RlokSnbf36vFXEO2hNdcS5u/rWDd4mi1XosCRpca/naT/rTquH1+CuoCRRcnuwTCj7LCmsLk31PW sYuqcxBTI5gAJhwYGJ7l0unkTBOhOv1g5EKsfKmyUizi8X/BcwVahHJHs5eiaXus6cFIy9ZLpVIN Yt2QOtjJTbmws7hJNthqny9n7y3j6lZTHJEuA9itQfzz+tKrnE8XevpSL7VE37Tspy8Rz0lR+vw1 kIxyhFSKwcycXiwXMlTfyx0T1ylFp+RscjkpHDmTM50tdaB659nhAM3xONic1Dgp/DeBueZoXcZ7 g7vlNqgfY4C6oaTHxtW7b8WoDsiQm8p5AtyBVZNUtNIBeUddWkQ6BUo77TzwfUFu1CsJzEQLsJYk RWX6GgR0hFtDzZrm8Cgiq2ja7H5HTs7HCk2xQeUqADvg/T1ZmOaDc/ZA3xMGqnh+yyzhIwRT7hOZ YHky05yQRfzK64xsRRXLw2M/hKANGn9ddiPv5QtJjcGDChOYwnlWArSbZPCUBH4VruHUzFbnCyEw eR8ebuFUC/J3rms/mblv4jvziah8c5zzXGgPy8Y+KxrUN605ctsfx95WAx3yDCiwqd7Klp5G8kkl rzPHuCRMCpzzwZExsJcX0C12V1ILrmwzLR+u4BQgv9N/LCt6CdvXt/3R9MB0nBpFSsvpcy53704q tHgvncTI6cJF2L8Cskqn+kyLHgxK1wMrA2vOCaDMxy6wbwHR3GgdGR5D3c3R2c9Nnuyr14UF5u0R cx8ybMMnzVIY+1rC3efHCQZld4cxE2fnlG+ip2lvi+HNhJtokNGmIl49B4FwAKTdMXznJKf0GKDA E0WCN2DjdwGA3265bdT8MnW2QM0k15ZyFGc4NunJ+33t13JiZzQWMd1bNSd1zgJ0BVDpTr+XDMLu ypXNiOmXEv9A83Npbvfqfr78Cm5UCxgPtrAThjc6P+x78GkC3gxLQ2TH108wWsQVYNDrxnPcJCKV F7wDs22Q4LXTB5JtmcQ3m29gZrQsgFCQz7F6h1PqbbuHnb7YmLEthRJPDfvPXLr8/SZ6ipUMFoHE FXFugUrzQxQ5QmHPBsciMShmTpcG3AbOy0Umf6i0kqOtkQWx/UsopTvL0K2zIoUWDLAPQGuJ++k+ 4Km637bOUK276Zxv03hIai+WYjsN4Q4yyk/8rbf7aV3XckzRNMyJcMDVzJ7YvpF6OtaXMKzeVOKm Igx/lRW3f0MKPGDktbI9sVlqqCZmeu8WvvuNc7s3fTdBPQnGMNqO5DkRJhZYYuk0GzG0pv/vBaj/ asy1uvXXp18T5TVEIxqERNdVD1eKemfl9S4oaNkJrDNtNHCKcNfd+mXeO4toHnQDBA9xI8nFzoZQ iG588bcXrf7zpagFjdG7rnry2nAYnjiq63s2/ejZ79HdlGxatTUIWTa2ikVzUfWuXUVXgWvc1LIH G+IASoRABJ4uunYZWM8c2hrjR0wrtdhKEHSXJTsb3fmKLuFTr4WO5Gd+avUy/z7a8kmUest8AYU7 2n+GFt0llGIaEj29+ZJFC9zsu4KY/jX0z65SbIX/3riqdRvwLnXo7Hcj+0L45rR0LKGErd0Q+kJ7 fD0VceRRNorW6jVOp68MUXU6MoXDKjGaea3OKm9gMTBnvK9Kqu9TS8lxH0iexyH2f2Fu07RfY6Jf vOfViQIASVUyo3EBF8B+VyEfnKDaFm0TEZsQjiOodn+QA2EedmEoqGKHijKlasx6UPog+TfVuFm9 ZaDBAjIYNCfkPjI6VEyb3gcK9Hmz5Z9Sl2Xbmw040UZe8+rrtkgTh0EbUwBPzE7Z3raj4Qy8NpD+ hdkDqpW/Cq8JKgqrF2BZpxWhkqvEtYA8fw0wpMLxMs5bNkDwJlqD+xWfuCuPLxn0szDAqeB/F8Cj /YB14fQCR8ocIkkak32SM3m7OaXszckyOCkgg0qP1UN8noaLqWnyyZuHS9b7RUG2OHxsEXCzQUy9 1GK7fZISQxchUfVujuas5YyKieYsvImRaGI+dupipplR3SeV9paHaYJz4BVcgCiSw5Ajrq4/O/M8 kw4o4kXcN5fqPNag1aW7BoCOjKePPt6ggCm469lLCiC/wy7WayIgFdpvdu4FHJlMktXY+uUcqI3a jW9YtCTbKNtCssm74kbt0IFMGOovdoNRNJ7BFiVK9zcEVwN9xbtSdfbwLnJUDX1gNQ5UK+enyTRk qNyHiyeb5QV5EsMnWMHOVUecuz18hWtGxnbeYXQWnflq5Ci2a4XJt9HXoBkIggoSuiPeB13W/6VX 4EYve34nbXktxP9AdvGs2YzTwornEqY5hVJMeQNn4rzH2wLpcsM6SVuqrWyO9KNVye2jSf/Vj7Je nlGP/e0h3WRrTufgmbyVEsIjGq+mqTHXB480I8462MndIsEYnUm73TBZF9gUZtP7/cr/Hh1AejzT rx7m4hnw5TpGOPzhTMBNfu3B5WXjGFqsNa/x+j05LyvTEQ+9nSnuPf9dnyyoLVpbRqPYBDPaf7Ah h3eVz/UCeIrO/8Smt/udqKhwm2GCjjq24FQU6dqu7l2mtksIzYA8vtxtWv3BjdZkeWNV3Bu2h2mx nIjFJPnom4QIwFACpYd1/JBdUwir1KSOXFnCgSChaus40B+rkVLXF/remsD6IGl+GUjSL9YfWW3F ZaErRcYh3IjDQEWtbuw3I/uDmnKnsNXI9uCQMTk8dMe3+mvnR0gnqUabRSQsxopJBgP3JpiuC2kQ HsWAOpyQvz/Z2FqAJhuXaQBuQIpApi5tObAl0RdN5oN1P85H34/bsDsI4uxxfGg5DMB5+GrBJOrR rYL+jmKVduAycHNkEhnFKOCHXfaiac6A2oqzdtInu+nbXIp5tsfVsDDgv3mChIEWuYXKmiavkC/6 9SUTwFTiiJcEC8qFyxqRsPk6XSAlugg0VtDnyedXp0x4qeCx3wIvHIOimfsmypyzBmTj3vlAQTrf OypDWYeYeEgllLCThgj6dQh68i92Nn2VWHgOgOpRXfGkBZWzn746DF5B1zpUMrJpnD7CMoiKEpU4 zlm+UJ5Pgj5nVhlFAIMDBP8FLoqKtOfqIKDVdnZ4vaHv1c8ZamlgvcbzAAFJc/95vobYJ98x56sr a+aziB6iRGdEh/mypKa+W9k1VHDinU7jFrPWrw0h1nJ8fQaNJwQSSF/eYl6lOGeTXZX4VrhMOP6M +yDYcR1VCWJhg3UezZJlwG6dFuZAH4ZEtAQ+SbQKNE2xA8ylWV9Jf81tDwPUq26IoNSz93p7Zhdj VuHJCJWhEHPVmRc0lVZaTWlSw3WbUFfAWMj28e8AoF5GKze5SGPtisbEZO3OW5Z4Yrhi5E2ygNJj lKEanwL6yS1Y9VgD8K8MxBVlW1f0l4VgI3iodSZadEYSZzJD97p6sVPA9DziOftfnQ1fPhWJ7VPY 5qiUhHPbLmO7tI066PsXD3fkoJeCNTfO1E5cn+ZSwzOKZapxamajmdr/Ams0r+I/JJit4nA8keGI 9E95qAkZ3KZSW131DCZ8VN5drL42k73b/l1mueTt3ZHrkUWiHdmm3RTiauRArVSOBC/Quc6WPO5A jj94FkXpmodjaka1gsXsQU2qoZ5atT4AcIsktIPY82pXGMAluaPO+h+KajvNFLNyVj/gYx4s92D7 mkKTLKMhCpz0NVrTiNmf8W6Z3HbbxlC5a17vYNK2NTeTFIczyIlwwSFr0V79t4H5rsIe1ImTrsN6 HbSM66756QukA424OeQA2rgbjCJ/y1CdB2BtzLnWaZttOCdeeQKR55cfOv7kUq7klWtGT4dktwZ1 fnRdoJnT64bP+u/EUy/pflXKXl7pvGYAPVZTNlD/S8Xy/bpQGlDzm7aVG4tSm1o3zd9lPuVg/GHg Q+88Lf3YZUuf/gIygPfBFI7znMbFFg7mamYLRHHyWfcUNi1QKID3F/nTiJawiLYpLYd1f6kFx6PR p9DlK3eIycWiIqVLIQPwzdShLMrLuFu9hPNy4RsQ5DozESy2jHCiLr2IRmGrYMlaoRnOUqVDYdQd TMcRIOD5kUWjNePjGwLg4glAdh0xiQRLOhFIwFKz8FHI6yEHDLUUNJEBqvo+J1awCk2BNRjT5l5n hHhjSPTyvy7KgGx0xVS2d2GkUnJ+9O3658TTOIRrHoi6vPXQGJQrMkCl9bwN/9eNFh2+sPVtbRlG yMOwUL6f0nhOE1vyxYS5lBwJn4mcEZmDQypWi53CKo8CETJmgM0qmLTxHyW4iYlhCMZpO75Ge8VL AoD41YFk9q8mdUKaunnHBZGLAM4YEEcGKdfM4Hp9CmTqviwkv14kv1gJQbn+3EVznQAwHxjSEfoj Oy1LID0PxYFS3Qb4olpkVV/VpSbMURNG3EBZmuntOTLJNc24QUEuBqQeU+SCUMlw1ov4/9AIO+bR 5KbozMRsJgzne/QdeAd9omkvBeKjFWQiJwPR7AZnl7+Ck3HRRyBGPAkEJbNY3RskcL0j3TzYXLOx UwOGSC2FuVy8mQr1eyjH/dmUPJUwAvTIBk2TorqCfilyev5fWWXXCGdJQSY4WnpH6pRVEKS09Z7z PusSldFG4oaSmuqcLeXB1MgOENbqgdfT/h1TogLDRKBi/9Y/VPs9LHthc8e9ePCrU0cSHiXQECsa zAMAwbt3Z97xLFGhS+D5UVclMrdlTx4xDsrUrkSPIRf6cwB/Mxm8moLkuKQxxMASzEOX2cEwSk8B xqYD6vvJ2lPPnaJRJWzNk/RjWs0uNWURsL2AJyT90CeLGGQMTnZPyIgC10RXsWorbQbkHQzexlnh /Mi+Sp2QXgR5qRLL+IaO/4Ni1HSS+tEyy+ngfpJUyqGy7zpeZGy+jlaYN2vaXoAs7S7ukdp/RSXg b0L0WRn6FOXWi5Q2P1mfAPA7DhBJS4bxLHSaT4YWpGAdf6IbxCsioahFD6elUCnm8bV+cWRx14iM UytH0xrHDz1WU6WNqQbnFrXECMxhHJssfycMTjccsTiIQFpsp2czGzZiUCkSaptWi4Mr5DaQs0qF c45N3/v5Gb4MEsIRCVxRJETrW09i1GXgrX1u0s+wCXcRKip9YiZvOlge5wFYG1JwnpZfPHiKjn4H XFGquvQ7Bfp975alm70NClzPdJyqoG3L9KzjhniCKQtvYOD7qawlxPCOVO04bwhpuS+t8GNjvX5W a0c2s3NnERir347nfvxucPxed1ofPh+ciITjm+klaNM/0yBzbC44e+E/ajqk+JF5tdlzOlvd8IQX DRrxvfbSK6hNtNZZVmDtTFOVPjdHJ4gtAm+/Fsn/w9NyQIgRsspCd3aWf06MAzjnfRtZ3IFK5uMe GJlyVGPBItEoxckXwzkE8aDi5mFM7A30aVVR8cY7QIse1GgUyBxIQSKAozzqKnuRJ47vaj30lp31 5JJYT5u/kwIN5BGfz8G0DfJf3uYmf5lvMMWiR4ojDFAJpGKZgjSPtr1V9yhm0x2jjmvaXfBVCHnn 3uL4mmo1rTIA2JMt0x9STbzWBZ71+kn+BuQ2mbnS4IQj0zYPTRNlca/ksgFDc8P2CL5xfPajd7oi +aEpzaG9s2esx+lIt9822gB9hGrb7pj6Wm921iY5jVY3yq6zy6YAcb3oGT3VjSHsl6+Qux+FtWXw L7RF8ZwEUqcFDfQqDNVZX5dxR6+VcORg6+4NmQPjBH3D/SlrgHI3Zmt7GsDmta8+6bEw3FR3ENwh rtUOStrDgJqpR9TEzJ1o4XAPjyKsQ9E5Kdi1WdcQu4vvMonKiM9Pcpr+729YWVNeSVYhJ8D+m5K4 K2fpsCi0mcIQVu+XDZxu5MEzxUMjgC9Bkkr1ULetSkOvNRzQIDPcNMS+RREG2RtM9lRzMKFTdDpR nB82z4rvlJERm6vXtykdYFstNz57lF/NmNRwQTDtx1MJ837aohChGG42TRGz+414lQ4ZB7la7vjC KbJCz6zX2KswJ3XRT9v4/U36sARkF1n/Bvdl6HBxacBUt127lHFgZjlO4WdUu/HJfVzmYxHz4M1y ZD599UMxE2NezONEknXHfs6GP2Esa6vVHVG5r+poJS3gCCdigitgs0+5ar+Os94muwQmTTPxtN1j 1tCpGG+Fz28/GqQTEFaDmjUwXQMpGG6wsjkxuSY7FKWlAWRXJmGyb/WK45k8ARPxkF0fHKOibmWm c8iogAxLW1JcU+0AmGU7m1LGMPPt72wgl5wNsD/zjwqNGMhLMKRrPjTwhhDjLlBw+xKtHl84nOS0 e43pDlOQHW/Vm8sROYPD9JKSAKcTZZbRKVkR4k9Cw7gpd063BNj9HUarZsikesz33lwhh+2W3LTT JFpvpDjD8pVOCcusRe9XNZGBxH2QXa0K4roQezQAFeiGDoTnkWPxXqq/dhIpAvyxeC4zZy/YpDrV xTR+SjsI+0PjS0htNLnsGnyjLCrs/X6T1Klyattd5/pGNnEVqImH4z8SFiWMd7B3NgKhMVpLD3Sk o+e9pnaw22bKwFc1OpgXasTuc5BCgqwzuIxsFRHO5slYYp+2tTSM+p0dVjjhw51uDuuk1yEm/zA7 V1S3Pz6gdnigsGu8BBJ5utAx9gysZ6qof0sI8XyYBBRpAZtaNk0SZRpAQNudvEd/qgA/EBm6bpHA yuTDsHG7SQxytlOfoptosxtcbSZna7Sp43ypCO/cpaV7vIY5dtQES0ZjDD8pftssFD/7Ki+AVuDG LEWomZjY+QVuKVzq9zERoTk6XTH+iYP9Q5ZoQX4hSFmnrZlzZ/SG5t+CtwiAXvgy58aO3vKdncxx 7AbDT8FJfiXnYVSZ/aTOhllyEhGYMfAF9FKIs90Qovzig65MYcwRRwUHekrQxadZ+nnFGMFDca0r msUtuU+dCXR/sC4EJmgoatSytdFG1m1bgn5RuHrWnKKCUnbABbNbGlUv5u1FU6Z10Dr27Bj/LsgI fprzOHDJytwcevefVK+2S/jrBBaXj4jZ7p6lBeKP48mx1UBgZTZsy+sNFS5L61y8yq0tc138BFC9 DHy/JVNR0UlimAaLEN5dLgfLI5uDDJpLiOlB5Fx8wY+Uefn801/pVHdT1XV42FfTAX4AlYnf3by8 rFwHVyMR/O8ziUEIDheo4+0GMQ5R16dU5N1Zx0DwANStK4RFARI3SniIp1Bpx7pnYryquNKDIClc KTF6ksX9K/MQ00iKbFKFkaqQLPTwpDmue4lfLthDg6F8BgOS8T1lE6wxW0luw4d0esWbanO8YnHs dwCIeDsyex7bkCrpFQRVf8X2XEtV+coyuLIV3XkMXxK3Ep3E8NFxgPG/rggD8yTyAbGwrXwDCOOU 0QMXQDYmqctF1HLqQUfWY66wMCvEGu/pJKTblDYyV6XxtXzGRQcvvHoBJ1C8b3hfAT1tP6sROB0A BlXWO5aTx5KBXQtY/MKA6nhLM2N1NkqzaNDCPtVnEhhdGS+eSbZSnAWxt//2Bqxv4AiSH0uaNM2S aHbRaAL7MyOpcQW3iVcg01d/37O5QZjrfsAOdgbwR5aXElaozxRk82EpfMQjSiQ5jpqktAlaH80V GGsG+b5mN7O7xDdg33h7+2RQ8wc72V4w+iAOKJTj8Bk1KjKE8C1HFRA8nvCaiuX6EeC2y0d1Ii0l t1oIMzfl5BmbzlFxIk9p0fqgzLL1E5PqQyeyh2mH++Z55HY/5/ycKEbnFECAxXRpK6pWjYI98ivz 04C/ZjeZZ3C82AB5Y/rJVdHLcrsWt9OZ2BCU1Dajzu1b1pn3tSAyIWsi8TL0KAIWSELFytgmIsoX R1m/6YugyRBQ7weJCZNMaFr6MZGmWyOrH7cViiMsLd9uA1rD4ti71f3K39jT6kX3k6snoIaWZm6M KdYkYcr8EOCKcIAXg8oI+Ka/h5xTd/qZGLKeaqoo7yu2TT3LJpNBpWNUNVCKs0nI6L02UuS5UDOl 9ICGIH00K1pqmQxCzCs1nLYHF0sNbLE6JahvrvCnNXosRWc+q3KSGDr9Pud1LiGyNfT10KmQ62Bo 63HaYoCCdpTgj89VpBBtsPoD/OVHFmSHu4pXabKvakLtloyAQduKXGZjK7b4eLpFcPxsufNG7Vg3 2g8BRnOXScgvHcPdy0m/NqX/vTw9tTHEckt72T+OOsUV9X9oCyYul7ipApIfEDIV4ro6VjBTDyUe sdUKQagagf27+I3wJeYxcRwzk0hWxoCOYYwcfRLu/n0hwXJJ+oPdSNx1aAIlz0rocwu6ZNr9PnHA /ZJLRxzGB+0sYR4OlvjVv/dfD73suOxgKOl1T87IjZzjjylFxO7GoNsUTla7esgHNMNDZsxOFJhV /0o/574d2CjMjmHO0hw0MZSYfEYjN34T32kbJlPnonBNS1EwqnJKD46oVxxWPU8UGXd7I/0/w4bt rYTfXirv+ucHsCTpnd+GclMjoSIB+ijDpGOW/tihr9cF1DzBPK4K0t65zEbBurXOeiA2rpmY1oz7 41XRvfA7x2f6PDifQVfvU6/1QlTJgJ3DFqCfQVCHlEukgxtIoqVVISd88TzS2dVcT6/3SFqMQOKD 0oiuFR84gGHyVwKe5eE7luObXdymIOVuRQ72JQGWWBZoTFrRb4BtFkFxH8zHPKwseglO/xIGhNsF qKgtuthdRKdF+/FM//qBfET+Z0wgJ4cUD/dXCpjdE6acVj5M25PPPg8wAMXopcsrYScfK6k67AwC sZim4RiWsCWXshX3kqg41IjZagBKyNH1fxOtpxMKAmFJa0RqYMsiFPXMeMyxxRUn4VHQHYj4quOs kXSY+jd45XvyRNH1H7qEz7Cmiy8T9bw3YOBF1IdmJpmU3uzR+K+fAXm22Vcc77qMM1gW8rW0Tcla cRHAr/f3ZzxtKps2jJG9Dfvi4xB9ZaE/rytrMa+XLyEJ97+P5Hb4+dvJR4RvQ/cfRhJfHMRPZYR1 iiHfYuexoMVuugc04zXm31Xs/AlZ2kCXrFUn2eMebtnLymLkabRLiny3jjAI+LHUf3qf1a4zBVem KRo5YuF9HbbcrLgJKgZYesQHdgpLJUtdXACaAurqzRMeELPPLI3kvjmltLe7c7Hn24Fa/VjnmfpJ MQH7kH5B3npDTl6r7ODTGIDGF+ufn7UfGJy8M4ygAxVBtTd7di0zKz5IMJtWOJTG7o66pbDkZNr4 +TZm0RVUTrKCBtFQ0NbgGi9hEnMYl70hLVpx3wlxn89tse3MtVQEX3AmAJBNyAxYBXpxeGGB/BL2 UAnfUhzFEq8yEcqap2b7IEXlsZyWj4j9PreyhFv5BsQlnEHtnGnWaH4YB+ZwLZ+QxZHgebcz3pIy SxjbLjpBJsi8ZJZzlxohtnN6blaaru4+I/5EALQkKIP+1ZeS/8zBhHPbzwMOL5drVBU/YP1D5HcU TWEHJ45yiuJXYsYD9AkIYgsuRLG3A941arDlr1D0zCQsq9EU1UHQC1Xj4lJ5r3M0hdi8ZOyqRgmt gyla3PhjmmH4t2gRhQ1xZq6zsvz5P1EMkrCa+bAo2sh+u99dVm2b15nO/L2NJZ5IfQObZJdN2FVt UD6ZhQbu1TlFTgqTPiLov/D41oD8RvHkJz5/+U/il0u7WFdz7eELahOgfFP0Mwx3lM1yPYECqvRt L5+fwpcNhNPYs3VvqPP/dI1hA8XDWo6zzBn4yW9xt7qZinB/OnismTCOCsBeUGJ5AJS9Y14V5/YK g6tU9nrkmlrpYCGpBO/v48PenOhABmhL/jtF+NMhF1fMjcmqY8+Z9BRWu9iLEk2vz9q668uSTQNc 09ueGBX9g5/BVX+yMyYGlHEpkgeEgXMpoYE2X8kWKctCc3vjXlhg5Ahdyy42Lu1H2U+vwtV4dxRr mllU5oSothagYe6nFaBG40Ac5BU7gmqrv76UMGppf7Gawt4LPzcZv6ALC+a6G3BKrR/zMVQSwfa+ KbN4kgVY8zQs8MkIT+FmVvxBX3QZZ9nk9vN+Jkvubu20Z8jajzE1Q255Uq9TUlNeXMff/8K01ZTX 8GxeKa6hsfh4ii3zdnsutz9slCzRuFFWDjewBHpeWhKcJTF9S/wOVAznXgOwVHaesTsf5oRnCyqz pDqBYqdGPo8jOQgzYdA6OMQQdhYapyZ0NUuxv5ZfXhFTMv2/tK9WO7mYjklPcolk0oYmZ7/G6jhl s5ocgGHqG2WpREAKB+6T0uTuqUujdqOfItEU6RJxLpnWM0ZBQcfTdycCEfGNh3n9VbC5NuMe66bI MVZMEPZZZpP1eG6p+PYF6kKEg5Gw+io9mozLHLvPTDhUZKlbVjVDMELbTEjurKoXEBLhwRAyTG68 +JXY2XTR8CCfzOcBEeEkGIHejjsjuujaU2NmaRY/k0SXzWQHuZqKCrm8yUnttTiTG5rnJ0aWDGV1 fRn9EUTZcBOJiDDAKnGPOtUudpK4Vedk6RyBXFxxMxk+FSE4E+X40c7Qby6H1JA0jmo+wxWQz3So qvUf1p526niIub/vWtnkhT5SXPUsR2sy4Ps5eLH3r82bzQY/S6UmMet+skwwY+m/JPcYihviA4aO 77TQzvH1i7OkOY3qCJAx/rrBpdpyjSy/m29FJAp6dMlFCgS+3fn1RN4HM07jGNgTl58h8+dU/q49 HvLexgZA+DN7iTOkXA7BN6iUCPwTVi5SZEaY/A5mor1mwKeKEQwxrK49mfJQdxdLC3+SD5R1FPA7 O8rudjzK2SaIUl4tTvrIj5iLVIt2ghQKImpi9EJjm+iybdRr6UXaH2QHeMUVV6U2PTWTHrEl5ddR aOuBl0PEpVO3AdN62QqNn54dr37jjqddnjg1F7gRnLv9P9G3smw5tX9EcyMPShPHLEZqTtn7piJS KuLqvfe2OXKvnbVHQ6bqtdn4FK52ejMrlG7CAsAas7WvBvqwP4BHa0mdbm0GWVfQ0ikr7Hul12LM 9tKgUDdgbUVGYSeLzqUxLWFgrtXHriRtwHyakzcN2qRrRPG7g9dYFv2xWwg0swfAz07Pbb3M+Tpg CF5rD2ohWRXSb2QNmnuAg8sE34oHIGy7B1UGfOdLtUrWLaY5Jp4qOX6qZyK1cBUAFFPZuSqSebr0 DBxb+PvPRmNNDjB8Pi34ge3rWvYcgdG3/CoEbfxFMY0qDGnmJcEHExX+w5IiXaAUWImHXsfyppw+ ranv/0m34nQ01FWSUx+YX+qN1yBY38bLGvIN2Esxn16wO9MATpBqV4uMJcva3pb0tr8/IGMLekxe TfDXt9K9rXaIMo1QL9xsrv64jxiHgqv95pLC5BrPjrqmy1+A3nzrRoTQ79n4cSe4bhUnfd8cOwFR +ZcMvf8PCWRaJ6mbMCQ5sz801mMlKSQYDXpW6uROlH6ZKvFbmNdeeUMBWGr/uzJWOln3CP/MTkGm r+fjiIUgnRM7xC8aArNebE50f4sucq1mtI88NtZcxpy6GzD1fkLvqaiI9lb/ZDDzfpXpAweSFrhY 7wfJ0VdMUmUX0WGmk6ztYeXU1nFAqZrChhn0DBEJ56zm7rm4fuKdu2CUBL6zrLsymnvKMfE15EFz Ph0EZ6vxcheh8m/2m6m9nFq1e9+6OSPhG7R/wqrZdrZD7Br6DFway/PU6Sslv4t8U57jJcuqSqeO sue18xKBn7klEDxoRvDKCeRuFJ5eKHVNccFE0B3a0nM7ZMNa08QM18wSClDfu0vuLea/uc46KPN9 YsAxXhJSEym9XxXz4WEiCBgI6dpCVtl65xdimnocjVWBx8mEA2z287CoR+QLL57Lu5QHxJO9on7O Tr2mVqj4uJ0TTBE3UeRf//Mktbnruv5B4VYhQYJYE10YChb1O2mGhMqX6m5UYuS1Q8grQ1SAJXM6 TtkAjtuPOP0mEioN2HEb5AmVJikP1ihExdYA+3xEI7hD5sakcTF5FgqfsNV+vGQVW589IBOID6/1 xm9mVd3uS39BkeGaD7TrF9+1n4Er/BCyb7pD5IRI8pLAlwI5TZx/7S4GNayeZDxkxrBVXSm0/XtQ +jo62PVAfndjYeCuqsaRMdk4Prrfp/bzRtofplbfZDaw+mGtd15iUg+jMVebSOFQ1VqQWcpDLL9b h6QruYzYlQIDXg9a9CuVYMPWcx1SkrW2h9V6tloSIPpWJKyhbpJ5yjEpTLHBkytYLFs1fj0fEqgO z2Y8HZ8vbqJq+qRjaiIHJhMhEpjaW+YmJll8GXIRV3Ue3ma75D4DZFTlpZY3PI8YteU1PGXoj2+7 qziu9rRiJPb9hqwVTTfaByFsmyJzwUIKCxgJhPa6C6JcQAQ6Lk1r+yjvA4PZO0EhImnlymooV0ZC GAGUeCN6URrt1YFofXK9GBGObRcs3cNg2LvNHKxk1HbNpFGSMaeRWVClpmabklGYOOuEYO8P+U08 1GbxIAbaxspcUU0u378QwK8w5abgDcttSNZoppaD0ki3Wht6X9QHQ4lQTspyz7tpsRKfI50sLFLU OClq+UhIJknnAhneGithMaB9rq4RITUxyYozrE+8Et93MD+PuGV+ivE5VkM+Hql/UPWfWpW8eogZ 27h71h6ID8Fsu1jRXzKgpOKK6Rzcta/6bx9hq/j1m7gI5/8Eaz4vPC+yrpj4SmIN7LhHUkL959sS Kvz3+jQGvbfriAnld+gml71ebYwxk9pQ4Nt89dVZP0FSw/g3cj4pUHWupMfmdGEoaL5C038XUsnp l5FAYIC1cEk5E+4UnZ+Pz7OrlSTYc6OEZlRKzjUVDldYthrjZOdUT5d2wzRYCUrSt6sdxZMLzfN4 8VrATe6gYmuoYWHWmpWkGq5I/f02I816aY1m+RFPXKR2mnYkH9o0rr6ad2F9FK9Pu1IqCyuvhooz 9bmohToAIZwSecVovUUdZU8+i4OjdxfC75qHnZO68ZL9epHmBjIuDdvnWkOqTvwIeESyzxRkoVaG l7Aph159C8H8yO8ANYhhKMZmQ82oHidfYi+8omnJlix6iHOd2p+nQ3HFeClk4dSkqF8pmH0N//ub eulwyPq+GwvlfJvyxqI5TrL8w5RypIWwzWwkNF+VfEWW54lVAA05sM3579mYsrW1ijEh4H7mR8v+ 3VWJsNfRINZIRs8L7ouMnTQ0izhVe8OuTlyvo+Oc90NqC3s7CLw/UYBZCsL4wNjkV/hClsKp/s3s F0gaa6YnTz3pEP3VTRSKM5HQWZwFDgO//cJDYEwLw4T9OnsX8o+FY783hFWULYShuNxNciAhDJys XJ7qnXecrNwOIig0QzPrPQ3X6G81OwHWBGhafPCpq2zj+jtljS+niakEd+PXvfm6TUMdunsEF0H+ T596GHAa2UydHK1vbinY7MVcfizZdxAmJUVAtSj/X1GjyCDhH/ko6WMMFQs1+ekLnMpGLOx/2Jcg 1PNeU687Rs4hHzYrzFX35KOMG7Qht47Eor6mtSvBKDVc4Blv77d3E83gnZ8S6dRrGyEaqKRlLFsj EYJCo4XkV/sJ5yZiEtPQREYlWvbzOYe9xw04MTcdcP3fuHBk6q2iATdlGVAUk6uSHaC3W+gSU1j2 3Z6rdYxAUi3sGbURf7FN8m/wqwMkfAl74p5I7wLpcK2WA8Os2xlCva/WVXxdehPollFnvVVkIcgz RvyvA1pbBM5FNikW2FpDoI8arQCOOiFiidJCXWSGZluJtciwrhQ3Ozqi2Oe3+2n8EiWt/aNICGHQ 8tC6tCkCUyH/rWdziMmbK1iC0RkMZrEK3Ga2nnZfa/Vs5w4fD1jeR7YUWhUWu9ErFun4+2KGizVM rv2yPAjn8vdRgA+GsDSK9HcXXkzyhBC6fXbwkP535cnZrvtp7VL1DZGPrWnVZiaESSqv+QZoKgBA UQ7z/Lz2ArE4T1M3VIZRtWIWwPy+X4LJ5EShsvGXGGZ0ziq+SVTsHn2nKYtG631GKQKxPZgDGaff t9vI4uBhJt2k7V3Fzw1Ff5mTpcScHyHvJyhDHTzpo+J8TYu71cchhvQYkx3utjGA9gzndBkzbqZ3 bpgvK8Cc37YbHMA0KX8Gll1ZW4WyXKnAmTBnsQgGzZ9ktmVFTGx1tzCwAwTfRrXWpDgM0wXPKWri qUDpUGXDVGXIFgfGlAujU+F6/3pgPiSG2oIu49XRnOuA/OmFkMOddeS67kcSrobyLrg4YvfTakhM Vqp/R6Q9deFvm+d+nikFZMkMHGxXnanZ91lNrT7NDtnUE66+vKUVXQA42APfot/aJ+lMwJIHkaD6 fqgP6c5oZgROJ8m8wcp783C7OArVvXBPb+Ppyksuu/7yWJRLUOHQLOdB1jzgjieppSY+H9FOjV5h hgcqgaR+9lP4CCQf3uN3SahdHsOmsIdcb2ljdm+XAHN6cLx7f20krTwART15VokOG62RPX6jJUM7 qdCK9tZx+3STnXAJBZgxL45A1eMlsR/bs4Wn9j2RobHIYwLpt6G8h+RhyIluE5OzHlBpd94Pk9UB 6VOFUfp5ZFuVDsAbekyCsFV+4atjE+QvU/JZO1RFLiGk2L7HFdbQmYoeNk+LD7EazFtXhjIyk7ee 1kmpVwOlSVEQX73O2v24NIxBjhqh0IKBPxaE08ayITB568NxpBeZCQ8Wjxh/5OPeUon7Uoe4gL04 Evy6XP8J+6U9LJ3XT50DZXj6o2dHwq0KLz08h1AL1hlbQi3ML8cOL3tMXerwjUVXwI433gwlXYzb wXNQ13ttIDTJz8NwXgX98RlMWzdmvT/rc8pqL6c2VjxR0mh/VzJOnTjz0MLCVlV1dYjwGNxeYHrN ObXWeSfxha4NX0nxz0eu2npIrOkgi8iSF0CL6sbYgDbL/R9OXIQU2483fsbE17E0+0EUfjpjTG0I AvwHFmj2GQ9bd+xGqcxDUgoY3B9Eb6N3Osjrh777UJ5YtDjXygUTLPhhEI+44Efa6DsBGArz2jf/ IREGF7gQonWph5MlMxjgZQeUWvHhLP0jKGBgxYxy1HFDM7KnMt3Z/gfqP3VgjwnuHyJ3za3kTknm Bf4TLyoKgM3x/3qK9zcgBwiR0mPcmqzjVzKbWI59t+zuV7zGBHaCtFVHrRnJmbFD9BtgXSee+xP7 XK+6O4MQceqFYoAUCFlWw5cm3E9vSDdll4eiljDLLEYH7jckHm1H8UguGx2O2a+prrPSATPo/TJq zaxDfWTwyTr4vCi4dX4A1WclPqJq02tIGTDh3dS5TuS7eUgk/qe5DhDlDQplbmRzdHJlYW0NZW5k b2JqDTU1IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNTg1L0ZpbHRlci9DQ0lUVEZheERl Y29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNTI4L0RlY29kZVBh cm1zPDwvSyAtMS9Db2x1bW5zIDUyOD4+L0hlaWdodCAyNDYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0N CslS/rc3UGPCb/TiwrduXI44qCTlGnrOzGueOcz2Le11RbuaM9SwQTmHx1EIm0FIVBNUlMSvYsvh a5J2r9mQ5z9kZO5QRq5XGiX4iITgQoT7dx6nOvf7dlEAkIk9QKHPqhST60swBxzhhoawj8OdmTXt YvPe47KhVHirqVtnUbjT6ICAASek3eZWDl+6djhqhPUVbeoU3QEx1WAVwhIg8TfHPv3cEm3TUUSs bZBEzlRDU8eY3MhqRqZYJlVmVw2X0LNhHudq+cnN07dFr+X64hZrj9hPuDKuPTuaeNc+0otBZmg5 yg8ZKXNQd2ZksPvApFRNhwLiMwiNl0sal+EOMZMx+fvuY0CpezlUATCDGK9R7dnN80JC7G+7meA+ 2/D1DOhss/m5HkBiPBm1FZYcqhtBJp7h0hmMXfy6ocu03TMTsi5zer3gtuWKrp+sITiUp805wauO ZRQEl+Bb5zhpCRzVelKpI43PCGSwwORtgAIpVtmsR3nhmCoUujLk5208oBqWmLw/azjoKSxSzvZB 1weic7s8PonX2Zd0MewqK3hGXeHDu0vlUwabIgVZFqCu+tuPJd8CYn8kE4Mmp+VKXBR0RXGPFJeL ujRj1TFbI1/v5wP8JlxPPDWVEiJ9wX+0/F5ROdXVdeF6qZDKEjwpjbwCT+ieKDrBAgHAjMHvnOaK ngGw/D5OfUqRygpNemxmBzj/Q99h58nQCE6BOoWFNy6oWzLZgInI7xrMG1bs8s//ssWhLRyDRGTJ 7F4/yk24zFDT+EHXNTwLBg0KZW5kc3RyZWFtDWVuZG9iag01NiAwIG9iajw8L1N1YnR5cGUvSW1h Z2UvTGVuZ3RoIDI2MC9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1Bl ckNvbXBvbmVudCAxL1dpZHRoIDE5Mi9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAxOTI+Pi9I ZWlnaHQgMjQ2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQrCyHrtmeY7AQO5xKZGztXuumAjmyeVUJWV 69cqeYMTQwt1ektD/Tu8cCyp8zst+NYVw5wbSIdcIupIrRkC0FZOsdz/IpDziL5byfIgeV4mQIHv V8rQ9f3eDVq/s+/cA5ZH8JK+WuYDRortQ/xy6zqxzMpxdVS7/nZUrTFZoi9xeg9LlqS5gtg2eSZO B7Y5GIQ9PCRNk42xT/wzpIJ/ZFOq0mlYJ8Jw6U+h25OA3HUIIgbSfzsYTIt+HKcAKDx/902CzJaC OpYTNHzecmIVlXTDC4KLgTucLZqM+DULbXdG3RUjORZvRWlziQcpBl2ZwWsrSignZuKZ+xeBoR4U l2cayXGyjw0KZW5kc3RyZWFtDWVuZG9iag01NyAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3Ro IDY1My9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVu dCAxL1dpZHRoIDY1Ni9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA2NTY+Pi9IZWlnaHQgMTk4 L1R5cGUvWE9iamVjdD4+c3RyZWFtDQoE0EoKTFw+v/18zYUMJp9XF9tuS0bRbU+TXGE8y9yMd1YX znzy6nxNZwIJ/rhziyGj6Emz+NVb2gx+nrmOA9+bhcmujQ8CTZtpVqXHcXNpztw8G1xqWyyRbXxx C7EkSrt17Xa5tUMq/PVg17sUWaleagJptnVjp+YBqqB4F1/KXv5FaWKbbmMzUMZeNkFJB1uXXGdi deHwgFuZ6WOew+Ly68P/NxvmUibXvWGgNPKXU8VWoAhPD2S3rpynb+2Q6F/SJD5dG3lnorC3sKI1 5yMg1kKq4f9LkN4ZmbmsKSFVnz7FoNeJOwqGaF9oAIICpFYnTMOTZC0i8QyluT5wfLgI0u4ARsCs ano7Vrrzlpa77HLw4HgWTnmZpX2/40fnN7mEDkkqKDO+1GXMNflw85v/M+JzLXfX46beWFaAQLQl ElDC/6WU6BFYporMVObNqBfQrrbsBhW64sen4NLzDImugvbBtsBsXHXPmh2HLZ9LaAJbF/865jrQ yLTl+ytR00AleA5AcF2GAhwnzLQE+ritMdWNEkI69nEXYr2yTJuiUTpZotxHo7Se6iwQB/xnsuQ8 F3nUCMp55eeeYja4TRoAD2O8BFY9ddeO0cXDxU8vXRSjniOqRTidKsdRPj/Yvv7veBZmNWqcq7kx 6l5hS+4Q6bwK0Q8x94C2Pu4Hy3MODaNi2FguUSIyjnQqMtt1GJPJKaogFRk4nA69CvyGaktpzsU2 mHM0COQrFPX452msT7NAxx/wQirOKHsBW4l2XszkvGuIRvuK0aWN9WoWqesrRK/WJazqVDwXuLuQ 0+x2lysdKJDqAL2BaDUwV9EI/v9EveeLp5l3EntvqKlDlzpBFNzBHohDl5sZ85boow0KZW5kc3Ry ZWFtDWVuZG9iag01OCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDEzMi9GaWx0ZXIvQ0NJ VFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDgwL0Rl Y29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDgwPj4vSGVpZ2h0IDE5OC9UeXBlL1hPYmplY3Q+PnN0 cmVhbQ0KqtvaQEpkOslmgApwmJ8UPNI2puk0o4CM0vfZTD/oQEt+UXgzxHC9eLeZSpSHOXXUd7ca 76v+OGpjM1DaSNYjUIvs0Usjbw+DF4V/Cow1xjRdba62KQF+9pG4qH8a1am7DuRCx+rwuW6o83da e9Ttn+aIwm+H1zXdUinE+uLJVBubto2ZDQplbmRzdHJlYW0NZW5kb2JqDTU5IDAgb2JqPDwvU3Vi dHlwZS9JbWFnZS9MZW5ndGggNzM0L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1 ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNjU2L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5z IDY1Nj4+L0hlaWdodCAxOTgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCocMwWHmPr9Lk0jLS4GwHRI5 IH4mcaUFWtJJN/0zeO4E0ebm0HuHYvulJ6YpduBn9rArmG+SOdIr9AUZFra643Ox0ga6yWp47ghO s8gJrh4ZybghLeoHbn+aoRUIBWpOeBEeD7WZrj4cWsLenr9xpHTzqnw4XkTZg3tVzUB3cimw/JA9 VjBdAtxhbDMw3yIV+UYHsgnXedsyKxyAG+WnCYJSPuIZCeWnIH+Ov+jxicZBKVU/1ObMp6EJGkS3 aFdWMbjEwTFPXKe37Gcb1ld58EV+/mus92wGPWka+lcRAgXhXe4bCnNuoqtiKmwj+Aitn8X8TpV/ pjTPEP5Nd29XYdUbzaqxhN90MixS65ACdg5QXcPrRmpLbkF1C9HVz7QbhrkALVhPPLZnYtSbIBtw frkbcK4GbzTxMKHfbbSpoejp3WJal1BVPJ1RB3RQZwlhfWk2ZDQ8m0S9S89VvRzSb99wJXIynq1k HeH3Msl6Kh4Z+uXPGXWdzC+5kjiTdyRghq33gGi+eEZoSvxcKRkO942afmS67f+emIlIjPeuMJ4G rUbcBDt3bQMdMqSouPJmfxbS9+vksMT8IGcUmVw4dioPmu70+ehfPFeIntiX7jnl0iLBwSd4cFi3 M4S04WdgT32i9YzyT7I/jWv6QS8MzvO2KpYGVioHnGVypaETBbIsewc+wgn/zSnVVaV2A4DydHkW j2USDUHywz2wCnYZzdBIakY+49oN0TW/k+wdpdxmUWj/O9CNERMR9TLfImZl80+/aenL9Xljz9tO GwcnkjxaCIetd6GxiL/CF8JvBmfJAI/f4wg3xFEufSOFYgV1Xa0E05Ka2467talCAddAfNd8a1pa pgh2AkIb/RK4aZRNn/EqylNZSjcVKzKXheMUk4aWwPvCDfo01RqX7Q+g95IRID1pGJcqxCCjjuqe 1InC9xXuYcxtnYVZ4jj/x9rYVD2+MIMb5Hah/++OrC2FDQplbmRzdHJlYW0NZW5kb2JqDTYwIDAg b2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMjQ0L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFn ZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggMjI0L0RlY29kZVBhcm1zPDwvSyAt MS9Db2x1bW5zIDIyND4+L0hlaWdodCAxOTgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCov//0Ey6pcy x/ZOXbBTTCjxPdrZlYy3t0ptf6xD3OJiIfayMGc7tiU6ItbuFnVEF8atVa+FNmZ0LkhZ1bQlzArh sd7QX75Oo/+IDA2d8J2/UaeeyXYzTMtObcf8AhNVvWohzOzyI7mhro7bkWQlRSPpB2+RjGEv6+T8 rfCQAkukzQVaO0RbXeh0jS2bPjnChRSfbI3rEERHPne+S9JTkHibscgPlK92jM2pvkiiUfMmnRRU dTnjIFrraAs3sIfJ6pYZ5nzNQZMcjC6iTSYotJHUrf1yPv4HkRSMe+cAl/YTrubxQbULYgvwH+Ua qdOdE3t/NDINCmVuZHN0cmVhbQ1lbmRvYmoNNjEgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0 aCA2OTcvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25l bnQgMS9XaWR0aCA1MjgvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNTI4Pj4vSGVpZ2h0IDI0 Ni9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0Kl99u+lsqW922ug7bHFzRlDTHBuidi1xLIvSVEaHP2phU QreAvavSrPRulCXs28AIEkjZtvNvlIlYNBDzmOn0IpfN1dCXmtqLGnZArW1zECp/GZD81OC4h/kP 1SjlLQy/0mBUYhbD97RMJES4RP+xabCHXT4GbcIhCw3NZn2yJg8WbVYGANBjQJNs1JpIbPLUsJOK FKeVouBZKOUfMg+8tUOA3xkG8fxLslvaAWkuu13FCHbvH+TPwO40wS6ILu+BbJtv/I7tZ1qyomta yhEPFQoXAe2sGJle6wxxDtYsmthzLzBAiDPwPLJAsd52GOG4Is3v+Z3v4ZUtote6qrc0Rax56mcI nm1qMbos+fNCN+5QGnNRkjTB9Yzpkcv45cMWzOhFKrXBLTp9PJkaf/8RF+lkgZxqPBaWW/0ZbHTF z6jqU/iRLQxHWM0Qj/9mp4Kw1d9KLFgE8vegK6xG8LXRwT7hNBrEq3T57Ta/a9z1Den6uTGDBSDk bTfQ4esRZ8pNFDtrI8SczJWbp7aAWHF8JG0Jf+cZVRTQLRtYbYokordDHMZ/2qZtX53KwaGvNp03 zsuLRTfC02FWcH/JMlGPQMXYzAJsh9xjIy3SaGFgDEIlRUbFSYiTyp8FeHVEka4YGiqofVDCephx fvJPnH0TCf2Mm6GzkClR9wdtefEJN0xU0rkfr6t8TQ/p8jxQWKUPepAMEO8kW44Vz94wkMOnk8lZ nwXEf/nD3V4E0auxxxLLrQU10kH6EUeuMqt6Tw33Vhzn7VH8DpOQF984nf6qfdwGRiVFB6QdT326 facZ1Hs5e/6pz+FGVCE8V6CuyG/D91wvcznEY/4KD9a0MtD5bSw5XO91XIWZib1WnZg4Ah0zO+Hc +9zNWmAlpk4GVe/O8r5lkjWtf5QwddSYuA+F4NU7aw5717TmtQ0KZW5kc3RyZWFtDWVuZG9iag02 MiAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDE0MS9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUv SW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDgwL0RlY29kZVBhcm1zPDwv SyAtMS9Db2x1bW5zIDgwPj4vSGVpZ2h0IDI0Ni9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KGyHsgE7N SUAoRGKCNnDr/lIRaDEg4GV4niYz3P8Ig0HzKZs9jcjCNU3s1Yk5dLwx9sBLHe8pfZ1qNOCTcfQL /6pZQWQiZ9xRvhxc7iDiSYP7E8DMN5YjV+W1uNcflLlhhhM5ioJQZNI/bEk7SK3vWImzJLRBcTcx 1A96HQynXceCfNdOPkYeEtWX0ixYDQplbmRzdHJlYW0NZW5kb2JqDTYzIDAgb2JqPDwvU3VidHlw ZS9JbWFnZS9MZW5ndGggNjM5L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9C aXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNjgwL0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDY4 MD4+L0hlaWdodCAxOTIvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NClPAz5w7Cy1LbLM++kZdm2pwxUjY AomQx5EG0pQklGe/pB89BY7v98CltYAHfsk4lR6qrmgrNveoFMbnrYkwkzjv95FDS9+kiO3Dy5zG CbY9syLUOwZAhhpVyXb3mplpvqkqELVDUbYiu1INj7bCabEmXk37/y/AkptFi1J2ezN250Nsf1uD m0dvRkm9HbbFoiFNOGJOkkcFcGno2RUHR/pl2khPT+UmjoR0LGY14KJxel3V3HO1qCmL1jR5OVuh LXt5YmH+Le+1QxK73cKi2IPnc5OeAAUyAR0dqsrMntwiPmVOn728Q8YxKEyK8xXl650goM2uOwNC 0wKCl0QNztOv8jM0CIz+lxbYlzXwjP5iAUXMrdxlqtohW8ibNcXfQPe+xriWq6dAyD1eqiKz9LGF LXDnT6AnP+S5SyHR5Z0Sx6GlLrL+50k8h0mQ0baTL3bt04dsWcRvjfxpkm+HcGlIfqzR1rsQeaRy z1Zr3k0Jts7HP8sKc9HtQ7XejaKR/Z1/GmedjVLq6bhEXvzIVM6Ikmf7mkZ0gWXGEBUZC2CR594y MVxMWianB15Fy0QegeyF9nL/4nC+gFUOjaY+GpgLdVmiKFGHhYMuxJW9PrQb5UqajV5tQUn7I7Qa 1JYmY+Pc4K3Y+HarqJf1z3Dq2piiM5kmOkGtWbVk3yHiSDS0eJzTEjIOqGxNus6YxTwnRIqD+kwR O890y9ZYs/RAi+XCQ+iI0XB4safA6AMrO5gaNvuwvn6Sb5N5mw9KtDrbFc8pxP9B7tQC6dnCP3mk vKBNGOm61ew92Wvlh2CMyude+IUWBSUTUWUGVAwIT39mYR3TVAi2xZtXOGuEZnGnjw0KZW5kc3Ry ZWFtDWVuZG9iag02NCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDIzMi9GaWx0ZXIvQ0NJ VFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDIxNi9E ZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAyMTY+Pi9IZWlnaHQgMTkyL1R5cGUvWE9iamVjdD4+ c3RyZWFtDQppACiPWbdwdmUUd3WnzjMVnUqS0GTVjFLLlFCRXlUknpLFxt3NBnDnGa0l6pJ0dkHP ox4YCpfDLs5FBndd2e679DvFv8etqcPE2t3N13UTA0RikBZNIQ66zJ034mYzJyMYFgS+1GZzi7nv XeikEkPxnuVsqVaKuRy8DVDYXVYBPkKi+UdSgPBDjHc6BGw4fpoS5PMgNTIcCG87sFdRVlW+z/kT ZfzRvMkYc+KoYIC8pHG5fjvjt77HAonvQGdH5OM2mc1313tyg+IeIi9jKuvCVDgbdvdOegkDz5mt t5/2TBOPg9jXIanQDQplbmRzdHJlYW0NZW5kb2JqDTY1IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9M ZW5ndGggMjczL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29t cG9uZW50IDEvV2lkdGggMjI0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDIyND4+L0hlaWdo dCA1NzAvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCogr7QdXAMtg/yucwOo1/lhygD6E7K9AlTdKhIel Wxn0M0I17zup3wsE/kiewYz5R4EgBncz4p1xLdtTg8xwYCRCFW+krO0hU9l4zz7wipFu55gqZ/nf sa5Y545cgwvgZPDRUuhh31N4aTnQkFX5nWAcwvPxuhHxDtsMA0LIq1Ha/hsK7sXrsBHQPDJN5Gdx mKJOGcWtloRcQZ3kSextMhCI0oxJocd9+GwsTb3JsL3P5BVuh6m+sgS/QTPm96BuJlueaC9q1U1c SZ0jQho/1yAJEUxEI6ab6uEAWi7rxtE5BkAkw63a3o697IRlMSFAI4btKvoh4tDjzlRrPxyAxFP6 qAktfqezFqfHI0ZxRMIVyg0KZW5kc3RyZWFtDWVuZG9iag02NiAwIG9iajw8L1N1YnR5cGUvSW1h Z2UvTGVuZ3RoIDI4L0ZpbHRlci9GbGF0ZURlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29t cG9uZW50IDEvV2lkdGggMjQvSGVpZ2h0IDU3MC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0K2mGSxIe/ 2DNQpDcgIUWuy7LDDzbHD0ngzRfFfA0KZW5kc3RyZWFtDWVuZG9iag02NyAwIG9iajw8L1N1YnR5 cGUvSW1hZ2UvTGVuZ3RoIDMzOS9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUv Qml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDIyNC9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAy MjQ+Pi9IZWlnaHQgNTgyL1R5cGUvWE9iamVjdD4+c3RyZWFtDQomR00u+4DqfD5QaVoeMfJlPfF2 FqjTmQ26y68tbAmmZuuxmMAyRVfp3Udhud/Lgl4wNe4+rtKJMs0Q/kO10RHZ0rl/+rJKJ69grkEt Avni3XGYLXY8La5g7R+RTR8DA7WitKx3ASK89q/DUBvLzTGKZ8aiTVs9B9Zuv5BawoerC5klJu15 5YeQjpfiKcQILUCWvbMjCuE9ADuAuDhqzvSPrMcoMjub1AsISj+Qtiy+Wq3/wqE1e9BcOfJBhln3 LsB7RZJNqUQ/lgpmOaEquvb0J9asYvI6cI8rPtCglyQWUtoMy13Ewsc8mmmYqGmY+qXsACwZOQ/X n/BLJhDEDI30N2yNYAmaBSvP2o5PQ4VVe2aGvkbf8e2f5HIFP9utlEHjkFe5IJ5mrrsvJlviPbyO m2bYCQreSK0BjODfQz0eNkun/TXXHJ3praLl7GOoH9WykhUNCmVuZHN0cmVhbQ1lbmRvYmoNNjgg MCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAxNDAvRmlsdGVyL0ZsYXRlRGVjb2RlL0ltYWdl TWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA0MC9IZWlnaHQgNTgyL1R5cGUvWE9i amVjdD4+c3RyZWFtDQqupjln2GQruG/WL2oqWcttpAAzr7Boz8tqXSXWzoXS+rgrGCM0OnUt4YrA 0EbiV4SLgKBUfRTTkosdEIjDNUPWyZYTUMXq7+BQyf83Crd46gKgjcQGaCH8QzPb8GOKj1TqGEhy ZHipKVdCoa+9/teW9ujRR4DK0S6BdJA/f/LYxfXfpywlLKg7DLVEyQ0KZW5kc3RyZWFtDWVuZG9i ag02OSAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDQ2L0ZpbHRlci9GbGF0ZURlY29kZS9J bWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggMzIvSGVpZ2h0IDIwNC9UeXBl L1hPYmplY3Q+PnN0cmVhbQ0KFZ9II4kwQzRHw0Ikrudt+MmB76OPuAfQCr4rq/qpxTYx3VLKtV/n OK2HJQvX+w0KZW5kc3RyZWFtDWVuZG9iag03MCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3Ro IDE1Ni9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVu dCAxL1dpZHRoIDg4L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDg4Pj4vSGVpZ2h0IDU3MC9U eXBlL1hPYmplY3Q+PnN0cmVhbQ0KLzssss51Md3K8cl7FT3uchhZdbt18mdqC/33zoija3fcA1QN vaKYm3vsdo6mgrexmthgmSltgqcCo92QtRWyBMBvMc4O7HF8tAgDggNGJYRyURdLGktUb8G8I86j Oe0Amo7dSiX9u0kqQJesZKjlqHLIiGc2WSXvy1J9Q29nHaE30dJtnCEvqnutJyURSi9SJz6NY6qg PJj4tg1BDQplbmRzdHJlYW0NZW5kb2JqDTcxIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGgg NDY5L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50 IDEvV2lkdGggNDI0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDQyND4+L0hlaWdodCAyOTQv VHlwZS9YT2JqZWN0Pj5zdHJlYW0NClTtxLEj+glmiDGD7KvqsC0Y8xNPxjR1IPpivsOR0mHLZE0y 1tiCvo7992zrn2DyCfI5hFvpx5Z5qS7XExtlIrYnVdhfz3Sgyc2/1ybCF1PDHJKrcBRiLxip3zJs Zoadfd9x53drji9ckIEQZLljj3XhmhB7MPrUS1fyON5kOQPlQshsJdBPEg/IxfESoqzRD/eTTU4v Gn1/6N4HsCDoZ+MDdFfF4FDzaWsVlqiEFrU+MClmQpuiVw4NfSX9xelgpDws0wTPkhv73pwF1aMH LXnRP8ELtgTXdoNYcUL1xlX4f3IhSLpqSZGT+aL1Q4C2mPD7gPhAXXPbzcyxi2rOrusVZoa0ZNW0 DEZHLA5jctnf1xNOCSq+2kmkuP3tbV9n7PvZCtGtYUyKYuiMtnFsC7K6+djqp/pyOufx2GrgJDl7 L8on4Je84F7ieF6DNrUbNJ8j2uDoabQ2sO0IvB4qH30rBk34dDviSrV7YJPhFvEgbsGWiiHtRq7V Hpne9IFqDos/7yS/MHkOHUFzxmEzyPUuqN+oZmv2D9ytxHV0Kf1dtMAVVfDxyrqul+OGlX23S6Aa hJqnucQXN13pPMxKmmElK37ZAalJr7jOf3rKfzeHyrj+KJQNCmVuZHN0cmVhbQ1lbmRvYmoNNzIg MCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA1MTYvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0lt YWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA0MDAvRGVjb2RlUGFybXM8PC9L IC0xL0NvbHVtbnMgNDAwPj4vSGVpZ2h0IDI5NC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KWOTKw2gY Akd5SVH9RrzQIXhUxy1qPOT77OKMVKnR5opch24DyPNdcrsWu0xSn5qqxW9Up+MSmgBlCeml+iOO tVlMaYl4a2aopHFszxS+flKGoIFQcZRz/gXVnWoIJcppKBwvoKXKBHvaJPbBdLSqfM9GWC2+yH0u aOB0IgeT7s3oC3t0vj+e7BoBpYEeWd2mhAOuZm2J+ZTCqNtBWWKjDWcuA+6wmRKl4YemjVRYwPNy zKOMpXrzfnxn9YJveamiXfiFxw6J9h1Bpy4vSjfyLh6/z38azOqYIrX7aJ2f8v0Z0Ca/TRTEh1aI S9OVrgA6QqoQ9VBAoFLESe5wQmuQ+zZg+exjpLLvPUA2NE4PUC30h2hrzrpq7S5245oy+QViWc6L UZ+o9uQvQUdjNwz6lIH+M+Uis5rS+EQglD2AzfKFx6bZAesqbhskHVVP2rGZvjn66AIQFeu3KCM9 zlI3nKVWu6qSRKDqiUrGZmWFMtfPSYYBfMi3dd2xzVBMk0hlwj2JjWWRKWvSnPzf0RCB4cD9GFed uBLfI5X1+uTRpffjgipr3QA/j94n+5E0+5OL+MJfVEUGdS0t9ZBSo8Pt/Gs22j01AnKp2uOcqE8x vm8JbJcvlIHbaOo1ZnUhQuT4m27mXXFWaphg5Y1Q/5cMe0sOKwSHw7rtVg2YWImyJD/EflQvDQpl bmRzdHJlYW0NZW5kb2JqDTczIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNTQyL0ZpbHRl ci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGgg NTI4L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDUyOD4+L0hlaWdodCAyNDYvVHlwZS9YT2Jq ZWN0Pj5zdHJlYW0NCgfUbAcbTnLkMDfhzq/RHRDA07dC5VxzbmR/i6FwpWUGM5lt3x0EwpkWzRYd yOo4ntoV1pb7S6ub3EWEbcwgTH621rsmNlaaTg+bnww3FyOZDRQixBLDaWkiLNv5Iu8FVszUjxEB OrdFU3lhabOfLszNxkNywb0AahDdsqnbP6vGxhSJAvvOQXVFHSrL0eviTyFgKyiwdzad+p2kRLL8 CV2Fv4tJ5XshFL1T237yARde27rgZiFSYDftyurqaC7pFyN12wZ7y7qAbMqK0CoGXK0GtdQPfurg sjS2LWiTe0925fc1/i5ujEYfRR4kroHYW8ierd0I2OWgBik7dl+DTCUiD2xDy4xaFdqdCkTXDwfa D00Vsyc57uScEHoHh8HwtUdDJPhLwUZ1LgWmusCta4Hi6GEqOntWWmXjsrxwXa3+h8awbfGis1/q xbmNT+AwnrV1WNd4xtK2OscwVI/Ia5xlTt502IFG4BI7kaJgE0ZnRA6oyrZKPdNh+3Fm4ASZmtRV v8AknoFd17AAu9V65jI6un4EdP0ES/FPLuWA3ixXOSOI5hHdSFFWyz1Ix0Om37NbWbek/1uWqvUa IezhwfzeFSV5P4DZy54sv1d9cUBzLJh9HtHogZmOE3Txggj28BNQnEM/TfTb4hoO6TI9ikD5yUEy yHSPgDXnyJss+JFmdVRSKCfoZ9HOFf3s8hole3nfXIDSS0avAGbmo0y3DQplbmRzdHJlYW0NZW5k b2JqDTc0IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNDIvRmlsdGVyL0ZsYXRlRGVjb2Rl L0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAxNi9IZWlnaHQgMjQ2L1R5 cGUvWE9iamVjdD4+c3RyZWFtDQoFSJeJWrT83IcRnKAx3zELj7bQou5TLM9WPRrdkzXymp44Jzbu SANxS5cNCmVuZHN0cmVhbQ1lbmRvYmoNNzUgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAx NjUvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQg MS9XaWR0aCA3Mi9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA3Mj4+L0hlaWdodCA1NzAvVHlw ZS9YT2JqZWN0Pj5zdHJlYW0NCpoKQ8CsLYDen66ukx62jrizBP4bXotkF5Z5xUMp2WZ0w0eLJCgh onwoDHwA79P0d95SITNldaUkE6CFWS5wOLTUXqBGb6N34SLcHFSn2xwE8vJBjH7yBezknmOozTO9 ODt8RVldM/HV/ysZ0g52L6BSm3O9Y1t5i/VfvTgeByne5mqVJYKNh5kLdLbembDOIqk7BhdsB5jR Lv/eO82sEUBhbluB/A0KZW5kc3RyZWFtDWVuZG9iag03NiAwIG9iajw8L1N1YnR5cGUvSW1hZ2Uv TGVuZ3RoIDM0MC9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNv bXBvbmVudCAxL1dpZHRoIDMzNi9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyAzMzY+Pi9IZWln aHQgMjg4L1R5cGUvWE9iamVjdD4+c3RyZWFtDQov0WLiimGR5zwJ6OuN0JT/Y0FpbCEUtzQ1GwCz IlPjbapYh/+x0z2Zg4Dx5nLRQpJkcgqoRmmQ3z+fk2u/Cr72pgqa8vYkvF92CCLuFHMjDRfC5br3 X5NYOeh2rP6d+KUIpANBz4uUXVrOMs8wc9IiCpfqrJw4PM+sZwAef9tsx1zYyeWsEUvzkOqzJADT Ha6OtIIQ/IJA3D7SuIn0msMWe3siZ3FMgA7R5RR/SdRITHYRlgJ2PLzyG/74mHm3c5qV0ircRTLY r+yM2cQ6KduFMvxLx/bgL0nwANWon9IygUAARDyFptudghmn4bV5f2jP2auWmaE5rXbr0mFhp1hA 6a5QfCMfnYUg1tcj1nDZjhjFwEsMNMWLrqZlY0FR0cbFikiWBG9wiy8sKLqYWB3PYUQ9CsxArLkj 78XZ9y/0qJ820MFSLbjcxzdwfgjEQlvoFX9xDQplbmRzdHJlYW0NZW5kb2JqDTc3IDAgb2JqPDwv U3VidHlwZS9JbWFnZS9MZW5ndGggNDY5L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sg dHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNDI0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1 bW5zIDQyND4+L0hlaWdodCAyOTQvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCpLQwQi85GFZpSWB8gcP ZDcHuaduk7y0oR0jkrpXmc6rSSlgVmvRljP17y2RFbf7/YjOozZ5gkBwip4o1fodvjk6MFINbexB g8hHsAUHYzzqS6cQg8uY3ei/NuWC0aouqLcwqWVm+Rrl/Nf0ewlhz8LwhXvEVnGQ2QpOyJpHdrHM KASmXzjPXDzLFUXOLMY6uzcTq4B30Rf2E/BMVQk33/jAHaOmnDvk7SWB0srFFuK3QB1trn5Ip6l0 IG0NY7HXTWLQm7PoiCcmAsW7hk3Wax8wuTJ/oTm0smgCbIo5FKuhWnoEtaNNur4HRIzuOxDknTOc fCnKYXzPqKV6/cuZfRNq8RYUK7nPth+9KPho7H6EFJPCb+9kddLeJclSfB3r+c/R5uSVEKGcma1Y anTHacka5+qf6Z3EziNw/hqr4Uj8H1NFnVFRKQ/FdwxcHpfZfRAUmhW6LgtT2TAumMMliHhoPk6E Xd+bkL1el2cF4te28hWB/31mqvf/A8hgfAVUPghFH605OXjLKFvjH+iCrIPid4w4/DFCE4jAd3+P vHx9EaTxOQLqCg4Lpqyl3wCetu90o0bmm7042YZGKaJVngLIEeJGVevPpASPXflFLnTb2XfQWp8N CmVuZHN0cmVhbQ1lbmRvYmoNNzggMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA1MTMvRmls dGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0 aCAzOTIvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMzkyPj4vSGVpZ2h0IDI5NC9UeXBlL1hP YmplY3Q+PnN0cmVhbQ0KgR5dlNcLDm6w8uB4cTPINNsqYYS5j4pfcrlsmbxf797bnb1DTlm15ICa R9v3JzetYfhKGmXl6TWMO6JTlBsERR8DhssLKX5Wf3KwThBT7mhwcH6UnzqogpYZHYISbLYH3lI5 OPC3YhLNMN7U0Q05L76fBe+00Pq3ybkFbooFrazk8PFKumB6ZdLlNx7hXaCQX4wqZ+/QrE7Zj+3Q hqQHkuM1oWie9H0JhTCWmx/q1GjHOp/ANeJizJLtnE0aCEJGpw4OXRWKl8mvLVUTyUpgZiRSXxqP Xu3rKYp4ul1yxNPMJ+YEk8U4hqkjvSTm8LslPh7bTJpJMi6cscOH0tAAuluT+V7J4CDmzA8xUW6T PjnqkR8q1ao65+uMx3OquTu0hWHnACQFNN3A4dvWjBI90LIY6INZORdnC5uh/tr7aZ4xICA1nurX cpbIFWlQ/teEsLtKY1sP7JcVm8SoQXbm932ubjOSG01cm2q+Z+oheAxVd3pCgIGNamuhV9GIEBXa fsq3rGyaadiwmhOQBAdofBwyyAwRxtlX43FB+snRCt44KWHs2z0wqk8pDYd8P6L9lRX5udzaPHlY MzMft3yTFgZF4ryv/AyB8vkZHPlAaf5KJxpYX0T6Fc9KBTF+8MsLvHdwcjf9XvAsTYRc38LnRTRu vqtjKbv/zGL88a3nkYk1DQplbmRzdHJlYW0NZW5kb2JqDTc5IDAgb2JqPDwvU3VidHlwZS9JbWFn ZS9MZW5ndGggNDE4L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVy Q29tcG9uZW50IDEvV2lkdGggNDQ4L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDQ0OD4+L0hl aWdodCAyODgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCg5rWW56NyRD0VqDq2XuLZzSfMMI3kJFMKJk BaWrqgWQYNlWF+WktZkzBBNFgY/43/ZSgcBSdpLx47ok9M6N640VHD9hP3tBP2DjSXuUo4TvhYSu j3giKHFPeQkYSFLFVTkyg9HNcQHNIn703tdPqKgsrhuyhjmn7O5BzbX6j+mITXvmfG/nHC5Av8QP KKWbAISr7sJVTpChL60Mg2FEHY037AOdRXluGy3JhA1wV5ScYA0a7Vn8g+A7Ygfap6N7NoAc+Mko yteja0AVIzrQ27KwiH+zt1480rClyidoH/Ag47Ph6Noqi6QdrIhCzizFaUvioNhPldqiACtjSkYB HnM83MaiTTtztpTwfQXlNYOX34k+Da3ef1t/vSry52Z3/iWUpI/xFlnisYNfatHACsNBAAOHVc/j 2NCM6Y+eFyo3joFp0cz4gu+DhXMuxA0Se9CD1p5hcNqWxs0F4EGObTpb5KQ8jHFmqT5JInGGSqrf MSizHAWEf3pc2rASgW/9HglZB4UtSKnBm0NHGXnsZbf9hU74J/YW6KyFwfm9NOyBHuANCmVuZHN0 cmVhbQ1lbmRvYmoNODAgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAyMDgvRmlsdGVyL0ND SVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAxNDQv RGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMTQ0Pj4vSGVpZ2h0IDI4OC9UeXBlL1hPYmplY3Q+ PnN0cmVhbQ0KwCMpoqXzDLCZh8sXGc35JP0odHdnYR+KNfdV+vm8iJNehdO5lPWs8GCtnaI/Q8Du UceEVC1pzsw8XMsHZSDgwev7/qthJBOjKiDtlqKr/LFwqh4X+teHPuRwUZ7uFtxTPvtBlcRHvAul n2nWCIUgUcN8HJMLYJ6OZU6473qz6mlNnprvWMXqEI49INmn7fA7AbfjBBLgXLbMSrohosChCXhe xj1/aY6coLYX7VAEzff2V2jDmSycndobMD7qxXijpCD61s7D6O/+Jl+TkNTYww0KZW5kc3RyZWFt DWVuZG9iag04MSAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDUwMS9GaWx0ZXIvQ0NJVFRG YXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDQ0OC9EZWNv ZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA0NDg+Pi9IZWlnaHQgMjg4L1R5cGUvWE9iamVjdD4+c3Ry ZWFtDQqward8XuoPkllknoeDtPeGwxf5652sTnlrNXxrqRzGyzPWE2fzcQDUz1JhKVTVpiWe8Cn/ 7wMKbmAaiquYURNZuraZHyqGXiN3xRALAfgaphYY668umpaG+2ZA9QXdTamnyvzzyzmlZ/zYdEQk TlpmvhG2hsypoupJYGoj3PuEc8R9Xj2Wk6WP9pe5DuGIgEDCajl3xaLVdG2qKxw5eKsdC+3fhGVa gsL6cObX491qMzONXAAd4Pu1Sq4SeOZYJZWzZFkVN9+ObLjkz3F+BYEXStdA8UL+UUu781VejIhh B3kM3B2JWdt6RlSMuDMLNNru0GefwEeCNpOEQj+CPqTgzuwuLsjxrR5TnkXfvqIRblEnKtgEIL6N 8QQKZkkI7iRNpOo1f7Ursig3/Iwlh25sZg0bspI//FvJWCz630rXFeKi8LJSFg2UNPa7ipo1TMLE cf2L1DWiF8MoLvVYHT8q8otaFS//dZM0rLkKpn1CwGnHEFejTu34giZFH7i35syJO/u6OYynYw+4 vQJs9P7t1NHzMz4rZHOPVJE7/Mo0PzuhiVWmVEwK5B94QUnjYC1tLORM6iMe02OyL4ES057l28ZM S+cPkdmXDkIn/qsO+rnD202XnVw6j6kdtmTRhAKEN/IRYD87HXhwEESxOofHXw/scGkNCmVuZHN0 cmVhbQ1lbmRvYmoNODIgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAxMDMvRmlsdGVyL0ND SVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA0OC9E ZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA0OD4+L0hlaWdodCAyODgvVHlwZS9YT2JqZWN0Pj5z dHJlYW0NCjjqE2WgKNMgfh69KZRG5aMXI5NhTuPvK9opGVUNHx17X5hf4daIfiTafGOsASRHvAVk G3sXK02ZDXr7FxHCaj/9RixIMA++Yb3rCr142Y55kcq2dd53HObFKa9HkXwTd1hUS8WRixQNCmVu ZHN0cmVhbQ1lbmRvYmoNODMgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA2OTcvRmlsdGVy L0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA1 MjgvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNTI4Pj4vSGVpZ2h0IDI0Ni9UeXBlL1hPYmpl Y3Q+PnN0cmVhbQ0Kl+x2N9q5CTrzs5ItKxyzrKyV+LbggHWYfYD/YsOIMWEII7wu0UakjhSQnth5 KTYQxU6SiGfuokot1YB6oHHA5mYxTtvvXoH6QgvLtGnhCfLY6nw5iRhMVKsRd9LeVRqH/nOhjIj3 rfKlahApDCQ7hR0uOfg2yEj8sxarlUpjeXAI4/1H3PtC1SrP8538PAC+VWgEPfxmmEtxv10rv9Xw JiwpPCJAGxp7YlqBSJkXwoTOTNrGAzA1Gc0Pxw/OF4/YBTf2iu7nn+tBiPZUTmQd1a2m+mGvjYsb n5o/t6mlu0PkwG/GnoxU1FJ/uyDq5/9KA7ZbCNXlAGosv0hCOLKIKD3zVWsszBHfT95Ybkc+ZJFX SJrxgP7raN4a8wuO9zr5xKdW274MzDYcCTZRDGZTBdEhFhgW7fqELFy+050OrxwwEfKjsuMxSyJI ihAcSi6b2HcG3fzuGKHEcC9pksbXVU7y8TvWdUI7HmV8vCA5jwThR5m6bDye8bv2+uvHJ+ba7Knn 8hWKX7UeEeIP5OBR5mtZhnR0GaNFsNBUlrVTsPAmbNS/WirC9oZYOgp5NY8+EYL5DndQ6M1uIQZB irpPC9MTYMLuuInLUmZeE708Wp5JHPDW4R4vZ3HmmW6nZG5SW/ri6fkWVDU0pytl+SNH1fnYxICq B+utJi68MXV7mt5TB40McPbdufIo/bSaH4P5DZjgksOxBvBa7+Mi04QH2JbSv8byk/uHIow2bcFV bDR8zAE9uPlx51J2jEqxtDwVpY3l+20iId+RBZ+IFfJjuMlt2Q9nXPAaT1ZF6Bo2psFukYVQ3qq+ fPk4uBziTBjYzJaMS5yTV9lTjSEUP8is3aRo1puKufRnl4WjpBX3s/yaUo5yTPINNZWEctuoDJss YzBqEQQNIdmts0Xs6XU7RL+WoYWDPuEiMQ0KZW5kc3RyZWFtDWVuZG9iag04NCAwIG9iajw8L1N1 YnR5cGUvSW1hZ2UvTGVuZ3RoIDE0MS9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRy dWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDgwL0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5z IDgwPj4vSGVpZ2h0IDI0Ni9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KZjasqMH1uZ7hSGaD50t7XqTt 7WvBPrJnzvz03UQyDiVZrhUSmAFAnILCgmZu93EMglJSCTcSyWOU4mX+1bz3zEqOmmuLYRSdybAC ChOfpTMg/yXTqnnJ4z5QSNgfC2gTkNoJHFYK/5FG6qdwviP+PvY+YsKaGCjND+JngBKlLI2XExNF HkMURNBvVuCnDQplbmRzdHJlYW0NZW5kb2JqDTg1IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5n dGggMTc4L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9u ZW50IDEvV2lkdGggODAvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgODA+Pi9IZWlnaHQgNTcw L1R5cGUvWE9iamVjdD4+c3RyZWFtDQoOv3J8s1LKDgr7+35oxwaT2wLWYyrrYsCBQRP9ZJaL0y5L WKqG8QANFfMZKKi8x+AB8GR1j3mpjL4l7h9ncipkipZ+yOvOKtF2w1yNbNLB+NzQwtfzBSG6kTei ouq/ZQFFV+TB2b5I6VekXWHOO5mAW/TWVIuTgT9kKzFm2uZRsETxnDeH/vSoQ1KqZvmXg5Oiev6r e2zxGYt0znQ7hcb+jvZDihBBhZBLN2OtuhrpgNHHDQplbmRzdHJlYW0NZW5kb2JqDTg2IDAgb2Jq PDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMTgvRmlsdGVyL0ZsYXRlRGVjb2RlL0ltYWdlTWFzayB0 cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA4L0hlaWdodCA3Mi9UeXBlL1hPYmplY3Q+PnN0 cmVhbQ0KEfgSPnaFCmgzIUzUBDsIkhtCDQplbmRzdHJlYW0NZW5kb2JqDTg3IDAgb2JqPDwvU3Vi dHlwZS9JbWFnZS9MZW5ndGggMTQzL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1 ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggODAvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMg ODA+Pi9IZWlnaHQgNTcwL1R5cGUvWE9iamVjdD4+c3RyZWFtDQrUUXs8D//Jif1+AWIyQMY5PBuH NAdWSDjLbfOefipLgMFfo2S5bDufy7AekixA5VO/9OKND/exmiOmMVX0I3yCNmShGP/Cj8X4u9FB xL6F2Vpf3BNCD1KEypDLHetTZtkUJ6PKhiKUCiyhlb+o8og31FZznUq2CtWwAhis49DSf/JHLNPQ iqCwo52NTEK6Lw0KZW5kc3RyZWFtDWVuZG9iag04OCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVu Z3RoIDEzNi9GaWx0ZXIvRmxhdGVEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVu dCAxL1dpZHRoIDcyL0hlaWdodCA1NzAvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCiLxP7wvZ/GOSx3n oEGN6NveJxrIdSFAkKFJxDImvKvPPWT4jkkIK9EQTZ2UFVEs0ZUw1qqI+Kq4ztyJ/mwTFadc5Opf mAspyFofo8ZonzxQuAM+B/k3cn60NudtqOyN2p5vyDyw59UwZy4zEenJo9oZBKGeGBObpTBGVyxl pFgZvejALBWJ3V4NCmVuZHN0cmVhbQ1lbmRvYmoNODkgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xl bmd0aCAxNTQvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21w b25lbnQgMS9XaWR0aCA1Ni9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA1Nj4+L0hlaWdodCA1 NzYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCiokitEJ3LNeWlaCoyRyhh9/LdFdcLm23tibGgYyQDjS b3OJHo0YNaE7ZrT5hndDudpDjPGVI8G8FG03AtZbmb1Pe8c+FC0RrUC/PWok9A3SzlYAZm2SBPP1 /nrTeqqjdeBMiCa1kNMJBMzoLlANfXmekj3u2SBDuWTIH12UYxuZ9XPCtHjdyiYor3WiGeAQ4boC o0qOFwUUGWMNCmVuZHN0cmVhbQ1lbmRvYmoNOTAgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0 aCA5OS9GaWx0ZXIvRmxhdGVEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAx L1dpZHRoIDMyL0hlaWdodCA1NzAvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCv6S/QdlQmM7oY2MkRO0 qZsU+D0nx16CIRtXI1zkkq95zTJtsPTKPbWvgH2jgggRDFlDOacK2cLEL+N1gWF06JpWwa7iBQpV YLM77FvpWFXpR85zo5HpVzW80KLkgf+hDZIF3w0KZW5kc3RyZWFtDWVuZG9iag05MSAwIG9iajw8 L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDI1NS9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNr IHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDEzNi9EZWNvZGVQYXJtczw8L0sgLTEvQ29s dW1ucyAxMzY+Pi9IZWlnaHQgNTc2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQqez//rS20sqb5MYWik iwA3zIXqaiF5o3MtBRmw5QbsfvGVjb6TWYJf7tCrLIRu4u9qgPudZxE1vyFGwMStfjSZI4I1dVV7 OcofDiv5pLKE4RtSmhwSTiYZps0uSS/Mme3fEgDf4vTZmTamQiMq1EFVDESo94r51PMOCAmZXv4f 7rVlF4XjC13M6dapjtKemPTEW8RQfxpZhVC0uF2jM/cVySfFDXiKBEpmgbYlsXZH5OKmY8u5fL1C OcgyI0zy9BCAC4PmEtC+aYsNIxAXSxQAIvurhzIFHdsIMnhJchj1QcjIvd14QIXtswbcCdm3ed0e dnbwmaX3SNPfJfFvhnwNCmVuZHN0cmVhbQ1lbmRvYmoNOTIgMCBvYmo8PC9TdWJ0eXBlL0ltYWdl L0xlbmd0aCA0NzMvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJD b21wb25lbnQgMS9XaWR0aCA0NDAvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNDQwPj4vSGVp Z2h0IDI5NC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KP8WTXUrXPrDhWGAE68HKBylxKcoOTTDU9Hzw VylcXWDqyjKEErrVPuhffArsoAA0S2PP0cYpzbOsTz9wximaPhAg5FELj/Sl5XMT7EnD84BBHIKm DdTjVHY+kB+S6KFKCCGCp5l4WyDux0zg13+4OKpFzavGx6sCSLyjQpRxcu6Am+1bPfEDC4SciPoN 2l6HA29kUw2Vi2MRT0jqjWnVf/W5nCrfnsNZVqgD7DfW8VwsIYoUcWa36/FPu2MP8Qoea7aVTHIu Vn8L3xddg9R37/ceknZ2yGbtzMdeY9RGfmY9hq6AsAXkwW7i54ClYcxtFSNDRKHvyL4PDPy4jlbD 9MxJTHIVynF+NtrWRHQNv1kE7IJZQMt1lzin6HBCxXwVRtcnddxosz135L0XlazbsBl0SYNmKZwE oWklr2SR9VuhV4f+OuBEAZZs4pk8d8k1z3QpK7ZFNjeEi0Jn9A/6VDpKNjj2ZIRtrL5LkzHvjBNQ /bWJ3SwQcpqZG7gEDMigLADmYdIEwjD0C8Yq9HuS4003/KVPQaC6842UyPAHcuCqPeNponSGGbWo g1Xj+e1FrwPAfW2Ya9ronuopeFlEP1kMUuEp/rinNhjzuuHTKzLum74Z4tO/fRsNCmVuZHN0cmVh bQ1lbmRvYmoNOTMgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA4My9GaWx0ZXIvQ0NJVFRG YXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDQwL0RlY29k ZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDQwPj4vSGVpZ2h0IDI5NC9UeXBlL1hPYmplY3Q+PnN0cmVh bQ0KxUFCm0XhydGc1XC8jowvGAJKdp2m/N1uLcQeys+KBtqKPa+Rxo8g/TzD+4UeweW76Esw9524 ANsO/8MhtaqwoutvPmEal1ED/qBCkmo5xZRDeesNCmVuZHN0cmVhbQ1lbmRvYmoNOTQgMCBvYmo8 PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA4MzUvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFz ayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA4NzIvRGVjb2RlUGFybXM8PC9LIC0xL0Nv bHVtbnMgODcyPj4vSGVpZ2h0IDE1MC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0Ky7wjK46A8r87jwhZ 924cbVSUy1eP+/dYVeWvE44cULhZXXjvBxeVGksolsqwq2VNCues60SeJLOs1tcDlehcFYs6unTA Mw4gengK0eW/W6rXMj9xjpw9L2EmDW20Wi7YkFBx04aBBq+KWOXkVrXJ6MqXUcEhUTHtqEMqKbu7 HBY7v/vnlmHnmKrHLQXyRg3RJXfR2lC2/Yo5k1BhRyigY64nZwv0Qm/mQzdMUZ6zwG/aZ7PwX16p mfe8Tz2lqVESdBpcpIE5EinMWs48077214/y6gVAavzEEAvyMRr0WJMZonvojCwXfAYtmCI0v0+d 5euoiXh8/IA8IkW9rY1yXYCBCy6lTBc4XSBo9QgGc5Z1cK+61fzJ/OKKj3dvRF89gjhD/nF7cU5N m+WHxo2k7NEfHO4lG418483sEi1T+cqm0NuUIAX+PLn9neFw7fHpswzimiggAYIXErj9uCGkv+Hy +1MmhBQ3qmPFWpmU6VJuJgivToSdmRduFyZlA7qB0qRanmMlziCjComcMTi4lP06jdUYFLmOIeDT SvIUFgSLUzvLpXLJHKuE4/XRFYa/ULjyy0+CJ74rpl2a2NGaxYifm1sZaPxfFO5T9gAx3w5oNbJh 3oSfX3z82Xfqt3Zr1QkLNT3TX16GHYqqyI3TO/tDCTRemixYxwpi4vF8PBb5GhoaKqEQNrHdL+5/ usLWSkzAShezfpNb9XQ8YSgONHalSE59w9Fp2P5s2IxTgnGEGZZiKBNgg3wLPplkv1ilNctGxJ7L 2SYtUN24vpdrBhnM1xz1+44sfBRJr7u/6csns5ISTYEXQc7SBXOfp2vbKEzAdgzQXywRL+G7+rT6 XFIVi7uAl2ebG4UEidC3S9S7kyXeHQl3ptKO4kPgQ4OTkS4iqVlde1fE8EtLCUIo7Gx0w+RAT4o+ XkiQRIpRKl5FEIi7ncB4QVe5F6yG02hJpqc73C8L0N6C1p9XJtJzYCDPdxK9j65QJbhC4Pe3w6IL 7tPS1suY3o84piM2wAh3Bm6vRAjArL0FPnkEr7FbGIj35p8LA+tyn02Df1KVAdP36LW5D4XDbPb5 xT/AhawMdj6w33QfgRitsyxymJKKji5WfQ0KZW5kc3RyZWFtDWVuZG9iag05NSAwIG9iajw8L1N1 YnR5cGUvSW1hZ2UvTGVuZ3RoIDE3NC9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRy dWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDE1Mi9EZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1u cyAxNTI+Pi9IZWlnaHQgMTUwL1R5cGUvWE9iamVjdD4+c3RyZWFtDQqM0ouyNWssYnBjZ66eE7WS umg1zI3JA43TR+f3Xd1SoSRk92eNfGiZXXCRbAtMirQPO5gt+ZL+PJeTu1Nn1TQvyHY3L/GmgS6n ST+a4jDEJQTpxE3afO8RL7aqT2DuN+ajwesKUcMxqTiV7a29cZCqIfsv1PvuDq5HOeuNi6ECmKtV kZPqiD5PUP+q94EeGXVAF0oKdmKD0a2x76es3ujudHBAFQRJ9kvs92zjk7QNCmVuZHN0cmVhbQ1l bmRvYmoNOTYgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAyNjEvRmlsdGVyL0NDSVRURmF4 RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAxMzYvRGVjb2Rl UGFybXM8PC9LIC0xL0NvbHVtbnMgMTM2Pj4vSGVpZ2h0IDU3Ni9UeXBlL1hPYmplY3Q+PnN0cmVh bQ0KqZtJdMbAcAsQ7sy0BNZkNgktXANuKwbJfYR6m9/dE3I+woQEfm1EpUTfYFQJIo1zYD7ztOWD +33yNHXY+JGy2uXqerU6We9OhxkepK5Y4BZTO7Yx5YZF2nsgABgY8D1NdJuwR1V1uHZ8Ipt14ZPH muNc9CpJMs6PpEGpC4FPTPQYxVx2nSBFPDB0M7u3MEjf4O0mZXsAP+Ju5h7nkjmRHWvw0rJLfTnq pOJFceWnbh/IPjY5nQak+d4xZwTO66o3wXOkIffRjJ4D1kcEtbv4NtXXCqQW8NaUO8gcrZCs86cn ml+x160WRv4fQLMAfI9b4bZe7Rp1g5QQrn0o+of6rqOvq4y/DQplbmRzdHJlYW0NZW5kb2JqDTk3 IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMzAxL0ZpbHRlci9GbGF0ZURlY29kZS9CaXRz UGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9EZXZpY2VHcmF5L1dpZHRoIDM5NC9IZWlnaHQgNi9U eXBlL1hPYmplY3Q+PnN0cmVhbQ0K/6IGH9vosiJ1OptKy4qJzlft2VADhNlWEJRxrKtEc96OK5K7 sgCCwN8XUQPqx0Dp/WJHNomyQBr2iq+g/y1gxY0YWhrne245qyobo8g3+7a2xBGaeJiH7y5JeR2g A+2QZcV0R+w5KpBrcgCqSIV/sp4R3mw80Z1l2ySKWRLNBvPIdlcN62zLAVVXFwNO2lLwet3kIVqW ta4Z/nesNvlxsHASc9BTMa+1j7JqdB4N/U5q2Dy0cobvspdU2dFZfv+NV/vynKl/XJFJCydTYlZB 8fmXoX2JK4yegk3h/fdGSY2D4NA1rHVckquNG7Df4dloAUdH524DN1QHEqtH+phUpOEYroLAMeqY Edh4GnoKF7gvnPFdSVUWhtqD9yX+dSVKUz+o2GuX3BJd+6Dqug0KZW5kc3RyZWFtDWVuZG9iag05 OCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDQxNy9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUv SW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dpZHRoIDM2MC9EZWNvZGVQYXJtczw8 L0sgLTEvQ29sdW1ucyAzNjA+Pi9IZWlnaHQgMzYwL1R5cGUvWE9iamVjdD4+c3RyZWFtDQoQ75/X CTezIHb6GAxg8pLmuhltq1AU2BQZouWavogBlHldQ3elE0oa8b3D9Ci3GimgsisgYJ3gnyM46Ypn i8O3/HYI/LoMIpbDpwiN/LeLzpopFk2CsN92Y+JizsrF9W5P+SXW8IA8Trx582xi08LDXqGX+vmN N0RoLnv3W1DJfkR6Nub6iXf0LdQt66xbZS6HITkS4YeNzu0hjrVzVQUPcj4vb6U6T4WbW3qyp3Km s72TxICkLAhHkbrmMvWqjf2CyfLFCawxGDPCAmfbEjllhGjp5mKFGk2wjFyKa37aU9avUZlduMTn hU/X1JwvKXCgUDox4zQsyYxVQ6IYBogN7YHtpqiVFdQ+bUjutcaSy2sQzRQ5Ge4YB+A2iVwQfavU znGA9sA67lvSmYhE/9YNBFE301SdwcDmcRTubJAvSqUUPuDcuBsSgzbDBT5pSvnxT72j2ozsLQpV fB9Dd1bEURHatIj2GdLUDHpk5/lnoOIEDpiOWoj32co1gfjdlGxTk2M8ZjR5mcTAw3CSa7q1AxdW DoL27web1svIwQoiZNgNCmVuZHN0cmVhbQ1lbmRvYmoNOTkgMCBvYmo8PC9TdWJ0eXBlL0ltYWdl L0xlbmd0aCA0NDQvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNw YWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQqQ TUtWVQxgcmyOX0w1RX3WMOap+3RPf4Kk7COuxUp6Tdt+BqhSdD+40/+IlZv7BCE5W99/lGIEN/ik oXNZU1SjJm5me+EoCep6HI93v8ER4pDU7XHkas/lr42fIBXKX+Ia47dpdnpIVpPaH60CpatTBCdl 23gVRpamgfxh/6/Gzokhfy57MBHidWzI8KLSb70sWF6yvbdphZDDOXhKs4RZSJisq1GSErsiO7+C 4fxo9jU2829qbTMVG0R4FbhtyNROxm5a60hG023IvbXdNp5+pWEaQDG/yUedS3ofTG+cWMLIXFh1 O7HfQZOeCU1Q3JlYU1P+u38Q6yZsXd0+i9EJcdfOxpCQ8W122GDzq/W6BimBF9pIHLHGtj2U85al O8iTmynWIoRv1kulbh1ryRzitqAEUH06Iset1eeoSTaJWaCFljo+dyRiRwjSnDEwQt5tsiKei7Ah u3IeBWtJ1uzjrzO9ozUuFrnkYHzD7jYTLTQTAHKlyqPanshPPXdGfSOmNx4NZH8M3hQyiXFY6mTs kBXe58S0IYH0HbrXxsly7RCLFoTwGoE81waSqBvvKqlFPD2J3j28ly3J9yINCmVuZHN0cmVhbQ1l bmRvYmoNMTAwIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMTkvRmlsdGVyL0ZsYXRlRGVj b2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA4L0hlaWdodCAzNjAv VHlwZS9YT2JqZWN0Pj5zdHJlYW0NCrkBzAXHfHamwT9cdG8e3vzsUngNCmVuZHN0cmVhbQ1lbmRv YmoNMTAxIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNDE2L0ZpbHRlci9GbGF0ZURlY29k ZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9EZXZpY2VHcmF5L1dpZHRoIDM5NC9IZWln aHQgNi9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0K6IvAYHCV0Ktdj+2bW3JgUXpPPH722yiWFIO4H0QR 3ehEKrG3ZvU0zvLhYsFfyfZXdfCG3B7tGQg1IXZz47bkZBO4zF8CjmqhnJlo4gjFLJLx7a483vlu uJhe4w0nPcYCyDf9QSuQrVGs7r9nhjaBFsul3QQ5IIFlBo+3pLyuVHl55GtetTt4Disgdvc9yRLn 8hkIQxFmZOaNDuL7nPJtnWzJ1TpMXrGwIynfmhVRn1l49nJXtHHo8piqRntAdN5wyRIuo99/qRw3 BAurTydsAaBZHnLcKgudGfqL6wQ0xpbEW/nd91lvWsv6X8+iYIXD/83I/1t0lwCpdmLUukyJB+Hi 8vbZrf4lCDPBZnZ3L2E5jdBsVvH66SUbqa6A3CxnSER6OLCtX5Syak042RHMkvAtbwSq0qL931+R O29RD+cd5WCtIpBq4MRNPzRkandZtWCN0wy3lsIRa6C4k93hKwEXne0GTZ34WMe0M7TBH6Y26Hnb 8btQBkumc/VTDBFL5IHtWb5kzsiIj9T/U5WczEuCaQvexl0bo7ToQgM4NlcNCmVuZHN0cmVhbQ1l bmRvYmoNMTAyIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNjM5L0ZpbHRlci9DQ0lUVEZh eERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNjU2L0RlY29k ZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDY1Nj4+L0hlaWdodCAxOTgvVHlwZS9YT2JqZWN0Pj5zdHJl YW0NCtbl73ctBJHo3yQ4ozLHOYG2Llu+b3TEMq0po5nJdJa44x1/Qhk1qHmB4dd15WIgXdP1Tf4i a1vjzYGehSlo+6t+JD14lzOfUjHLR7Pd/hHmz9stqZpCY3EMnl5d6aJFwFP/SYSFfFE1RFFVPhGY tCxinjsmrWO9XigZtSG4T7SPCVkQNB4eKh0AUEl8NW5cxfCHLFqt5WyVLOctzMMDnkAyQ+Qq4qDj xTZDgPnLhNd1HZDlqY4BUOfO9vY3LIRTKahuvJyPPLocgH8JAGkp6EdnP4k+XalXMTBosHBWFxsn E7DH/MEQkERjjzwFz03xTwoEKitPe+8qczh4phP0ZIlliL9AH3T7B+X/CkWEr1I2QLC7ga9j/3eS 7gOd2gX7vS+84OzqI0R6T06Y2CSz+h/mdSW2SeZWL1KFXXFThQi3ijAB04ivVqbJPaf/Ttg1najm 1xwV3Fq+lyfolsWGJFmlA41C7NEFd0yMGNUA6VbOuNkAhCSTbRbpIkJYNSgaMPyrd07s4TO4fFtA tP0LcpvIUC56tA1aI0Q+rWf7AIV8NUaSKowxh4bbXxy0H0EDbAYKFPzI1kFLaBe5hbkio3yM/sHp gp4QPDnJ6QzQ1mM6xtP6sxoIhqDQPJfw0+Jp56iXVNIm9n/F1LDq9zNRRXwUA6mM58rrlge1IaCF hi94k/P++Ib8eDojYYrZOIE0LwQTkdfJnjy0iDhhGT4mMsmi8LjQzbqSIMLc6KrzfrGV8GPkKwlG 7qdijcpUjz5AOurT5fVbVHC+q+9a6I08IYxSMWaIKdp/PVdXUFuU/ArZ0u3Xk2O88UMs4bbFWd2D sxpzvDA4P63X6AssNXdgPw0KZW5kc3RyZWFtDWVuZG9iag0xMDMgMCBvYmo8PC9TdWJ0eXBlL0lt YWdlL0xlbmd0aCAzNDkvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xv clNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFt DQqI3Oy+cG/CV0Ub/8YYyH0xVb4AVokRgnH4XUagwOgObm+a7P2eotuhyZ+nUil81XXGRaM2lWm1 vtG8JUsSyRM/5jJakjl4w8By1UVcxa5/k2yxnX+YAIvAjAEy8pxrHqB5FqS+9rPB7uvyce9fqIHn F+B5zyq6EBrOOcak3iCyQ5ZiMhVTlc8jdGWNvGa1qLqXtlVQ0r/SyNt+a9bTEXACt9TqWJTB2fZW L6yC9PLHdUUcB5rpAjRBO2crT/IQFW5+OsasiqqRwoxph92NpRpWHRAhHSF0WTZ1I9LcaBhrTM6w i1Fu3cqU2aNR+doXjKxDei99Jt47h/X/9ks+USKahoO/++FUrXISvOvrlK3n1lmVV28KXBoQxXqN Dv9f0Xnib/pTq1DIXpfVF0lSa69VBITwMN46+bB+EeMOPjNK9IvCXqyeLgaLT1RxU3rMx8+QN5Ov LkdMQi31QQTvDQplbmRzdHJlYW0NZW5kb2JqDTEwNCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVu Z3RoIDEwNi9GaWx0ZXIvQ0NJVFRGYXhEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBv bmVudCAxL1dpZHRoIDQwL0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDQwPj4vSGVpZ2h0IDE5 OC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0K2dgpZjrbYciQmAfIdVXtLYmeyAuE7I2A8E7qx4luOWq/ kYxv7lkNTX6oxJy3BvyIYQabeYcqS7hfwoI3ivFRfuW0ct/79pXSL9DHO4PzbSACP/AQKAYAlubh dvrPXET+F4YjL1LQhq8HCQ0KZW5kc3RyZWFtDWVuZG9iag0xMDUgMCBvYmo8PC9TdWJ0eXBlL0lt YWdlL0xlbmd0aCAyODIvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xv clNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFt DQoV9BJRlyKviaoPcoxjXVWrLGJFm2cki5V0Bal1IItdUdPt434xpJhJXojfY9hRE3R1f5IbrXtG Mo5KuT/gcE5VHAeollgV8rihrcbrGw17HKzxqR2E7HXwWYNSlAbzZ9tFgyKvWuP5mZEtRcaL0xG6 FvE3tTh1iS0/6SWNCVZs/r3xZ5p+vsvKGNOAtjN1C66OkCZjKysTq1HXiv/wU9qanLuYbBIKPtnz +ikHw0K93bWea0EzQ+d5P2uCRizt1rMUuMnhiF5szrqmNCeNg3jDFJETRoyAJbZBXzsdz+ZsbIzi IL/7Hp+PVnRkuY6tmjjRyKBykR43dVGtMeX7C/beOgWXEk8YvY+q/5nJlsytVidROLlZ4Xb2k30N CmVuZHN0cmVhbQ1lbmRvYmoNMTA2IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggNzcxL0Zp bHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lk dGggNzA0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDcwND4+L0hlaWdodCAxODYvVHlwZS9Y T2JqZWN0Pj5zdHJlYW0NCpWieJ/8MqxLgOxKz0TXznWf1fzYCShnCU6O1RqWOsVTQDDfCstRJexb eXHFhxDrqSB+cOaXuw5xP99I9NzHeur/v3h3x18gUwBdXIBVSpbkoC0LSfl3GCEtWAI1Ec5HlIjB toj8fFg1Ud8CwUtOWbaB6YLsOjsoAIpUBpxBXGBPCHpewLz0IvBMbEcG6LodQLQfhoMISc2TTCPq 7kj+NeYrPU8Q3c9zkjIyXZZWie2BK7DTAFXb9ksst84oRl1SGoLY58XDxceEptVECMKL8l+mXkt9 Mhu3c+aM5W3qpnAPHF6/dfB7+Zugv77CwIRFDDlAEEQ2QNfr58TxifJwyjI25fkliH+jrnCXs5jy WCfTwee43Z/QApQz3SCNNVYTtZfPKtlnrha/jCvcbTcglu++dedItZUIEu1Pazj36HEyqeZBfQDq c/yKbCejnmd3iVC+m1mULaqc9a6mn37NkDpmttjFUX3FGOmn0NrKclKcYW98YzV3OW530/TgOmfg LVd/3t+dyubmNLBeafB8KMD35sobEXmu7LLt0FON0auacO3vXeeFbDWEJrsQKZo0qRijmWQ+IE2K +5SbsiC17subPmUPzRWwESHJHVHUMMZeG4lKzmtKjDrt4MRhCOLQUFTi8vLSKnAEdMf0uWGA2Hs7 ak3qU5kCRthh8IjRO2a05LgHLOOvi8/TFgUABtK6zR9GI/xyD1nOUGZlOoGzGvI9XWipT6yY0cAF aXJjQGDCaZNwjd1eR3XXdiJjWm8bKcAMNNIT6lBOS7tI6iql10l0KxY4acPtzOn8qy/F/sixOnZX vxcHuYFgt79ZRYy0YFP9CaDRue89SfyKuPCJHmh82TYytxPnxS1yx05e98uSqbnmiWJm1KM7EIIf imO4ixj+ttE0qrRXQ0z6zzLlrpUBEjXWyh8OgTLxKA0NDvj4cKlROcwsQMiPVdvWWiDaxB4Gode4 aGxjuuHcv0jpnH/1GNmJnloVCkl2cwRV7VS7+uD36WG4k16WLVXvlYIztp0jhQ0KZW5kc3RyZWFt DWVuZG9iag0xMDcgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAzMDIvRmlsdGVyL0ZsYXRl RGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0 L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQr0f3V6WwJC9KBQ4JrkHtflvvp2qvmoMi68 8woVNsysW9RYvUl2BeTIq/Bbo/UMpfpO3GzVXfEpuQrl9XDG92BHvB3VydBnMTzNS0fSMSED/Idw 0yBlwLwN5/f0phihxsINSyw7GbEjm3L8ZufmOZwOC1Z66l7nfMuBq5ZYqymlfHJSwpsglQ0bDh16 oeNSeT2ILCBn2gRKOkNPLyes13KYKVIysPonehKNC66vHxoh43AXXVe1Nv5q8fv2xx1TvabWd11C rX00fg1yJEgAfkkLypYLqBJwLOlTGO+Wz5k2b82e3fVMnejj3Eo9K56kSc4yCfU24cHeckS5K4nd d1M3johPUes7mw+tXIZ6VjSi5tvWUI18xlJzsH6SHU434k2SuBxdmjvw9Xo3VBIVjA0KZW5kc3Ry ZWFtDWVuZG9iag0xMDggMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAxMjQvRmlsdGVyL0ND SVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCA4OC9E ZWNvZGVQYXJtczw8L0sgLTEvQ29sdW1ucyA4OD4+L0hlaWdodCAxODYvVHlwZS9YT2JqZWN0Pj5z dHJlYW0NCsVKUB1QTmNyUwLbAaGDP1YpUQ2wbh6tQp2GZYp98bF+rkcOBny54KJL6QeZlSX0fDyX IfOl3SS+wrz4pkDk/M0ipRtHsV6uOzP0YJmrf2NrObzn7hD/1QlI/jn0F24bnT8Tmy2roWktENVh XEL9W3tjZa2Bljz6ZPr01XANCmVuZHN0cmVhbQ1lbmRvYmoNMTA5IDAgb2JqPDwvU3VidHlwZS9J bWFnZS9MZW5ndGggMzA1L0ZpbHRlci9GbGF0ZURlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29s b3JTcGFjZS9EZXZpY2VHcmF5L1dpZHRoIDM5NC9IZWlnaHQgNi9UeXBlL1hPYmplY3Q+PnN0cmVh bQ0Kh6VpvNkZIBk+Ipp8S+2zSysNH1P22yF4KOVvn1w1ASSNQinJJbFydx9rzlFnrc+PPB+g6dxZ X1FuSi1yL2YxHivAln4FBfR9QhvgYO1aLr7BlcgI82mXjZ20aYG+q4rWMHkm7ed3XMmGQurM+1DH FHnYpUySka9GCIYFt/4/zP/zEn+WAsVUCHhR0N+3wR+srRPwFnTf10Xs8I0nJzJxAoTFBnYKUf3S C2EY4fU4sV18PgavIOI7v/GbtbbSu/g739UCebsEsNGkwpRhe7HELucvSabvFLrtSdKwn6Zl+/BC +yyIyLwsA+JtQpM/hSUX6+yT2XnDTYQryTfTbz/0WX+eUNwEhqjq9CBUzstc4dOLcgGu2AO0JqZy jwDplFjEDEdNe9kOF3H88vMon+5Q3H0NCmVuZHN0cmVhbQ1lbmRvYmoNMTEwIDAgb2JqPDwvU3Vi dHlwZS9JbWFnZS9MZW5ndGggNjQyL0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1 ZS9CaXRzUGVyQ29tcG9uZW50IDEvV2lkdGggNzA0L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5z IDcwND4+L0hlaWdodCAxODYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NChGyqkuBlafpjLI9aBfNpRer e63h/IxYpmE9kuIxyAcAMbph9iJn3PxidpEjhno7AXBhdd/vlq38lGCByECTfVY3BmB1lpoJKAhp 0xpzhg0LCBsFezkHbZdDWTdX+QbULiYBj1x2bjtzDrGVNzH7BuUo5SwcrsKjV8KPF7Lx/PDW33J1 Amt4aZ5GTnLD7ZC7uueZOApTrda2h13YOfGWen3xe51tS33r2EgENzJePTt7d/TjHXLPMIr610Wk ZgdmpjIDERnMnl1dX4nVF1umJ5P1EQsW4kmF2CiM4PTRByrmV/rt12t3FnlmzFwIM2SZBHtwZgYG gQEJYYvF01EU53e0+r527cvv24EMlAV1Lqb4P5+zt1ntsCKhXsWv7X5oTbx4aVALNv3wk4n25ge4 B3xZ2MR5EL13ttzosJJJ6v+Nb9KgHVLGXQnwJk6G9hT2k4rKlguBU2o/IT7go5Ny5XXbQc81Ls+j UwgCBgN9+ePojkPaOIcAvTX8TzPTxr57WcXSeA4FPTETSIyiKVg5/ZX+lkyAlGht6FCirYiqp/WG j4HzLqtUqU+4Y7q77D7HFDX4186tUkzazzt96XXJ4ogsUwaZCLUcXRtbXIkchy6OaLlgwiS0gqhc w3UBBrU9QnlIZAOCh86FRjmLdWmXzOQMI9PJfK9JqVD3ziriNP462485Opwd0Pb6v7kG1EU/TDdg sxU1Y1DrFqS/cG7CUpA3lRgaB/kERRuq3rLb52BahPjqzS4nqlHraMEI00WKRAwPy362nCd9Yr/w EGpYg6vb/M0mCP/IeXAyTWqGvPtB5enjNQzkzqzcdM4+JrF5x4riOr6YxPop8qUpZcLt1Du5CQ0K ZW5kc3RyZWFtDWVuZG9iag0xMTEgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA0NjEvRmls dGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkv V2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQr8Pc8HnM88TMA7Mphnm7Iw tpYgFY9CHucZNvEz+eROjpRCOWjhe7lo4YvecZN+XbaHBd6e1RdsfNhGBvSKILF7W2TBOdoOQjaD TQtje/2E6h8ujw+nqrCK+/buWh1Nr2Kb962uN1QJeKWs1x7HwBsooLPeFdbMypmWAnlwUrFBCYM2 UFwxbzrllCf8gmDJWe8ZX/AYZX3pGQxioUV7V24ikErv5jJX3yAlgwswIP6Vj0I4cOnLzNF7Uw22 fDcIWbjG57H7SMKrSQlnxTDM4j7MsCfoD9YIyS7+w5qivfPJMp8gD00FkorFDxQ/B2sMRtPm7pJR CrseYjYHORAkRxSPeXzWIfNBV57p9Dd8Oj3xwdpCgFUf8vTWbRaz5V5juoJA/73PNUeJ7nSsa+9u Ew7Sr0u+HMBk8OZ1W6PBopi3chWi0G67w0jpeQ+uupilOYhP43I6BRUD82krl3ZjjdrfNpyJ64N+ AsGMvKeQ4HvU0YrwmzD2+/6Ks/3adR8Y89HDoGtGDdbxrDmeAtbehL4JxRf6vu80k2yMZmof3xNk n1KSicDgTa4WuY89ubjogyuuVOoXCYl8oJiUZ5cRdp17EbQJwOPSEKMXX3LttA0KZW5kc3RyZWFt DWVuZG9iag0xMTIgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAxNjEvRmlsdGVyL0NDSVRU RmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0JpdHNQZXJDb21wb25lbnQgMS9XaWR0aCAxMjgvRGVj b2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMTI4Pj4vSGVpZ2h0IDE4Ni9UeXBlL1hPYmplY3Q+PnN0 cmVhbQ0K+xfcgAVABnBpwR1pOoG8uuoPswhze8AFSOyaHt7e17oulgxwNX6svmsXvib+O6oy3Sh7 /PvXndrJqoOrOFlUicyxR44mYRYPP2CBszHU135pz2hEqPux3MKs2ap38rpFZqsaIJmTrTi4F1+H aeyc18Rn5ka2U0AYIhsCE8tNBfy3x7fogNZVL/3KDx8EbveNriAQv6TtKjn0Bztcds/6R9ANCmVu ZHN0cmVhbQ1lbmRvYmoNMTEzIDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMzE5L0ZpbHRl ci9GbGF0ZURlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9EZXZpY2VHcmF5L1dp ZHRoIDM5NC9IZWlnaHQgNi9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0Kp24iLepPgRs9YOU4SiH6sO+K L/tF+AiH5Xpkc6sIb3kCIbC/6ts0D32CLE5WS8u2US1gX7JGIhKJUCrm+2cpFJTswoFwUz3k9CJg sEgP8955k4uorVlBKExA7dK9N/vQ3JlnBrFBnznf+tK8nqD67ezRfIecGi4GERBfyWKqd9I57LwT sHi17ugl9xMgzGuHiMsYq+I3van4wOngzdZswKcf+Sq3WtTx5wS3yt59hSO9g1NsAhlf69UfjUtX 6Y0GB3/6f3kWw570I4+Zi9BTYToDmyztHoNjSak8qOS12Yy/elANsm/ipXQ1uipmrr/5NnTz8GG5 7glUj0zd4B47+WEiix/M6vZDVpDiRyaIdq4iKk3wD15H0bndcbpd/ikHu8vSwXVwLqj9Fl4d2gb8 3KI+HoLTrsZt6vkbkjkerQ0KZW5kc3RyZWFtDWVuZG9iag0xMTQgMCBvYmo8PC9TdWJ0eXBlL0lt YWdlL0xlbmd0aCAyNDAvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xv clNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFt DQo4ycaQPfCSrj1MUqv9oC0P0QFjh0obp5q+wBW7M8ah2yixTzc2DlSv2fVf9WH3t6BJX2nHcOyw 1YnZ9WMQjcR7txQgiRK90EogYRVmVlVE4+HEk0qgEBI5DZzSlnItcwkcD5Bw8Cit1T4SYZrY0xFN 5PojRrofF7T/b+JTIzXb3VNh2z8Q5NcisT+/dXAQ3G2BPGqu5KMb/jcnvdZmrGudbgIrDl7O+G64 4utTH6xZckM0f968rNBCgMJtyJdM8AncYFYLX1yCTTtnFJ1pYPQEGosiQL+ggR03e4ruszhh5x/P SZ796XxBaUnqlt/nVKMNCmVuZHN0cmVhbQ1lbmRvYmoNMTE1IDAgb2JqPDwvU3VidHlwZS9JbWFn ZS9MZW5ndGggMjkvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNw YWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQpc dr4RUnhKLKOv8Q0yX63nUsZoCJcDpInzlVXmBw0KZW5kc3RyZWFtDWVuZG9iag0xMTYgMCBvYmo8 PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA0OC9GaWx0ZXIvRmxhdGVEZWNvZGUvQml0c1BlckNvbXBv bmVudCA4L0NvbG9yU3BhY2UvRGV2aWNlR3JheS9XaWR0aCAzOTQvSGVpZ2h0IDYvVHlwZS9YT2Jq ZWN0Pj5zdHJlYW0NCmHVDcmkLJO/SGOc38K68aaA7uJ7mWamXEMbknW4Kpm/mALdQmYgB8wWVlt8 iSW6Hw0KZW5kc3RyZWFtDWVuZG9iag0xMTcgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAx NDkvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNwYWNlL0Rldmlj ZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQq+DJHJlkpT4Adb dnrbgODHNEMSjAXve5XVdbDloSjrui+FLGPIrEkxkvttUJ0unsNhFmg5Sfo+VqnhsFqavKrmOdmE robo35IjdftEp6HIuxwxPIF+cutedIwHFvlnRhwvx/KA39JHdZDWS+Owfk/rxDGKkJ1/eFP6f96b cR5T+Z/o8/pMoTosaZtRL+4htM3TpF2s5w0KZW5kc3RyZWFtDWVuZG9iag0xMTggMCBvYmo8PC9T dWJ0eXBlL0ltYWdlL0xlbmd0aCAyMDcvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25l bnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVj dD4+c3RyZWFtDQroJCOyBGLLoH38u8lMv46HKIDK3/b8Qw5/QoNGQGbBMLkq+dsY3B1j2BXhP+zM IN4IyRnE4aBigP3mVjohOskNiHdokpVEIDpkGBEit9AQh0DkEvFdL4K5vxtwZZ7iBvA3kswDkNio CIBzMCCKk6v1d5Z1S5QX/8jJ7w4PV2x+/LK6diBqd6Nt6v1Pbj89Lb/lvDUrXKSSk6Blxuw+KtY6 IOkvxcjUGu82MFf3CD1JgMwrQFS4/WBGnJsgwn/E2MKU1ycMR6n+roZYpLpj0ooNCmVuZHN0cmVh bQ1lbmRvYmoNMTE5IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMTg0L0ZpbHRlci9GbGF0 ZURlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9EZXZpY2VHcmF5L1dpZHRoIDM5 NC9IZWlnaHQgNi9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KXuss2l9Uco2b9oDwSZmoVJ4dHbEMB+L9 tqQqTF5JiE4E9q309VXwY9d6y3MyUu5/EqbySsNATo34oXQbMPd1c409BS6KXWWmFK49e5SLc+s3 jhFdY794i4w47w/43B3C7tXpjG8bMnBCcWCf2Ph4lqQ/1iWKU6bKYZrjH6aCODXsiMRgkHItQqsM +cCV2AiZf6iPiB198NOBLdgPhCC8H7rW1Rkdhv5Iwrh63b51NTXSxItCNxikZA0KZW5kc3RyZWFt DWVuZG9iag0xMjAgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA0MDAvRmlsdGVyL0ZsYXRl RGVjb2RlL0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0 L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQqwmRYamuW7hsN1Q4nDEeJjVMxc3/dgQP6H fDUF7+MvASX4lJORjhIC7APSb766LL4wSPQwcPuIEQKn/dt22RpKYAUrmexGNqSyKlE6hoUT6BGY fgax8EBkfRP20KoqFdRVUs5ax5v8tQBRehuZ8I1gSEog1cXkiIFqCIvLvVRC3WhmON0+A9TABqGL cBxgMQrpSLPsEIOZ89ZHIAjFtcL2WWVUqkVdjjJiZFPvUmSVAS6DjT+AvDLP3gkECx3d7BEDs83A DDP/srCx/XcnGlleoc1TviFEq2FB8pZVnoaIb3WiCfH0tPQu5/AWVEhvlL85Ym7R9UScAR+wz3gi gG3AvzjltxRVKgAwC2syCtDVE7ekG5M10BvZPVB7D7MdciE7+xs45QeoKQnRlafZzf5yzosAmNAK VSJvmhbn2d6b9od/v/2HH5YbZ/kZxZpHsvCD2UVs1WQjnbXcGhgjQjhRi3GCLOYn3ufdjCq/Lza/ PMfpLZt/IEooASlTw3hdoLNPvMipylTPT8MREzCKgYhrDQplbmRzdHJlYW0NZW5kb2JqDTEyMSAw IG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDQ0My9GaWx0ZXIvRmxhdGVEZWNvZGUvQml0c1Bl ckNvbXBvbmVudCA4L0NvbG9yU3BhY2UvRGV2aWNlR3JheS9XaWR0aCAzOTQvSGVpZ2h0IDYvVHlw ZS9YT2JqZWN0Pj5zdHJlYW0NCrkcpG30FJABjrksjPskGaTOCOcQjQ+MUoTEVJKLPza2+fbSXrZo m3YPi3irwSFpRoNHJbRiYv+/Ld6ORui2JCAnOFx4Pkw7qqVGhHSEN7EW7g7mH9e6ayR36U8qIfwP AmqZdTWkAEWHNc94IIwrTaV/hCJj0QA/xTvCuzg53SSXjC6KUA4QtmSUS/IcPC/iLvu7diCew+A6 ZgHBt6V+A0yn74Ik8ndWq3r/kQbnHHTPQqPwkcZ35UqgyrmMXRLySCo1sBfyhPFlaYH1zM8SS8Zd ZZ5t7pXcvCS1pWIjiCkRkhm19oMmrM0cw3w5wfZcJ2S8tmVz/ZzVgNAkmN8r8VmIzp9/+AkcwouK 8w/XgjIY0yJZDmc1DJ5C4eU7jK79Uf/4WSzFqj8O+iz5Jxerxhr6sX0xjcXt51vbLhZOtuixd/TL yMj1rqmbv+W/IWfMUYjoGtkxPFu/1jxs3HNogwgxOkyfImcJnqdSaflxBR05mCo/BIKj3JALAREd PVPUsZWdIuBQMJivxugyBoukqiuuL1P95cpFbr93TqZR4LvLSWyrPgtmwuo9Oe7OQRCrB8ZNjY7y n7JaCj/YDQplbmRzdHJlYW0NZW5kb2JqDTEyMiAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3Ro IDY2OS9GaWx0ZXIvRmxhdGVEZWNvZGUvQml0c1BlckNvbXBvbmVudCA4L0NvbG9yU3BhY2UvRGV2 aWNlR3JheS9XaWR0aCAzOTQvSGVpZ2h0IDYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCuH9xkwmvvpH k7TcDCXx8Syh9m+bkClr7MtcTeMRToE/tZWq1pi0vvVw2hkz3zwmvxHv3+m9qqHtt6KsT4wUtP12 85Xct+Rpgcb9i9H2VLvr7M3XNoxKfr/7Z3rEmN6GxnXO2q1JK+hQgKC3HNSWvt2SBCJ2E7a1ZazH 62b954RuD15Ogf6XV/CXUM0sn3K6ZSz//UvIoqz5c54jvJjRh808pfr5f1kd4uVO1IGIqR1k7kVP TYgMXxkODwvz3jmmDlXZFWNsuUh1EIB8de1MVbYbkx2NjTKchA7nIiwNiWgB1MjR+//PzoIHyJPR rnwbN0wUBKY04RYJBiWUurxrJsODwadn6Pw+3NkU33DL1XpWA+kll3EbYS6fPCLQ5hwlFCsK1a3u TdOIjKNrIufzftimGLlRMGNcy0ouuswYvztM+nq53MtF9CweoxSX/xyrwV9Ua+cIxjhk4/B4dI85 2oIdL+5L2DOcSHoHgoGeYRbBnS+2IsfWhka6czl+0C1wApM/53DrBGGZhtBTqOGb7LwBpLP0kheS aJxaR1b+6MN3SLbBylco/2b9sGNVpc65bek1WOpQMw8gd/kVTtUemxKltOZ2BDp/bD+5O/RzJdux lW0GZ1lU3vdIXM41r+mDbxxi31SEgoPNVSm198CH4z3h8v1Hpo0MIVRYDIOsq0dX1PfjBKG8qdCI B0kmo6kVkiAuHSqXJSG/9dwMREei4ZFDMKgQdtI0xdBTs51MxYnkgJqXHCI4eb61W65LVtRvxZ0T cQfbP+28dw1y40ReJnGEBoODeNT+i0KhWRJfGhr8qTN4CcTMSjdsAi838jut6WiWH/b12rrZAWaw 8GpFEnmDPzIZYrqn2J1m4LSxu8XF+TwRg8I+559Va3NOlA0KZW5kc3RyZWFtDWVuZG9iag0xMjMg MCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA3NDYvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQ ZXJDb21wb25lbnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5 cGUvWE9iamVjdD4+c3RyZWFtDQpJQFIPptrzhBhPdS+djmXzVM5ri2T/F4b5zqQt0IOBHrxLjr4m JaGtXjNk1OY4yhNPWx/4skv4H1mu/0tT/yB3wHY9vnrk8Xj/hkhb7QipLi8AS48YZuPOj/42jUsv pDg/ZSygkaZxenlsC/qWUVQ0yoOPa7oQZGbRKW5azEjEmIxMmCxhm28F1VHZ518cNeQzZiLq0Olw XX+vq8txN4fNWNCRyZincCsdPxohzuVK7HbKZ/Jru1ni9Dwtvn/vQERZyWIY9BfUuE3sZoFDVFYn uFalLMZriwbsxTx/NlHUURscGG9peFv+ILq5T/iETFJ4/PPeLHBIuybhk911ogE+1mqb01ZWVve5 GXwuKJyjPumC1srfAvmWVbBgrdeaCKZ7P46hUKpphSrSOOzBQplZ+RjhkWmDECPU55zmBdashd/R uAM1vGQ7fMAPc6D+sTyhP28bKOGWC4jFK2m08NyqV96PPz9AyMe+AaXQUmilqnMivBvkyptILPrb UpiT1HTBVeCyj4NgPQyjKHvQdf30x9a24hFjCmfsVtCy2jLusXCZ+dcZi2H2fzn/Gv7gUXNcNrNv PC+CSX40WWJ2+n6uSCNu7F3DRRglsCqTHoYWy3aykm0QblvV4oO9WF2gUhCc5LXD18sKJkuBvExP 6ATvB3ctEyrqVYN2T4QukzQr/B50WuCbp+UBijp9ja6IMkF03RaUQ621lZpVpMO2cmpEqGlWu3+3 N9GeXvHkg2CLvbxnpYqtUq+OVrzyJkwkOl62Nd1oV0ZORDbIm3oZDFsobUaBGtbbCWX0pK1kMz1/ kDe3jGEjDGDKiq5cIaklkq1NWqgLtONastZ30L3bGMrczsWHPnR8aGP8M9T/9a2xJe84dywQUnLC zLjjG1QOanbi9Oixy+Oxt+SlWNBbcPICpvaNIvDs0zi5+/ic4XKwNZsRzAvSKwj7drK7qzYAerQe 7NogD1Wcvd8wkIcIW15rrjqXVE+qX/C3Dw0KZW5kc3RyZWFtDWVuZG9iag0xMjQgMCBvYmo8PC9T dWJ0eXBlL0ltYWdlL0xlbmd0aCAzOTYvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25l bnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVj dD4+c3RyZWFtDQpsOnAbscBDZ4kxvQcETDs1lB7VA17Ob5gxUj2dwjeJ7BMNWPoH1FaQxbJbXPoK XnVN3K2q+Yh8nKJ8rYpwju2haZDF6tEaw51kzq3mUAP1DUxrD4t9atwkJDOqV/sdBFHsgs2Wxb2D bh3JgyIxRklhcCwcskmOoDQ7mUc8BXaPgRQPVMncsoNCLug9DjCz9zfHUG1STiEANmN6iGTQTjl8 6q58232EtszFvVZfdR9T7Aa8aKIUp1eHpREoZqgd3eGLuVm9GJalms5t88nRTC8VvqszXJ+Xfu3C 0GKPWTCu37JQxFX1uN8mmC0UqPp/8TxzaLk4mhgMS1z5exYiDMmFvWA7SrZYsFefAaXWHBXQuQDd Ne3oJfeVu5wnxsPt+Dh/ZYytaa5vMkg9SEBEpXw1Q0hUrK7MmpJPgB+qmkc2OZcIr19IG/76hGFx V7pUpiPUmH3qV7ZpBL296+WdzefxkLd63DnN/dlT88J4x1RuRg+mgjTGqFvagbrBDvdg7ztwPZWf gcwBDltpCk0NCmVuZHN0cmVhbQ1lbmRvYmoNMTI1IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5n dGggNzM1L0ZpbHRlci9DQ0lUVEZheERlY29kZS9JbWFnZU1hc2sgdHJ1ZS9CaXRzUGVyQ29tcG9u ZW50IDEvV2lkdGggNjU2L0RlY29kZVBhcm1zPDwvSyAtMS9Db2x1bW5zIDY1Nj4+L0hlaWdodCAx OTgvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCgo22hIeN+vFa9j6nzLRWg4PqnooJW2WjegRdfw8pebq jb23GOIgOT40TI7otzDlWdKtQZzdm7IM/XvCFbNdfeqYCC/yhAYActgZ7qe/XZmGpzvmZ6mIKuIJ 4RnbuWMxA+GSJje7AWzbBcNow74vIQDeuVQC2LyF4n3+UZpNod5/2oerAoEJjfmv9qZc8wN0X+fe DhoVR7dUK9Uyq0ibfWy7WsrsK0mfQ7W7+GUKM3ec6rfATtINIPVUQNBicKlYprX8Q0lNc7TZPhoX BaYVEetOdsFEr1078xkJdPmvzySAxUt8rYzLyRJz4LCB5G+/o29wUZpt+EvgiMxaTNZvWX1GUNEB C7/dGZZPzLtVVH48YEV1929fOPX41I5MIQLn4Yz8gCF8nYEc+v63iAxciMmjALjLNIjDVVrV26be Jxe+uBTRGlVkTZYsXJbsMpH9YYBSec/CVUkC2VUFbUuGP9h205HWT77qjjQsP/iycSdIq7jGhSOe JwrJ0Mk++N3DyVurbkz1eJykkTnQKjLo3gNeAAXwTJ9zXbJ5O2AwSTKf6dluw1OS7i1YPEKqjefS LhW+eaLbsuqCwkr3Mutd7Wr2vLc1GbFKod2pHzp1tQrsOClCLc+YHeP3oHzE/SIfujmYXZFWiTmB VmoCipftQ0LwFcbtDGedP1bmHLfyxnKPRNBh4/qqaMcKaF6/Z8sNV2XTl3zqmb9uOta4wWT9COgX FxOZ/OPqqAdJiXc5ZDVa000FrDWdLzXpScrSmG7xL7CbiRG+f9Q0Q6sFw+Aus66LaHXOo37VIoQZ 9rwJ8kyL4FVBL1qq7I8Kjq3jMGKrE+7SWxmEY+CdHSDLwIV0fuEWtrMCATjJJKWtFBhh6OYS71oC T9sKuEEP4n10ggEcXVQbQJR6KqTkQs0FEmVATcaFpnCf/NVAlqD9RCDH8UxnptErXmcxL9tMUyDD x5MyhYibcYspEmg5NF99rFHqsA0KZW5kc3RyZWFtDWVuZG9iag0xMjYgMCBvYmo8PC9TdWJ0eXBl L0ltYWdlL0xlbmd0aCAyNDUvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0Jp dHNQZXJDb21wb25lbnQgMS9XaWR0aCAyMjQvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgMjI0 Pj4vSGVpZ2h0IDE5OC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0K+HoS8LC01hm2PENmAVWquRf4p0kS vdvuGGK87Hz4zdo70UYrOUVlV2dXQGUfc3vxiPVsDyW0qDtU4fNcJK1tbCTNbrz2WJ+itnxdXACK JtZZAmwe8Hs+ewj1Z3/AC96VNVkOaeah8bOo1ssmuk1OJhTpzDuebtnWlOU1u1GXNSp31VnKp5yp HcokrbSBliwl8oxLBXTj/MM+YLCtxtGY/Y8eCofi5WJoBhE2RSWZ/zk+EC3eyFJMmuEFd2ff05CZ iJDwzIeLZIr+MfAXx3nq3TTCKGACdm2y+XrBCRSDDvcbvXmHMHryjuGgzXDws+/IIpb7PkkNCmVu ZHN0cmVhbQ1lbmRvYmoNMTI3IDAgb2JqPDwvU3VidHlwZS9JbWFnZS9MZW5ndGggMjI2L0ZpbHRl ci9GbGF0ZURlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9EZXZpY2VHcmF5L1dp ZHRoIDM5NC9IZWlnaHQgNi9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0Kiun1RbLZtbN6zKfk3wiOk/CT RKmTUPk12f1p+iCB4m0WVUGWXP77GH8NtUkv/tGQ5+Jmp+imRt2bJnD24SQmx+1bv7QjQbAVm28W 88uuJabk5WsjzZKdBhqrMuSJiZT8LERYOtREyy6YIh/QQ2OYYRmB6jIjl2kkF+8vhgeYKz722kWZ nZJGZZZm2ncpR4DoMuOeMnSKKnbEONaIxukdhwhOcAiPL70uYIEaIVitJOJZETSxVQ9l5pUTVJfA oUNBUECkE7Akrl/jcdA24n2KY0FS3IGNC8Sy1Te/zMuz+Cj0GA0KZW5kc3RyZWFtDWVuZG9iag0x MjggMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAyNTYvRmlsdGVyL0ZsYXRlRGVjb2RlL0Jp dHNQZXJDb21wb25lbnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2 L1R5cGUvWE9iamVjdD4+c3RyZWFtDQp7KLd5WLQf9MSjx/bpnqcIL4DJKI81tCAK98lap5W10avG GnkKL6zbfTsRZbpU1Dte8OCXYR12ZE5LNdDOgx8nDtVsHG53cSHa1YPi0STm1IpjkRjRm2FoSDS/ bt3y0qcnK3MLggaaRNsTdddE4bTgE+r6mz3m0vE+NZmNI75Sd8fD88rKwgz+rOF/NO7C/xlVQ1Xv /wET4JuwV1zRkbEkjlfLn+BShlQMk4GFNigEWFZuidAfDYmJ9eRqfWU+MyFMH7/hHeKiC8tk6k8e rFS/msmekIRKyDRGBOGPgvAOnidhH0sTGq1CUoxQE5bguDRDsfLxSAEntby5vbWFuVUPDQplbmRz dHJlYW0NZW5kb2JqDTEyOSAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDIzMS9GaWx0ZXIv RmxhdGVEZWNvZGUvQml0c1BlckNvbXBvbmVudCA4L0NvbG9yU3BhY2UvRGV2aWNlR3JheS9XaWR0 aCAzOTQvSGVpZ2h0IDYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCimTNuqJgK753Mnm14Ui5OpRE9fi DC2KOkNPEyg0MqQAAWAlrggZ7zbpX35hyaxTETr3xeUWMMd50kUCF/juNmwxY+QTJhaiwVgzT4zV ptGpMNw3rca8viCed7yVFLuBwDxNsOwTFPyMeDA2x3aRaL/SeUs6kXQGp/CKHN4ppYuWNAf6Dv24 8cIHHX4GJCGHGKkXOWV2gae4BFGvjnkcazg3+nElrQ72ljyCZHKy3A00ZP5VM2MdckPqjXNybXSe d6jOyxKbOEHOq/hgNXYS1IAAQxp2pftTz4i+Ru3YMXdWUSx5D0FLrw0KZW5kc3RyZWFtDWVuZG9i ag0xMzAgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCAyMjIvRmlsdGVyL0ZsYXRlRGVjb2Rl L0JpdHNQZXJDb21wb25lbnQgOC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdo dCA2L1R5cGUvWE9iamVjdD4+c3RyZWFtDQrx/Jxr4R+tI/EPT8uFq7/lSh5a+bqSiOCDSa9iEstg jHy+4uR/trIl4Hd5fJJ91dM5emdp5pKhAG44OQEmr57sW8N2ShoXDqbHbah+i8fxJPQ0L5/NPLIf 3ysYO7PxcyD3MD3GPGnEsOtPA5CJ0PyPd4Vki5PDQ7z17pqhRgG1vKAj+BtvHbQF5xm6dVVh8ZGU JWbVgILwvIiAEuTZDk0WeK/OJJRMnNEWQod5zmGOTA5zmBA89SdwkF6LbZutOcn/j0y8Sdd0XEWT +jrgvJiSgVPftBcEsgTYpOI6VnoNCmVuZHN0cmVhbQ1lbmRvYmoNMTMxIDAgb2JqPDwvU3VidHlw ZS9JbWFnZS9MZW5ndGggMjIyL0ZpbHRlci9GbGF0ZURlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgv Q29sb3JTcGFjZS9EZXZpY2VHcmF5L1dpZHRoIDM5NC9IZWlnaHQgNi9UeXBlL1hPYmplY3Q+PnN0 cmVhbQ0KCut9qqOIK3pp+GkW07xSEvXGHBL67TnE95TiDENUl2LJPQXtPWWujBXxUTsh6+QsYeOc ymoa5k48bZjUNYaW6ip0yuajyKC7Iz2LVTBiMlwHKQLrx5C3obD7PbV+vLupFisvN1Qwk3Lwarpt DPpevCv8WDgkT3fbkcMxmr/HOG14boaLd0Q9lUMt7aobH3FyTPdieyKO77hn6EmaeOZfoOEHpOUp P8TK4K/YmhUrHqDcbykHltayu8rDCRgzRXEp/UkmoEa3AaZ7mBYxffR1kfkw4hixZ7Zt8zkwkfye DQplbmRzdHJlYW0NZW5kb2JqDTEzMiAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVuZ3RoIDIwNy9G aWx0ZXIvRmxhdGVEZWNvZGUvQml0c1BlckNvbXBvbmVudCA4L0NvbG9yU3BhY2UvRGV2aWNlR3Jh eS9XaWR0aCAzOTQvSGVpZ2h0IDYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCtG+7zmjNn2GnrfIzfM8 sUSwqa+smQ+HBEikiXobnfm6vj15hbDZOCCX2hkitMN9vmzhVkZVprNoqNeQNvp6qhs77nt1lV4m 1qo7Evjg0f1TGIbC5yTUyhKi/RPmZXMkfP8hSJ6khMKP8KJAPcKOvcKgXbpFYmkA5Md9PrfPwSvg EeyTibxtSXZkYvUvTiES2WsV5UpdsKBnUgsS9RjMoftfVvJ/urSwpBbuy/257zNBpSJQObhxrTzm Sq5twdKlfh+PvPW+yaFsDpHXOXR7BA0KZW5kc3RyZWFtDWVuZG9iag0xMzMgMCBvYmo8PC9TdWJ0 eXBlL0ltYWdlL0xlbmd0aCAxMTUvRmlsdGVyL0ZsYXRlRGVjb2RlL0JpdHNQZXJDb21wb25lbnQg OC9Db2xvclNwYWNlL0RldmljZUdyYXkvV2lkdGggMzk0L0hlaWdodCA2L1R5cGUvWE9iamVjdD4+ c3RyZWFtDQrMIWXWZ4ddR7/v/GBjM/GPQv3UXvhYnpdBf4dK1ZwuvoPOiDljny4oSEiy0X9IlgsR nnwN5eSL6N2VRFB0wEi3XP4w8Wr37tjtqwF241KlEohNhfldZ8mWZXfDZUYbMHW6KOORNZxHlLfq ta2BH92qcVoiDQplbmRzdHJlYW0NZW5kb2JqDTEzNCAwIG9iajw8L1N1YnR5cGUvSW1hZ2UvTGVu Z3RoIDI1L0ZpbHRlci9GbGF0ZURlY29kZS9CaXRzUGVyQ29tcG9uZW50IDgvQ29sb3JTcGFjZS9E ZXZpY2VHcmF5L1dpZHRoIDM5NC9IZWlnaHQgNC9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0Kd1JNmZ3j qY4GtiN+KmKm2qoPgQWBK4xiMQ0KZW5kc3RyZWFtDWVuZG9iag0xMzUgMCBvYmo8PC9TdWJ0eXBl L0ltYWdlL0xlbmd0aCA1MzgvRmlsdGVyL0NDSVRURmF4RGVjb2RlL0ltYWdlTWFzayB0cnVlL0Jp dHNQZXJDb21wb25lbnQgMS9XaWR0aCA1MjgvRGVjb2RlUGFybXM8PC9LIC0xL0NvbHVtbnMgNTI4 Pj4vSGVpZ2h0IDI0Ni9UeXBlL1hPYmplY3Q+PnN0cmVhbQ0KUMU/rUB6k3AuwlkV9V100JoCPJep 8x1m4eQoyR0jqZY1RLKey1hhKM0q4j2gWcxzP0iEimf6p1QPbfKJI7tbtm9GGqImOAP2tlj1yML+ jxjBhpPHAcDZBnVj0bgAKUrtKrTFyjAUsUBILGRl6bWwQV0/io2ZMNr57ezsN0bN1Vr3g7wN74PA LCdpDlRxd+7RwxBPg9rDEnXhQvZzVI7a0/MLHe1T8AOk/2HMaaOY30gKCGUs7tdY1A+aLI6Ho5I0 52oL0t6RGvxg92mt506mv4r6kkgXIcyOH68FmtNDDEDzavgEs7Xl0EPAk+8YyQrfRPMx1NZ7EekH X/TljzahvwT38x6I9mzXMnzN3TFoS5LTq8dIomTkDNZRR8Y36IdD4JPXhZ+q1iX4al31AZ94hBXI oiP8EZnY3KoriH7NxyKA4mAPsiInEePWcrJLtfgKsHpFcfaQE/Ca759prnmNLamF9ItruqDVBbRw MRQKo4OtCL6X5OGRJyyPl4fWmUcC4K+vqV7AjMzioHu5d8T8LRvnZhEQNtc9rvCUgozE6Twi6xkF Gz6R6Blq2rgTfwu4v+t92It4JpnHvIceDrlnAX4GWj8oNdm5hoEuSGTQ+vrzuQLV6k9eFQ/lo/lz E3WukJEc28gTx2XbGGTmnkyMSM2EVl+0CRmvN2pfjamCWT+igc8PQhg4zlBuboG36hH8Mf0iqMkb 4MSSRA0KZW5kc3RyZWFtDWVuZG9iag0xMzYgMCBvYmo8PC9TdWJ0eXBlL0ltYWdlL0xlbmd0aCA0 My9GaWx0ZXIvRmxhdGVEZWNvZGUvSW1hZ2VNYXNrIHRydWUvQml0c1BlckNvbXBvbmVudCAxL1dp ZHRoIDE2L0hlaWdodCAyNDYvVHlwZS9YT2JqZWN0Pj5zdHJlYW0NCtRuItWcq3YV/2teCRrrlxos RzbBGucA1FISbbZa6bxYH9O/u53AiZ7Txg4NCmVuZHN0cmVhbQ1lbmRvYmoNMTM3IDAgb2JqPDwv Q3JvcEJveFswIDAgNjEyIDc5Ml0vUGFyZW50IDE0OCAwIFIvU3RydWN0UGFyZW50cyA0L0NvbnRl bnRzIDEzOCAwIFIvUm90YXRlIDAvTWVkaWFCb3hbMCAwIDYxMiA3OTJdL1Jlc291cmNlczw8L0Zv bnQ8PC9UVDAgMzc1IDAgUi9UVDEgMzc2IDAgUi9UVDIgMzc0IDAgUi9UVDMgMzczIDAgUj4+L1By b2NTZXRbL1BERi9UZXh0XS9Qcm9wZXJ0aWVzPDwvTUMwIDM4OSAwIFI+Pi9FeHRHU3RhdGU8PC9H UzAgMzgxIDAgUj4+Pj4vVHlwZS9QYWdlPj4NZW5kb2JqDTEzOCAwIG9iajw8L0xlbmd0aCAxNTcx OS9GaWx0ZXIvRmxhdGVEZWNvZGU+PnN0cmVhbQ0KhhR613iuMmYF2g94jW2y4IQoL2yYDuDMSz9/ FDNwudfFhjRHKm9jwlRHAu5lZYLXyLUm59vzQ+h/gi1elvhJkFuPeuC9KcpMWyTcFRIZB79eo2uk kNi2Fi7kwZyq3gH2/vNkvu1lPQUKgZ2w/EUGcW8RRo2ILEAt+RWELQCn6bdQPEFPUvNjguzZQnoP 0AUEHAGArnhPBOTulLKVMbEcR9SHSI5oFTVLJA+OeuyLj7mL5k7AQSt15gLtcv0uv9NIEQRTp2Y6 5CSTycf9q+cvbC7jo48u7wF3FXFL578yui5j6Ogdfy1UzC06DKUGUWPL20WE2kUcHOqC9Wc7ZZJO JFjBxYID+xtLVswr5IHxpLBjbqO1o5l2/9KVihwTuYdzCoQRn88zrAPkDFIMIdhjFkCIm2jyjM8K QV7ZE8E7M+6zPh+l6kvy5JvP1G8QwqoWhk7YkzwbSUA3YfluWqNPZAxWz7i7yrYLnQeilwyjAkoR 2jKzlwXkd7pFVn20KeHjSoSxkZJ4pYnMSXZ9vKCrtj/ME7ydYz6CQIu8rJNYq6j4C+CrB9tVhehO u+6Et79FpbekIhsW+bSh9W76zkEah2UQ10c7fvCGfXACv39uI9UeFZ0M2JltlKy/u8vtMddf9Nxs a1EtBi5pog9OwI8uva7YlutjI43oxpWNb7ddApQpub+YpjkW6nEXdUqo8ZhrhFficeVcMPNYvPAB aw4qK9CtmavhthEbPOASyPmh37grPeoAxBIn3GdzF8SMMjVPNLFsl9pguWo+4UplkT96tlnprdwD a+VGnMNy+BhjIBr3lsIy4DaTLne922RbmWtcIfd8qUT0dMBklalRsj+rgB/OWuE2QOxPXfSckLOf sd4HfGPaDM31rBj+RCY/fCqsj/Llk2dBBJ7rZV14SpBs+QfDjhoi/M1DsP8gN60REFcYk78Nc06d ittRTjAjUL5ZkPOKZprBxMNJtGUfBuiPPI8CgM7WsOwfy2wtDf6vfGi8WwNFs2OWG1bmk3L5Ob+j QVcTelzCu3RbGTF0bnuT9K7/e9DH8YMMy4h92QeQ31SCEM8qJm42YALlxJwrKBEnOiIy5MUNGby1 crkRpRGDtWtvL+2Bsj4v8yqOLHQ12SCDUW314TK5HGfZ3NVGN9vbIhqu9VX5yysPeRG3v3rG2rVq p5EqxFx25lkpbMZ9spDbND6CFTXkJR+tsyCV0m0psjrcUdd4gWzh9aln+uCmt+IVbZUzd5j+33Is vTOgyRDTCk+r6cq9oKNbIEgvBNvr/OZPVf73aFOTMgHVJ+Y87skhojXSeWhNEy6IPid5fs9twt8i nRus/lGgkz38ZFeG/GzWp2W0ec90rnen9iD4tQHVFJLScabVUOevsjSN5Zyn1JxeEnXD7YLWfcTN AWYB5mtPskNBeIo+kUGVPy1sIyhMbSfpi12dgxLaJ+lhPZOBXsCF6ssrMZ3GB9Wq5QOoKradGI3I Pjzhq0iZx4WI2iMb6+j1wk2xq0zTSbiIH+U0+Mxk6jMnQg46zLQMXAtCNEv+oplMYJN8t3u1mSzN XIyPo1F06QPKIbEQT41FMfvdfQleGN4u+k8yCAJtaK5N/+ERUjcJafcHUL8gpbAS3MM6l7lB6if/ fC5XZUn+Mgt9SBOxKaKGtqa0vZsFIV9myVa9dq3jN22s6QktK/XWTfvSQImB0N4qARq42a7C0fyJ XtelgXjNqBp6kwZhzbsLv/yQSkYfnEbFE2zohAaMb4WlJieKP38ovjS3m7z949NajTgEWbaT4707 NLzf7j4B6NjN/hJDCke+TKmekfToZ6HmD28yrskKaDHps5DvUcC00Rzexg5BPnph9JsToWstP4wQ HU56uzn/n2cjvfT0tH4W2MIKIGWUskm9eOvq/wQ5pYHxFl956kiP4z2veRB2OvDCE469fzSinwbY /0WtOw7bQZs38s7WehFYj5gbfx0EryhXqG07L2kcUp+dCDZSI59fq2WSpOjhNHKlSKaSAo3WhZ0h X7trVQ/Qw9iNJT28PV2K8DCJPnSs+orwpykmkSJXU/upaggV46pkTJTatFU1vCeinxYKU2/8y4LO +NOY+eNPOGt6bARrb3xwGap/rD3BTLCQEg07ycoKrcLz8zNQB/t69NkDKUg5pJQX2ixP6tE7vmVT lGfq7sEy4n6mrqx3F24bi+lqWtPL3ZcU0rQTz1JZEQLeIP0ZKB7m1CVykquBz+Q4JJAhd/RFYeUs fOSjxI3WWaUiFzLEEWXn1VMpg2ZZ06rZDDTBm71teVZ4UqNTfOZN8ah2EzV/jOGlex+gowS/OYtF +On/04XzxBpGpMeB/01Iz+T/UWeevnyynlwmIBTuKYcN6DEC1e6FWbPUzrEYObTX4ptj4+m060nu UEIBl83HYWzuIxxbcyIK54T22cOVGPOeDTHD87ERKPTkHSJKDKAbSSkeU2nb8DnFn/xciAhLrRtm qFI69jUJ4BY9A0k2pjNugdGRk/i2ZFNDqyHiBV2H1V+nugumM7pZCfTcVdLfDgSrnqY3hLUpKSuS DTg64OsFGsKc48n7areWIK5qkuCGtXBeZd1z7vBCajP1Wv4w/cZpS05LMQkO3BZGzsFqB+rEVl5Z 4d5zu9lSjqO9fKYa4Htw3QZVPmL51MS4YyHq51jcQCg+/ddUx65ATpkci4dPEji2fVffqQGuzccX kQCfstfB24DauPKjnZImI44iFL8G0vTalqmZE7Lc80EyIJnHYBqBGWmcIoXL7ZB18eDxmGQf4NSR 3fxfP2h1UgPeUB5+aQJFlB5wvA394LuUG/xZLRan2DUVe60V5Z2vmXK4SQRQRkEUglRce+ibI6HH zMFNmPRJQyLdBiSe/0Y1/XWdbdk90hDbmFvgeIrffLUerT/EHdXwuKVjRyKLT9RvcIDOuvcPlq/G USqdlyuIsCDcieNJhG3fVSsrPGD4F4dokEUDdWXADqwcxMeN+mUaEYgQWs8YXzGjBhFncI/2K/1Y JPA3Uu1QL/SO3EA8jbspq7k484swRc5DxT7aFo7j91R80Y+KvYDEQf3i8OQS9iJyO76DSPNdQV9P TOdo+13bDIrIGM1KdQ/8No4eb1aRHiNKX+H5HD+GY1sqk3XO4hlcQJ3+gxrEKm1ZRQVW2/RY3AbY PWItBOlDz3mmHYSvcJ+E+LX0Hdfx5vT5iTsLT3sq5a+JLN9iR3P96Wtykqgv2gShBhLPm10P07/O zIk3sKdy8dE2C1O2W/pFhteNla4UEhRHlmgPBv0wOWoEH1RzRhGTWlT2/NYSO7jdzJq/nbJ0QrwU omPfjQ2tyHjbtB5dtKQ1KRctWjk86euSONCWWNN4+kTen6FaVItSMdKB7E0yWZdWz3nHYmuONGmC SGqcO+O0/yrjJYJWIZ/Ca5RUfAaabPyHSHZIzRm6b+pb7ThpqIpER5F1DtAQLUFUBkCZn3yqdBv/ fbUPseyd/eJgSvyQXWn2uSeMelfdtczgzAlNN4P9QG6/N1opJLaovmP3Mo73mRw3gZ1fbM0irZzP BOjE7PayuzozgfhLi/HfYp4kcTeGGcYQJ6HfrRiH8FgE1jcr4/C/LaJ9lRi5ZeGOJBx1FO+9vPJD 1CXrdKTGqmoylzH/DY69DgO4m0AVlGrzg5xEN9sdo0rUapRo1opizPqpQ8rg6im4Tb5Z/GaAZRAt d8DY9AeI2/vwGVRy7ZTNNL/BISOIb5Ow+ALtroEUtSu+ouC75DocDXfjj3Av+pMaPpNodqWW2e2a t5N1gx0n5HqDd9p+Ol8WiQbG4yBeU0LOzKSZzfMZGLnZD/PJdhGfOhvQ2kzbPcpWIjGmbAdV4WKq AK1ltcXVPVPi2aXnty5c5AHgxYK3gfPLYXyYsKYuxuuArv3vmcChvTMoBtdc3au4NchBw2aVPJcP p/T/20+FfRmQ3AB4blArftYdmdcJhtH6M+08S2zgk+ZGZ0mPi/QMLM5QTgxo7LV4NtUETT9PgMru 4OOQeZWiwxPIvhAFP1Md2jgSJnK3F7fVL20H2Mmnw7ES1PEYoZ46myM9WFhQE0dVnYMBrmGg08L4 OgZeAGrckHl9cpRSgjo/7+pmIVUyvkBQJBb1hsLF9lXfH8eTITy5yeQVv7KjDXQoMzkvFqxphzI4 fXMxtAEBY9QksrYps8jHu7c5rfGMZwt6TqbWh23nocypVae6RlutXP1imY4QfAk7NdmaPquhPSSL YrwcV83JG+VwxpoUBHyx+17O6rR6NqomyvFVE3UwUi05mhNzVDJMKD5htGDoS4zjKTkFsEcyo3wC 2cqsLM2lGjx0jTMJFdyPmg5ZFYBXEisUJo2rdhjLmvZCeteeUAOFCQZUk9pxIUVhMBLtGRVqPM1V pStuhufWnNQGhF8o3zydWR4KpnYX3bV2GRNYX2UTLF416XmzxN9pqYKIRqckmWmmLNbc7OpmaZXN Lvt6DzYfMOPlnK7UPV7hcOkjc6hFiV+A9WTU+hIptZB6y4VQMx4qF0dk10WOxIVE+qIatSWpqFB8 +u0z4WmTCn4L13vPqV+AYVaP2bPABcYtVI0+L9pXq1GlyVL7g3RZdPzrh7Ff0H35Pem8AXzJN+EO tlRKvko+fBqTVAWfmYPRuuX2JGAAgjKGivCCEgcHBL8Bmayt0Lvyu9nO3LFMJFnKNTssZ4UIMjLa ERonE1VByj2Ajo4g0dfGifK1padBCk1Nr609MUo00KOIclIdlx5kXd52v+EzaymLryAKlWMR5un0 s1zv+AipTca1gVALC7Vxo/BpmC8UA0ZdiOpwIZh+pybpzA/tK+NM8/Aamv74OR27VoYbKBubFAJx 7ZF46Y2Q8Z0uaXNfpa99vOT0tIldjDWXuKJVmZ/dEVxqN9LBekpwRGt8huIpoHcc7XtELs4bFPRe 8J1etpCVW0Z3RS86qoskOChrpLPu16/wcfFhKogEs9M6SSFtAbACVm0/0FR6EcJEFODQnDYIS2BB HAIAxqRvlrftRNYxDFhxwdDSeckt8LzsJ0oRmQLn/unrGNmCC142NSO5XZ7aHptflw7jWaB8j09D +pDe0RsjzG2GNlhtoLGxVWAMY23w/83TLjwQqU1l2rrputv/+nQ6/HLtguy0OxmBouhbrxkasx1E dNczNnLHSSKwMIPqDg7L1VNWJIog20lurWgPioCb7w8QYcTL0VIdFACCQ/lJC3dtbIEvLhn/+PUz U2yH3YBnBceEx1e4Es7pEuSL6ylQWFwpXDbuwaja9IsA6u1D0Aaber+k+UBrGmFbf/o+s/QyHkYm nFgOU4iQNkB3WUmmU0Uh/xRdCFoahQkE1dul699d8Ca8/MKJxiDQtq6GLqZVeBo2pS6MWpkAiRwp ILkzFFYm2dJL+7X+FQ2ppzDanrRQC/iQGLNNoSheUPkDgpX4gF4lk5N6iLwa1hRNaw75x2mwOXcm eQruK4cbQ8AVQfToLBEqRgEpYLw33xhGS8XQdKv0Za7lVC6rZkavXIUDiz15ReHHTMCmwTEBOgVZ Y67PogA5o2R2m0ZMmF2avmXyMHtBh3BZlbH2gkAU4r+CnnpVyo665kCz/zASsqvBIQBmOhjzTDyy rx2cGGiI5foyef5e1uBGqfIKG+58tvZPHfzGkMRnY38FmIFaL2wFw69dvAj331aaDGtsg/IEXjhF KciwttZUgUJAi7AupjypNNChdgWa6PbOVu2NN8CMNQJ+4Gp5vZFVoE1O3fSWl4o0ISk1tUmacG2U atCF6918Uw0yWG38lxLXEUlds5B54qADCfy3A8xq4oCjd2CL0OzhzcgNUdNnvck6ACXoGsD9g35c XzcVQEBqkBbYQWDtU6ZDAXb0HUKs+KWzmJGpbDZQymDKLwOz1VZ3HdziT5EfOfeajuZgrDJKRMFW /UVY4FcLg85aXynHRNyhyIR4Mi7nX0qpN/uRx5drlxCahd8Zotkug8h6xFXTCVfHMoQvHah4qSlr wIe/oEiZRgqX+Ym5eCYcTo9u72UrFKbdTt46LIhcTq+nBDwGxsL3aSmRLB9QUuPskWZ8As0c0QrF kYcIGJnQb/JEzqV1uesyQeSQzK99cv89L4/7rHgEyHyTwbhhNi7Y6KmPjf/RGK4h0+uUkED5seY8 OPbnQICHbczfibatkLijPn5BFdJQJnDQb2pb4/7VVhfbwy7s4ZWVgHl2EBo/+6lZgg9w1z2Fc5eo JbgUb3a3ZUlVJmXd3AsRa0irLLdHq+aQ+qJea4fAI45RsnyMAl5Hhq+Ne2DxLukVm/8JJjfx6M5/ RPHkwjI13vlyyMCNDbK8/W0xjEYUR8d7GzSAe6jeuS9rkJSUcDvbKMeg+F0rSduf4ZRNyp6LuR7j Ftud4pngrrJxKH27uIs/HS77lKoVn21FZLO9od7JqMiMrSLd8yCJGurVAeCTusvnI8e9hbEgzHAW 8ln0kgurxXVXjAl7WRU80MoeiwmwEacPb3LamKYR/Pm7Ngk5cLMOrqItd0mMdfYVZUMSMN3lS9A5 2OSzFfur6BEk9OMgMAsicxcOgNdkAny+jogvolsS/7xgP4yp1qNU61ONaXvkvlqrq8BrtZL32DE3 t9bQ08GCVhd4YIdJ2anVKD/7WCljepNu8UUK1nYebeBOX85ghVO/O9r4nw7P1L0Zg3eKGJHi4T9W Trg1sUSAuTbDLv9Cu5dYvixEG1jZUhZYxg9SbcqtPnxFD0ZfK4WY7St2DEjrrh3uBFjA1ZFz4EeK xzwTxKcCZiY8FVVI8nCRCK5TWGVxSrK+i0fpUjr5Nfn0Dho608mpQMQhEKfEVW+SRYkPZoLsAeAj OThlswB3RVwS81evz6RTzvZw3vGC2QIHrHMHQhzVZ9s6Jxv7IBv82UgEc0M/PagLdxdfMwcLaJzD fnQVuEkK11YWCVnmTftS2J4pdzVJsNSMp2m4MxlNHu+Xs9DjHQSFyrMbvr2yKqbtFZfoVkees3QB /GyIOm8IJOU8ezhEQHv2omisPixYWe0Y97xofl0frs1svbaBPEGW/wgmO2fyV7D0Nnvg0JC8s1FY Xdfv0GQDMWL9BbfDpCuXnTDMNLDAY6szFG52yCPgg8DB+sGu4v+qh1X4oeizFynLKm0y8y0tFkwV 9y025cL1rrriSsavbHvlY4FVEuIs/3Z0twoGM+B8msc+z9w8GslqK2AMpaWYucoh0n3gOOcM5pVd YtCJly8+6yfENowoojEgWZLjoapmct6397Aoq3Z9XgONmB/9XGg9SINsMFYyXnHaSHSgUgkI3IPE kJ8QiI/T7YJ3EH4LXvqFUjsDyb0n3W7ZUxHCONIyDBsnA+A1UR/+s2+bAeSDgDAKU50AmKwEp/d3 8l+5aIuA9NnR1lclhSJjbEjhH5+K6AhkveWpZX72YSLtEmfXzPD7ziQVLqjkfxMyNgD8gSFRZuJF ezt/RgOcLAEElMZMnsaxFegvRXpOmx2ngVye660GZ7cyUAJuwCtt8hxnzHIwqmBjviAlz87B/FLD Uen8hVIERNslShPDgX0QtXHJDW5SZbnBETGty/HnyrKVgQLIRmfBd1V6zoCTUBLwAJO1KRVOihPe 7WUmTvW0P7g1OJdhp9alq0usI79OY21w7Xa9r/36tF+Paz4EXME4QtnZjaCIh9rzRqZVLPn6Ipk+ J/08vPa3+Or8WbtK6c74WNYJuBv5ESoiFCCIYyyVOKJE7a7lxod0u9f4fLnRhn6/licUEyc8UBc+ aNPHCarm55HxU3PU68/JgAxdYGM0eHdZ/F1XTmKFx9czQnCctp0BzK9uE55geTE4UJUGxahDUpwK LnSP+nnDpBdBXfI/uinInFrXSskNAf6wFVHj1RTuLs0INRFtLGkbRC7xXAizU1Mmh2DtRRfr7Jgy WpJmZrgv+bU1MTRfv8OEGAbER3eQ982nVA6Nz/tT5tmOAtPrMIDAAFG57WHZhK7tNTf1BeOnGhg8 ty83LJqioLLdiuk5WZQ2LNB3BFAIXqbkyrYPjAesjLsI1k4nGWXXGnzO4hJGb+IZgFW7flnhSZSb RQi3f5S5M8ZbNRAMAQNA3l5owu8tvHYJL8NngAjCegBd8HYV96fRONVNOC+ifgxCPCE5H9VqadhW gR0chJYBqzzXl/Bb1/HIdJXC/OVs7tNbpxxn21jWCdqKtbooeMIWXz6rFCvQgXyIPXfUiJvT4+Tq sTo4zSbkNx7I3AQU/Bzx1kSlXc7j4C63gvpZ9WFyZWOdAM32JNHMCt6xRZCTbPzE+IDcD8TbG1XK ZKxCQYRjhrqeyYXL2HxIGI/U7bahzgeWEa1dMpiawJe7jCPV2NM128zF9ekPJ+sKw5NQ9/JEz8Z9 YSCkmzopJiJIQ9Gt0tccylhdsNdOqSwDK9dNjHJ5TlQ2z6nXo4WxFh+5FVSZychd3IAN6atvJDbN XxbUveJ9wQflP4DGbo8jyj4lpQgpbVFXvFQTM0i6ipwzEGm96wA241q1+uMQl1fEwTsgExHLsqBT wPM2TLKRZJ/f+WkTEVZQfxtGz1Ehraejeir/PqtH69ORBP5Cygb90yNRR6dlZcudS5mXSE+JLglf Pg/GjRANMkc4CEvG7m2xBj24Fl3PhsHL6by1rVcKNzpbCz5shUaKbVGJJ3FKjh+jkFhlFJ828d0n RN4UOaw6MjdL3oFYYAkjC/wVxmpRffIQULADgLBDT6jGEdDdgA2zFmGJw9+amAWg9N1StKvzD0hC nHTD7HadTuq5OXbYOlB67pLCHMkLxrkVoQ+7N6Ft0onDjFwg7wMCRzgxkoj5OMUwsFDfD0f4yPMZ cwii1lMwyNRX5bSVkPHGMaNpVlUURkpb2ZvD8THkv9wV79jRRVqULogSSJd8urXOvvZVIoR5z86w nLOHMCDH00c/WLekgwOzogG8dv3+IyJl10kyla8yifMTpFFos8VBrxMgfzF0VK227i8RGXnimaee 89K0WdPOOGrtnvLePJ9RIiEwOK/VcfqzMsy/PuGGnhoxCMtDovL8aSiohzuil9+STMBmz4z5TTcS yDyJhpfIozi8Jef0UKn9LMndAQkZZR1+ZrdhaoD2TGICFUdLvfne4IFP2FqGJ5GGUxZX8+qPoyR7 ajOTcmBOv63xQRZW6RIXH7eZct9WG3l3v+Nw4siSorYjr2ZA+9PQt7WYZxjw50UOaoaIjHkV+Gh9 CKt/4jDFP8ZlxPVS/5AGWrkq8AvSsvzsY2FbLmOjWabZ4XmyaGKna9CdJ2+8z0qLMmScu7vBXav0 W4Ylv4DkO6hw9T0UO0l+DPbmyfuUFBvrT65TG4mWdqYiki9R9SPLFV2iBaCKLIUWHmVdC/zOjgbj URSlxdizIrKm6udGF4iTpwqFVFPgGlE1VJcgFe5eyvUyEyrQO/hjpmzfvenS0ouQqdi7T5CbQgOe xmm182YIM9Gdnd6XVPLXrC3IEmFsLYC82PJCs7lx6bCy+rAQvdLZ/zLT2NsG5onLt6pJjLbtucvv 1zgQ2W3pevkim69dlB3FOXKKr4l0xsigRWG6PYVulB33DSznBUl/Udt65MiXn0QA8lzVDTDa3pMW 0jKL4jHp8rEpZjcT9Y57i8fpT2dm1MRXSLSiCVbIGFzQ8kDUX/7r/P2iwOdYWEkmLCfdLuqbHVBO XKNvEm4PmmiVzHNDygeE8j9J4pR2pXBmrCG3Cz93P2n9SgkhQJTLngGbSKYCZgNVTbOZ4/5Zk54/ CEoxiOo9MvoYSvgxrRMbBEyypFoRv0WEMaA0wk73tcDzpkdR2Rd/c8wUo/0WDtBIddpiYd0iIv1Z Xw1egUYO++3T3d4xdWl/FlDTqCLLC7cgZ/WH6lM0mBz7/20d2HBwS6lS+Hn3MWmwh9fMXNV230ZO H8BAv+SypJ3JDFyJ4T43eh5Ao5BZ0zYktcXx/bMqogO9/CULwlFHkfqKvs5Csu3Aq227zfP7UlE0 bVwtGNqFWljC8+ZxbP6I7771vNV+sB8HO3QZfxLDYkiXGjAC0cMoPrZnwv+idM5gUuc7nOJumQ2+ tE3Ddh/QnOzfZ5MQgM1GIKTWzGtcqSOVo/bPDyYqr/vTzTOgGfKK0oEFuvWPZQPKrn3PpCk4AY76 hKdJU5fKfMVblkPZaH7BQJO1/3B9AcZwuu72YPAB/1EgVsfuJMqarXJ4qO4rzSk4todR/gSizIBy N+LAae403k2aiDyCK12gl5A5d/qrPUnOLV2eHaAGiGdNqOb1InvSgjgwajCt2a4FH/ocSK5Hswql BN408N+bmppoY4virCw2iXtB23KmZoSgBiswmswhQO+tPxhsJJ7y8a/lB8gAe24Ji86AaP8a5fml eI2bKHq4zSxv7N5QI2ZVW/FVwoTFUcfYo3sONyu0ERe/siJoKBbA6RzxPrkLwtCn+pzKmh3w88/x cw6TbGi+1oX/41txulkL8kAc4q1ny4QDoj5Ccqtopl9q5AqNCwr/OZ+su35VCFF53bVsl1koly1Y Z2LtP5shGXQRE3uNhhh7J5IFfDYDPBswZHKW9XAkxAN7JGMfpsh5FgGna9jn9GEtokyvnF2qYJu5 tbQqIvAcXpxbGYcy1YPFjsDXrhEOcIu0g50vQxDmE4YNugkc2SOo3Q1kFzNaCiz6fIz/lhV7t2zt XvEGReC5PFso6Zkb7Xn3GFump64VvyYFAlcwrFchCxVtrpRoTup2/oYJ3a3TJ22jRpCTzzHckHMx 05iL0xDbiFN49V2+9pXssnFUnzz/gekElNLJZp60HifnRFNjMgPqao8Wi3+J0AHsS2cd/IB0QxAV fPZOjpVCYfhdRsBGsNrraNF5GLZNkeBRo2uJ6lS7/PWmwHXUPjgRZksC/zTdOgZ3Y28AvT6HT5op fD0J1Qt1s1vdz+lMu0tHtxBEBH2vztB/uVPZAALIAE8JQ4108h12Qg5gkP2N3txjNFX8geyM5HbB 1jTAi6QEipaapHfdIlJtQTTAD6Al88UULV0fsKMihYNVYA1dq9Kg4GbrXH0Qea15wqDPWLiGmBeJ 5C1owgmERnpB+HDxvq5xDqxCasb+n3io4YTpKXOADUY3ZlkDtLwuCYJ7wmctbQ5ZA/ZDP6ijo70G 5jlkqiX21uQGOr4+uh+I0wn3DjceuaCdFe43uj10sLq7bCGTGCut7UoP5qhM//YuUiE1c0vyBMfC LxrzWbdeJTWwfnoksQnNi9hbM3R6HNaflWGXqEPa9Rx7vCRCHmIfYUpDRi5yjlNBo594/ltq72av 73Q98B5mWYHxQMepTzu7soN20jFNOtjq7SVrI44COf1a1tiQj9UZKqvXwcViEzqrRwxAhgYLA5Df idrqptODKw2btWl3Upiw+NtENvFBQeZ3yDmflCl8yOcQLMdMrKipRsIMd2VVm7cWIffrZ8BKWb7/ MajTry/2g4NkUXz0gF4jbujfTbaQSQSZp+Iz5H3vhkc//T+Gq3TtPgozF96APqiL7ABbARttWdLZ upXczkzTUyxYGrYs+YbaGYJJZem7dEcvqeLW5IR50gJzYSm2m2sijCgmTpDnIPumNIlVlDbrV2JG JmfVNFVG0FMaUICvmMYnTvc4/SZNOnvx7uId3/yW9kak1Er/PIFRTc3Np5hUNzLm9Sk1k5S0hOMN w6UG1lvgJaLk3PiQPzRUIq2HC7aVARZUWDwGlKgVvHav31sg+m4MkV7hwjDaNtMm/g6S6BYgkNnN 75IK4YJXMYbzv1bU+gt1s/uPJh8+saOgZuvN2vKj31YhEThcqsRgxXJul5vSJ3iStOFDi5h5cgzS ZjGOEYEr9AkTvBXePPtnjAasEFDQnS7M23ShxK4RxIVSWrVYw+BLB9091OXvC9GLI8R3KaYlxwpz CrCdbefIvKfEEDHVm3xz0E+dmgS/Ffc9nG2DGgswnUs6jdQUXks1s3xZbDV0CRIePdS5lctneqXj KJPfM0bW+RaEmKQA/7bVunKlXKxaRYr4FMLo25nZ05T5nT10JboEztS5dEXz3/lrULxQ9p0C3mxy 23BSWqrxjRPzHyhvfdJr+aoGN04G1jRdugdcEuttMA0DkLan0CorcX7kI3pcOm7smaJ9a8BGFwSD /VBmhaAUN996yISEhVJdMF0Y0YZwr+4LZIr1spuHJWsvbYZT+r2qIh21iIJ16s/I4Sm+OaAoHndw SF8WsvMVAO3GmGbcb6wluVVEB+vcO72Xt+qq0eNHTVEKdEHdaNOvlNd7ccqPxjyWKvSqpnsO3dIQ 3Upe8nr2v5DKPBjA7opXveFeIwKZxPgx2X08dDk12tCiB8AOK/RbyJw/DJYHJP+jmDGKNsoI/4HU nBdk6KD7JURqJgAXrvx6vQNhmZ6Q/Wf7+Oi4gQAEWWSJ0Yk2xeipWNpDAJZdP/Cw6DhFl0WdNWTd kG55aorKAX2zMAeYnG946Zo8zmCqGldCL+g0F9mql9j7M+H+r7F5sQfakitJsD+aEggpeoDQV/16 yXhsgZkQ4YIu3wpU/etrFrKK5X4RRVQKwWGKXFrquNaiZWmpN0+M2u8sTj+7Tc6XA7jGD44ViZ5r QuJCq37IcsFgPiklEw212aIH+qI/x9ibzj+SGqDaunYMWrfhzBYwd3hdukYD+zOIEoQ8Q550kFPa qMKOQbOa96+Ztu3mlk2crjGMxb+Glgs7h0Npel+hdwY3rG9NUR40FiYC22oqp5yC8W7+E9chNAdz 9+ItDXSfbT9JHDoG9m3HxL2GQDNvk2LI7pw3Uc9pb/pz97Gey6QCv9+MRw7PqBPnGu8ZBUND5kg1 ZjuRamsgOOkFu5XmaZCQ65g6AbxwNdAhkMPznzAPpa8h7PNcjjQqy2p0kzlH3FIaEJ0IS93GM9HH wPUx7CWyzm0Ex7ETgBiqw7jBOclP4m5i55brdO9SpqWBjJBMomh2+JQ/Qj6mdCL+aXNFbdF76YWO PFuS4ARZX5HRc0jUQlwDmgq8zZkcXBhgFB4TcWaq/tKAplsdvjo9uxxq7YVWrp0rotzFQCoeoK1C gRQiR49MYlsP7OypYLj2Y+gwwqYCNG0THDX8Bgs+ZV3bW16CPTi2JlH/gEqbS+Ag6PEyaRCeMLJc ESyqewWj0XpMkr5BUQc+0nLJp63P+Gqc6dKWWN1GxUsgmCjgLfJFYavPDE8/whkAzDwhrcwL2x7y WRrj67xCW3vLly1dEWieYDDx7xBd7OZF/NGi3bxumaMnaRqs8NDsUy7C47rsQUEXLb/0vSelXGTb RKih4FdTITfYvvGA/VbZo7idpuq4P3+jyK98HIgmJE8fIEYlJ3WQqXrq23RT1vuOUR7E2abqw/IY kpiMyOBb52YOOdk0TR/7Hh68gfkyCpnU/9fPpZAiSvzUZmwuS3x9K8kV0P/J1FUVSLsJah77jJCL wTQTH1F9VkiLQwtHrLmeJETnkUCAvTsi8FRLumAK2ODKl/SyShm6stohdAQvNR9q0w6kGmKsuWy8 2aT5EjottJZmknTd1z2/4Mjlqh0/UyLgpOGjQIswx6ZYcCxeeB3ogeB5NYA4UkrjtcvhTxVyEOOs q3fSMniSID1Zf+GJa6Q6ZFBZksISVNDVUQOdagv2wqSeJRQkJonjGPh/haSV37JNzRI5FHC4a6DX nH3vVLZ4ckW13NEg8KsrQ+ubzp4kphTKvYTx7EgBljCbWulqrhQmwwiLOicXk/lw4eb+/Dmxq8DP fIavSq0yfMyYAXglkIyt/bqYINxCCKxh167/MxwsySuC1OP4xCOOrZ6HKVdVFfciaOfGqWk2kiMH Bdpvljks9y+SSt7JFnZUSaznrIqAsX++2vDY8pGE9Eb5tM+IzVGz/eIaPd8dBChz4dXhpU+Qs14X sx/ysrXNl5HpEMsDkxdhfkpNetxB/uFMyZB1oLnOHWYjnpBn23RFC0tqO+BwvgOh7XCBB/2/dzRW C1wExGWMqqXaumBygdXY6+c6E3Ao7SYwr0HXY3upzHc8jeIHHnK53HAmZkJCgryJ/GVAPRjDimrr RAxyVVnPu6FWBaoR0ogxAXA9hb8ZXHdBtDdEZN/xNopD8jscgwqAWQQGyjPFTf/ThS7bY7rS9iHp QfIpIo/ulktF2b+SMGdc79IrfHbUAOsLZDOGoOSu1B0n4ABj+qW7VJE+jPBDpb5mgRyfz9fnZCCH KSX0RXFBRQVmwDLFfNpxgc7S8L7BLwdBj9zLbyiduYO+a+WKynawijeqBsYivpxKP7Mw4UYYK6mI qta/n73rdL4yCefiRz5YUgU98asV2MSWcgnXUbyuet5kJjPVTv9WiGmDIt9K93QFVZ+bLcHMRndM b6xbFF9XDvEpifCde3eMCYAaQ5dRWnNK8VB3eAyaVWlJ5+jM9fae2epz+TVB5gvt4LvFTnkjbYa3 96Z9whXW43dbCflxnMFJDJgZqE/jQyG6uVtydq0fL7fb2SYjJz44TxEfxkUbKqflDPJj1frzBZ7J si57o8ilwHYCmVlXe7+x5hiSVAOXGc9LKOFBc4mz4v4RdACYJBLaVCi1YlQQ8xh2ztdrpNxOt2ml 6dsfZAtD2Gft4UGPsHBDNzbS1Lsf0ScExWV1I6ip2sQeD8iKJ+bkI9ii7DkACIl1OxemuA6dIpoT aBO5ux85f7kB4EI9utrcIYlEc+FrqchQvOAFNi8jjjI+jZYT/pLAU0C0t78Ze8sBR9qGvp4QLDed 9tIwVri6+naiDrFz6yn1GH4L6WtmhYlWrPXtS6LSxJxkZpN/WpedeDAwQgJejEb23EGMdbC6cEUy ATmBFJiFuyTiUKet8ym6VTiw3SZGpyeC56CgN3RW/kqmB5znwKDde7nWmzutxG6abUNzcxl6U4mb W/kD6ZEB4bLJvrW1jsnC+mUZPICTgVAkP08JlyYKQ+y/futcPtRytUTgEYleQwIfaksZ5pcYET2x PkF20XpRerh6bka/4f9rWPhskD+dOABsAjmo1/8iLCIPFjltmfel0qvnrQ3IguNLns8A+e1PufAf XVruJEHQw439sJGvgpFKIx567RRX5Q+xTE3IdAvn2bcCOBb6OCz7HlnyC5NbrxZvSCOkg4Z3iB62 v+rdKwd1S63vRlALzxhkyhHkFmTPCNthcFh4ovN6qzoFxW7MuVvJhwbiAwsNCfC+JTjvqFafIMnq /xqGEe6vHhyhnt6iPzrgfiDqUOH99TeST9Pe3DmmX5IT1gVQDBpCcU3OugEUJbYlGzxamaZ+DTc+ zCsVNAFd3oiTi+nD9NomoQ6yHLOK3omFVokPr4dmAroo3+TP2Pn1gV3D9/0jun9NRYxO852pUo95 EiqTvnYev12FnSz6bR/gA2dO+T9MUYzrMm0LyYOFA60qQD5yPvqRtPMTueG5NtlLl9O+VdnwhvP0 2XyNSWPY2vEDmJjjbrzHFJYjuLJu+BbDEeTtR0xEjSIQ/S+R1QUgCwJhkrDuF4kDrtN8zxKQ7m5M di48tHygc79xNZhdb6C0KX3RpjK2DU2/XJFm3GfunykfVPUAQFaQe/Dz3AafohB+yci2LUdnpEW5 1iWwmC942G6XqWn7epAM05jYdTG8p87ONo2zUdJyMSXIAXcJT+rCVU7Bs+DABDEavv4cjjbGoBsy vA82UK64wsQargl1x5Do/YL+hmnhIpeZYRhhtkmHLtI34evSJIYXYmtx+l2H0EuRWRfhYQ1iOzSO m3A33xqZmUPilR1ZWAb3A1+1FVB0Ey4YNMTtS2LVVnpEEcT56lZ7IT0VPwcJwGaK3DpcxAXRe9pu hLU7iM/uJsme2WNbrc1BaZTVlGTwi2lqzfi8NepMUY8mHJYv3EAC+wCmhdjVBJ8gWdUeCfNFIL7M /FIc00WOZ6PE3zeW6P+ug8MDaGAiRytjew0im4o5fuUgIFG8J9mnXc9ShTKr9xnegXmi3ER4M+nm De7pdhCFgKMB8CNBbk4ypEXDPNR+MQ9ORIMhZC2VpkLQXCFOQhPMy1uCjZYqIr4/RHmQ/JewNX8F 4bEzjRXfP3RcbHzzuwn01jHSCS72429bNmAOZ/0LkLR6HZRbNZsujT2d0BUhsasCtW+dWLGCMi4M bDmCI8a/33W3WGeE/AT+gowv6bAX6yKqI171p3kqOsY4x7zzXuz1RZf5frEroX3T48bZUJASOe3O ISL+Ii8+ByxeSxW6CiR5kg7Oo9deaDDFFisaAR9wHMeaaFyZfeKdYJrUBpAyqwL7cSciCvnKRUJV 8Z9Oal2eMLXEiKmzgp7unjWqrQhq5qVytw3IPWw/PyM5OVApb1d0X1o6miGcqVVWKzMRRd+TwMWR MeY+loeZSit8ighQi8IT+XEd2MFm2hU9k0IH6kuF2Hx6cMQLeB1aRvORMMMtVIwSQQXNBSik7rj4 5cE7VchTRl3q1dwkkB2fkk7UN3kab7sq7iym14CA0lFyNWMND++CyV3xf7cJOhMKqF4CQ8Gg3t8x EDAojidhqEIBUrXNe/1Uh6GAMMX+ZXY4lNG755EQXJtbxrrORw863uAF6M8ULkJTxNstJqoigTFN tbNMx0nGkVoyl0ljeOBtRGQ3FVkARfhQT5Wx/YDvGrU5ULjhLABMOqKEEALWN7lBFQ/cdAC7Yv9c VXyYQVVNaOxw/phpVW9yzI5eyiXQeBScfUyJmtJVy3YIzxFRf+lvYH9o9N7FgVmLc4wyLkpvRSlq ZfWGtP4Dg1sHCgAWbuUXbeWke634LeRcJB/8fd8qekMpmCfUNoGEiugS0QagLqhhOe+Q/OQdzmTD n15WIql6+U0dKw83yHayFbYa/glhMbyDGpSru5cu+Dy1914Me+QqRdlq8HTQKtozYFnU4lyaO3hK gpxQNTtdVhJHdYCk9U6NKP/GiSErWocrS7RaaRZcDjc0kD2IZCZ3xYVpqxQpJgzpS0OJHWVHbcBl eNA3KX8aeM+M7u/Zm0/XzgcpwXfU+9/l9vRaUOgsyjchZ7hG7KROmK0rDYKn2hibldFZ5SVzmRLN mwUVq9NrFWG47xH4twqglNO/C/mGrgLNy3/PCRh7XBZ/ASpMcj8AmBQyPVi1G4g0YuCZHMVrlI6w JwUKNiGhl0zLrcrd8jw13/HWWju/waV+cwWWA0Va+ecSbHeiT68iM43Uc97Dgzvl2bTKNqlsM1qT qh64D+4fSFrmEC83VPRlE9K8x3Mb2Q9FrMagUWj3NW6HEXpSNprIyyaImhjjUW3kVmEaB6svsg21 ouPxD8HXbzFJ/2FnLlArXae7dfSOd04p6p0jTNshkdJiGSn/IM625vG8+JeAo2oCr+rnN9ZxU9rp JyRpKnzhdqw06zMkY3VrnEre2iDU3A7p1ykGUe65e+Od0WVgqwl1UaJoMFnSluqbbxCKgwYajKiV TbA+F5gHu6deUFWLdLj0NSsz8loybPzGIseKftz1SYs1ncY/e9YOkcp3Yw5jA98hqQSa5u7kU6pa F6T5r6kDyLcNoY1SnurpklhhZWhBwD1G4Abjzt9mhU+sNIGBbE4G2RC74J/h7X9UogIAUVLigyyG yOBzmaDgep2dkeOgbnVQlfW+9Mg5oWDpUg48tVNIp+YlmHXu7hG+fZ5jqc+SnoXXRWtlDhAhPm51 bs6eFjUQFvvPc/X4HTFUJsIjbUuWjyLt0uHgCXOeCifSuITgeoApaKXEHbDaqFAuW3C2xFiJx70T A27zkJRZv9PMa/HUD6zUKSpW694VMD0wgz4/DWr4fGTbXYvgxveaXYdLASEoLuuL2Nl3waN0QeeY MV2UGTqMGdE+PnJW4QwRSVyOsxRRWI4YE5uIERgwWP2hlnicCWeZFMJygeCd+tRwsFTvNFEZZ1Wp BmDpg8dkb53BcPhA6MCBB2+9Ueb9AjkLd26JyMvLNqiheQhuDx4odQSgSF5+oLUFbnRpqX/vMsIV QDE3UI+qBnv1+x/I/DO3NJbevVmj/ynB91mBxnLXEsZWmNqaOkDkNlXsUds9e1OSLQIlAcz7yn6C PNvsO5hdzkvdgCJRobqpxuMXfG/WI5k1ysSBQtj0rv0k87yDdGtaJiStFkdMgFtdkuKoYJ2x4siq xRM7z0/7m4Szm0Okjd3Q4H9AfoTGEGEJ4oLFve1aqRLJJ0y+4ZCZh5k9+cL2v0DZo1gmn1f4JLvO 6X7TvlO8BxFQMrbt6YHKtcl8/xDpiaRJQ/l9yjLc3XBdTL3ErMEQKLn1sNtA94q96TXMyghgytAe m5bhTQtgu7K+2ShourqOJYoyPQkfaTXql4dgUWDZBGKK+404TAMtcJc6GyFG8bPBL7lSkGdFIRye mi9cwC9gaEI3+5vwYhoOhxd6GSY4HWzIYaxBRF4LOEIX1jwmW9SY9juhSYUd+vDqg9q0frVbRv67 jwAADzn2yhJIR+tQJ62hskgrKbKDzUAIViYX/cX78VeZreJlnXqg1pP90Kaqr0RNaS4YNGECp/qd Njeg9QT0hDy6eVJp/TwNGXlB2oyUesGPoPGSy9BlQBSrBzYDmWyoo2SMma781x8DAy3+7thQL3Gq QauNotElvrAnznultva0JGYoNGGtmvX1YSV83V831/ScA1fPavLRWknGs99i7eWtKMzgEmsGEJtc TkzsekPCzrXr+e2W3UZPwwN5K0MYHm5ApnCnUENZekRiv6o8m4zEkT8QYyBhutnlDaeQNstViskz WwtTP/16m1Y2G6+y4Q0DRmVi/vXyH95gGnIt2V4+9YxcTJ19CuUcoEjMzfKrs74MmVNjBUfzq4Qm Fugdn6pf5rtX1rlqNrx7aVutCE5NC14/49T1BYdstvQwJDL2jI/1O5Ih9tnScZwF8E8rA/lYSq3+ ho8KcGXbBCkQigjvOMB8v8q14+XkNyuNnZopiqEiSzZNYVZylvdWsRjFIKNF/EF+DnP1L8k1WmaP f8urbgbRSFCda5dy57UKFScWISJ7JPhGiJlsLZ38RwTqe7qs9+yRHogGKcNqHkLJHDl9cZWLaDj/ RK4sLDm0ts8rrK9YCNXbTEaicQ1m81oaA5CKdDgwFNCTbdYmME1lS0XTcsL/lOhJKKxXKxR43tQs ijx3FNvWifrhdHRPig5y2Onq+S4g5bPQoE384PmS1ALS9LueXtz59hoOMu0a371aheYwCfZZ6P9C qKOWer8D1tgmKSe+9+TrGgweIzOWE2cy4t/lL1ZG3DrHdv7zTS4PEN1W+Sf7qNS+Hd8FjgsroQPW +Z7buB3iSNILfTyPqEx2h+VBOma2PYkEKdfKD1NLOPN6Wgq0t0eknlNgx+ARnS05N91ooXjT5AmN Br5YT0YUSxOm7ZdRH9o9sy7sL4z+vvUGJDf4jO83EaWUVjGdTQmoq5GcO+5htzN3SwLer+ZKAeHC MquKkBMfIWndquLOu7U0knZXhrFoeL60WiyidDKhGYB4CoyM927WmUpKa9lcOnidR+j62Jum1571 WjSXiBLBoxwRd9/cEWQ/on446FSUekC0OYHsSRtsqzyaSfYaAw0iC2RnpqvL6jgv6MtZ2m8o7DKj IlzrBGvrS8dXs0Nf/5RS97yDp+8JWZDF5yonYpjs180Yl80eds+ijQyvZmo8Gpw4QXnPujhAKuTF SWTWDhiLe5sx78xG4WpqjnT5XWx/TKDfJF+9KCZMORNJqjZxH1TXEEdEsBlQDi0OMBURG9RuKFcZ 8pmFQVmBLYWstczSAKKsH77rIjmGtjkXxTxhNoAl5eU8pikmYR94OPgFu6FETBhRWwpBsDcUVa7x /fsG4NVX30/8EvVI3t9jfr4D2MB3/n5U1NqF6RtIPmg/taPZ4QdtEcQDMVdZ0ANS79B98qAtoJJo FfWOHiBx8RHei1cCH62N/fjp69vCtgSEENSmH0R7GPjBjzO0V9G2c+Jk/YiU1b+EALFZhpeO0iEo 9GnxC+2WU2sLt/QpPcEWAkdYJkE+au8hSNmvFsieHZEaLaUC8l1H2U/Ca2hpwzQWi6VtxieO4wAX mXlyF8qJcw+1TmjKaQckrQHeKA00qxiWoWQ1UlD2bEUzrW3Y6JmRArM3+InhgCxoq86u3A0GubSq v49LC4XVzbqugFqqkGU7JJp2vQJq2y0TH8IsTp0G+A6vkpbHZgE+316f6fvgPf1ElWvhhEXoFsjK SXHAzHqx7MplMybNWOUmFSSO7c12Uik1MfFxKMevjUWgfkI1chGXS+cMOi+gq3PLwtOAGVyK5Rkr 1BaIMEuzJOi+yi31j7IIzMJ0S3Y0DhVBbFGt1296NAPHWMu+ujjrlsYnm00N59TbVAoXEN13HD0P N3cLrFpG8r3yw8C4e7JCusi+u7Zcx1mjZioZBTvIvSxQUHlp7qL8HYIeC4GUObDaPv2xDV3TQPGv VxZbeGnO3yQrwFwwXW/VrNQ4SlIHNLN2357c2KmZZnLkLhUWMTDuTMWTueELoT6McMZ2S55JRApW Ui0LmfB6ZOjkoV4QqfkA9AIYVCb71mMBjd8StIgOx9kv2mijUWy1ek6sl5llzG2kI3aNAJ8Db/Eg jnJ0XT8H50JPAvCwDD3ks4du05ZemrLG1FlhiRk+RRKPqT4oTPZ4irQSis+gUFvScOptuZbvxdAv nFQ2cnp4zXsdo27CEU8BbllaFxgyQYPfB0DgbfxNRm4E4Nw7Jjtk5r5N/CNqVOgLv49MsdwC0wtx ZIRJ0SuPLoZuRG0bZNGgKdqaEHjWWsSf751zLV0q0IEjApwbjnUC1+84FYtl3uElfWSPGtIauult nqZh98I7UBN13P3qhyk9ctQ3VDkrhOjnS2LJz9hln2ARlusfSZ/aypEEEFmY82PuDAsXSvSTf9l0 GM2DugK+ljb48WYiZptcbyBXHxLgHjmUV16wp0H8o1ElA31gycvWbs1Bcj/drNNYyyOh/xuLOGt5 yKVpiuputXXD3veWmeYxqlJzGLvUWM5GDr6zFPG5bUHOWE5B/RBgXVwlFp1QDnILW0gzpg2UtOeX Px4O2ZnfpnHIdoXK9Cpsv9N6ie4xz3DUWrHHgpzJuNQs/f81Dvt74f00nEj3S3i0iCnJse3o8EQq puCUrnUQ3OzA9sMhINBzRNgNCmVuZHN0cmVhbQ1lbmRvYmoNMTM5IDAgb2JqPDwvQ3JvcEJveFsw IDAgNjEyIDc5Ml0vUGFyZW50IDE0OCAwIFIvQ29udGVudHMgMTQwIDAgUi9Sb3RhdGUgMC9NZWRp YUJveFswIDAgNjEyIDc5Ml0vUmVzb3VyY2VzPDwvRm9udDw8L1RUMCAzNzYgMCBSL1RUMSAzNzUg MCBSL1RUMiAzNzMgMCBSPj4vUHJvY1NldFsvUERGL1RleHRdL1Byb3BlcnRpZXM8PC9NQzAgMzg5 IDAgUj4+L0V4dEdTdGF0ZTw8L0dTMCAzODEgMCBSPj4+Pi9UeXBlL1BhZ2U+Pg1lbmRvYmoNMTQw IDAgb2JqPDwvTGVuZ3RoIDU2NzEvRmlsdGVyL0ZsYXRlRGVjb2RlPj5zdHJlYW0NCjOI3ENG6nnX b97Vcj/Vi0NUEdI4kVlObgYWYnmCIwg3j4Udc9rLbqH9gx6CmB3qe+CXJ0Xq+hRl85TZDjGz75Lx 4vNwrKk+mKNo8k3mIYFxhvgTpRXa5ROLWugXFlLyUvxLNEW8llF6nJ7QztYdpl1v1IEYNqu06lSO Co1gz55zr02l3cVdXh6FpYmYPjb5mFRiB6JAeMgpoZFFSEDp2ieeeAs76Qht3NBGBcsrqUt4N4W3 KHdHA2ANNCyPRNLedmmncwXnKldDFLXznx/BTmeEqVcPDu2oaibVmPE1ifcrUMfxJUYxuHsLMZdo BWUYS3M1F8/edtvyvZBfGy7reBcd+nKRA6GR/CnnmTqoFCwCouVs/URrY4XWmtzb8y792TfQIYlc IssvziFgODJ8FMDd2CmRopiQR71QL1CiIlRQlUfvvNTN5y8bzFQlEl0N9VrCtrqq+urPK3Sy3ecD Hz1D4PiK4yHswLnR284PKvRNns6ufJVwWLCvn1d6qSjcVr/N4q7UAjuM7PPSxB7bpaNqkXFliH6N m/Atro3bh4EDdDE+sQ/6LLAG/3QlKThBWfNXFjHNf3G+XBXgsPvfGgFHtEPZdDJJ+2E55Jfi31mB CIihEgGhxCAqkl0/WHGpg1nt26IrZ2DVaPlR7OuW9F1s6FMeoFcr4ElgS+XDCFWB3m3rzaDlVICd EtxaqPCSHCwQXpSKQKft5nXXbwBX4BMEw/f4u1x5rY4/nlIHPKeqBl0GkOxnaVAwJ2n5HIP5Y1qm bD2NSx4cVPEDksCv3iXiDauOqiiePb7RJLE3JZaQhyQIbABcILQ3+QWOPdEaAhcuDAC5GBOTmuxU FrO5usNyxryp75ZSe6p0enw3kfFgzgzvkuoV/T4MF0kfvqBL5qcK8j5EW6zmrpVmMR7B9HZV9oDZ Rg2I8b4HASkn2/8hzI1MG9sq8mtbUu+pY7oTrJfwV4K+lBZuMxkUYvNk/V3dPMtH3qsW1FkrZTMz a+oSEZkJL5qKbGOoZ4ycogjnaQMWo5sGCIijVO2/BFlEJqmxrHl+/23covef0EmSBdSG1uVaMzed FgVVONBclouY5NM/u+WAhfPoyrkktuptypK2J+S7i0jmOBypvA50LCNF/HV8Mbp0ieul4j/4l0fB +rVFTFLkbpZOb+oMrovsks3eV9CRcLw43wrk0sHwOjPhTGZfjdPz92unm/5p/SRzVS0dzGl+2xIy ccAKQTXkUO9CrJK7alLKJGZpipKmW0kKf90cyjXusPOdXZKRlzkG3DYNktUHlyTLWDXCkoP9c89L G1T5yZ45gZjgded6dw/T6dKkmpsLm+fgN3I/AY42pqfYiRLh0czJOwlX3UyMFIT4ThBY10z28iHH yAHSJhJ+ykWackYVSqBkqasn+Yxj6zIQ5SvshouBb+zrx99a3OgijOpVcZLIhj2/xJviXSglVhNb 6WRHW+GbMjHK6OH26PeN2NyGVkMHvSpjSznG20+KoSIrAvzFBeqZpIHWZN90o0vHhpVTFE9nzwcf KmvEnh/rlJKJAq29ez46/365oMEBDgHHJp45jSQyorRgGURCYCXAGSg5I9bb83B0LqgakGEPY3Xq hUfddWATxoNtjlWIQAg61k8T829c1XEp8BExuShhLDjn9WSE8/hXq4s6ESiNvwVrg8AUusN/i8B4 pXywXPePhuMtYSe/omyDN6Gae3L4Qll0nO0xuWQbXYrIUka48k8DT+VZy986Xs1hszMfVmzMtsCM l3k8T594tp81fw6ltoMdTo0oVqMVV02Qmwd45D6Q5OUcxfDI81idV6LEAvYKAHL1STLRQxo1SP66 rk4fN6sU+5raXIB9TBTDQFzz0rw3cq71E+dQU4ierq9u7CUpZHs75mgLMna9jzEF1bm5d1ggaRIE 6g+97CBEo7vCRSCi/8LokOHzrNIbumgYFSqkaujw9xUQiVM2v9t+Y2qywjVtM2ngPrVhrWoXYodw gF4MUVrwUdMENouTfY/3WSwnsUWu7iu/CH1i1ACDxKner1tWPP8huJAW4ysvCV2J5iYNkbht0XKX hAJHJBRE1e6SVj/xNE5Vl+ru+FkkhErdN1qlFSQK/0FJ0qsnYJfUjYMs7uERU9yxBgT9kL1hMGvH 4hCiQHtOk4detOzkEWWSVdUzWcCzr9V4gPxMaBzkIUSKG3VyVufFY3WQgZwwV9RQ5+FTvY2JB2xN a6udGMNYVwe5DNn0R1hjApfpQ0E/rRkGsxBl65AQx5HZwWV9/Ks/Ev8Bukre6+LlX01ccVj+RqSO wmlQie1kiEAAPJm2aZs0yl7t5UOD1j75l+02KAKgtJSURcSUtARCokOroRf9acH9M4p5gaNJR79R m4olE3JluRw1YJeIzIRjKalknh/UD4LGMOgW5kcT9ntJnoyXatotvOZiUvJk8C1ypr3I8WaWLJbN FqWoXzLHluVDykbMNO9XGDDUrOl2LCBgeLk/j8cQwh++au4Ej44pprbzRGtYfN3rhRF04z0eadsz zWTQr2sjfCNV6Ufz5eZtVdBtJfWjflYELX+puddrtqCL0F6y3l9IIwQYI+BOowCqxmCwo0xPTETN /a9PRQhYbWFPy6X6z6rIQOoOZO8ugj3961p0zeSV1VRgfGm5hLbcYLoCy5hjn4BqBox5m+2V+5HX SuOhZJgjqICSwvWQ4DBAcpjInZCEjw1/TjjF/gYwXxC3HA+2e5In0tedEefPcvw5RIKD2tXibxXP 1ADEvWiqkiFqd9sAzuYzLAD+k0g7lWFP/4OnENDdfqawpJdikpKTWdSwdyt+rYC9kISJbR5ApJzb MabcCy25ZPfDLFU3eKE3JFvqBgtudlAq4NOBGzrKOJn3dgDumu1w8BXE3v5dojRVPQfx9z3pMNSU fhD5uuGJR3ZWCIS5J+/jPudOdcwHIfMAyHw+cAHS8CdQpDzi2Tt2tC6IY7hFpBDhIQ399Tf/ADaU tnLVcLQR7dnWLHcul6agDlwwqjjBZy2GpOuy5I+2owI/l1ANfSoJhFEu7R3FbQL7Bp34e3KSLTN1 Ko2oLd1ouzTK33HqzhW7WhtcnX3Atj0LXO8IUHd7I/b03MM2RKsttQLf4E5dfbmqR6j6tL8Nf85i YLJ7vykAn88tNrzn0nkKYUXTkZWOeMdenVTpXWixpAlasnQWoroZCgqzQ+JGL/qVStoVGxlQQDnl djyuGYaBcBqpz/q9oSYl7Xhj5BWeerrsZxpCVRxMGU3IOdxNi9rbz2jzrH6Z2EaREyrNP9mVii40 Cqn9JqqqFCFP3k8JaBLN5rJSlwj3IYUpVf77LkfRkw+7kaI2ILKRTLq+w3FifLai7noYZUNu/4a9 C07HVCPCydDbGsOGhoQMuRa95iiUGgdmO6of88dfvo9CJJfJNNT69bzy+6GzkezcdnaCgWkm+274 43VsQYn+V9pWUof95JOXdBEUAfvO4SWYd6kxomeAQGLUFLUijzEsLlKh63Pn2sObN02QRJiGItHD U3NSITSFE3Vkux7bioBva++CP+cSkudAmUzmZ4AU5mks14duADUuKFLJfdhf7lUTrIiuF9T7Wk5J NXVtUDcZHex7g2Ai+eS4f+T8KoIJUndtO9fGmSVCRg2S4jtLJ3pgiAPAQSRCD4k9Xe8fj4v9wift 6wHmzqRxHMsM3qRK8KiB+ZkD3tyEAisSQAyWffE2VKPmQaFk9a6fy3ZZxMyCIgjZODCN7JuPNrUX r4jtRD+kEfeExO76TsZltwxMbrO21qK2dFZS7sxj5xbVYw/Bj/92g7ijWo1L1W5msnVTKyx4g8+v kDjzUqwb62Y5TZZZpBu/cv3ESbZj1aR0hHNqJYrr/NJHHCcIyM1rq8y7TNUpPHONl8sDCZZx+evN pW+n3ZeIjKK0msJQyNkR0DXvdr3JlKd6FHvLCaDtvV+gQrv+hhZU+QhwWaYpGYkRl0k5kamqMTF2 n0K6w1ysfRCUBalFzHcKB3cq34W1aBGH8RVl+/IeYljvpUJvLhjnh5r59T7GL7EitdfUSifSvOxn OfLnLcOa+YeHaWyiSVnnBfcdxb+NKu3+BpVyUdIggLD4SwOcUO4cZ4Pcj5HZ5BYokgl78DMEUZSN AYVlZ/bxYfNNhI4jI5U8D0BnJj2Fv2ebQ7CE73uiZcfSz0VrQ6hqeJy/kISjgkOGZi0A0PhFTcML SmfPCS+VX4JiuUeKJzDE34/5dCxcRq5dyPReXJw8bqGWUZzWahOuySqHf3f0EWM2QzVP8ivNCAAY xcXVpbi9GtCH55iZ5xZrh0XPAvjPEEaGqGVXTJzODU29cNckXlcheWa1CaAhoDtz5R6D4nBdCws3 yi2xG/NFWK6DqJf/XULcI7TxqhwC85tp80p+y3AzbN6bRGxUhJT6lwCivtfy/vFvuWSMAPwBOiNR 9Jr17XkhVz85XUl6BRa+IxCGeha5ANw9Hnvjmpj1V6wY6MdR2qLG+U09dBmGMKCHN3y836JkKXCy tW4bGMgPBha3i4646WPD4IG6kqQsjODzOPyiVkMnnpHDTTJaOEflU1AracJyHmwy8oVepAciCwmn DfktqCBPBnKEPiXuaZB9KbKz3kILbru2FPsNCQXbvjJUtQQc0nudfTHsNGspW/t+AlcGnNZG6l0L 9MJbMN9+3B+MP+sSnVOK98pqjBrOhYNSNNiTTgSCLf7G6JW9WzQRpOV49wvNzKQSZqH0xJhE7bMg c6LTDJ3DXIy/TbNv6rc6g+evaKEu55316cwZJgV7vnU2dBGNuhS4qLh+FUitaCFupVLWBJPPl4Xj OM5IcMwOezffPfynm4+AhDJy/KHCu4LcsrZGqmU+f7+cpS/4UsBj5JrGiLSLPfabza14ggJV3OjY gOktNOFYVB5tXUsV3GiQWmC85MDOAd5wOUC+9w32Ou0mPxHYNegVNiddPF0LNWhMlXzfFUtoZrhk BM8VY6liYnuS7gCL4Vw6N+3/yn7E3x3QQi3inTbIaRAXxzYMo5VdJXLe214wa19cKYWQmg9qki0a vchWx7g7euXBYBk53FLOza0gBTFAvMbJZs5OMHqgW6OzQv+xTNkqlQzPGOGEKEJpwm1JtvXhwVls 7YHOZ810vn+nuA4uncQCgQNGtPjqEM9/QTQzA+NC3rhcE4RCemYPkgUyBGK7gAQPvmKJX2o+ucKV rtL3pS5xh40aoqSXWuVEzH4wmLFNX7+KcVZUdA2xUklJ05Ue3T4SgugOTuWI98r27yB9L0Ot8PxM WJPsH/yysvD2u/TCXJo2bSfZON+TixdEFBnqEWHGQmQcmoxhr8GEDNl6wDF/R6r2EVCk9+0k1np3 LZPWm8nBWjRi7w5HaqZMXP+Om6+KDv4k42qRAylO+8FIFb4iYCpU9Dz5SnxMapzgOqPNQZBjK0S5 LTZmbYmXj9KkRr5aOe90nrcUEIAq9RdIfNHGEzeV1NQnDbn1YjcxWE4ECv1ji6tCff8d5tNHZCCm 21s+uDtCUNQxAsvGZvqLr9V2hflOoOuZzVikr72pu6Bce/f7ii7ukGN+lKYH4RxkrBHDzTWazwRy FJy1PYvDOh5uWxCJKmTNu36IfHWY6qVwhC3r1vyFT4I0zFb8iFZ6d/sEXT/y2ISEdASj5cv15Eca S9wj8xEHTxAvCmoyPXgquzk/Y4DJ8DDlANnqFwxGTrcB1eaRVzKIEhJ7TTccAHyGahKx4yhcOPNY z8whzi7Mt3psd6zD20S2fRUPh3t/kJ4ApRnsBuKd2W6Fg0Bx3zdmRfOTvJNTh/RVrkFnRhMhe8D0 ZBxk2kzXdOX3GX6y5Da+0yZBGRJ7bEhUXLnDv6MVDQo70aKKmEIDlDk6T2hPhAYzB/2EjuP6jb1Z FKasrn4wM+dKp9K9teeTvmEewM68GWAjG7Wx+E2w2KsRdh24cv4H4yTN77tPSNzob3JSnU5Mj7t6 XSZYHrRNgf65lznOJyIf0k+jCFOipg9QOQPMZfUTIjnuF6T2HEFKUOOR5whTqRd5Y8HrEhMgCpNN 2lla+gHiAv713kus4kO/hOcfPLbIvzA9OtsSW+h/zzemOF/AZfdoL3DYXuPA7xG4GuN190hlJaaz fpSYDqK8iam0bPolMyfj5T47S8S2pGclnxH5Hb7EWtAT/gkN2b3M6JtPc3mcJT11aHIQe0CW+Iyn k/Zo5sWLBxgnZm7kikhbDG0zeiEF6KuAK25LqO0OSQ9BoVsE3qWTt3ccK4V5XkEv9zkmHE4HEa5V qhZ2PDKNVCxMVnGEHdH9Y6Ndbu4xqwrXQjvluAohUF5DE6R5k9mR/eVdEHUGpZbskc/aX+g/QlMw h1zw129gzb2Cb0zU/jtN9UCWlF7CVpXFkXVe8ehrEe2JmB6xhGaa9FMvBD3+3xGEv6F8HXGYNjOI 6F1uoLW1iOBp0lxGg8d+XaXiYoDFrWFcFqjThRQWyr+AngkdbVtz8toB4qE7aW/kcsZ0F3E/Hi47 lCJE7Ijh0PvXi14IPFzA/iZ0St3NEbP3Qo+FnMhUuQ472OPzAThAcAlTSAcxyjPONDjlkkBb0l2j F25v5K86Pn2+8UetMhaFMdEqUTlupZiMLahgQxWp6rkYfiihGayCF5TVQJniiCnhbccRdt4+s+KF OUXK40m25qr5FUTjj7yccoiRepjPU79fUHCVQPeyxhH92Q1K5Hy4q1to7Oq4pPuggZfcDATzJ4hG zYYd/+BdOaDl6QLMBI3fs1B3JC2Fl0ceE/MHV1wMUXsSq3tJOJRufQSQ6Vh+psY99/JPbx2p0sNB JmH04PfVMw/f+DEvNAE0go91uP5fx5Xyxv3ZHI4e6Iwivy5qdeDvpxVTiqM2F5mSXwLexub9JDWR ZdBXZ3i8ev85C61F1UD6ASaTaCS46aVdlnNV13sbW8+NF2EFpk/yxWFVd1vP5IAYhOxOrbKeeNGm CvmtPxUcpsrrEUFDeVRuZZ6foaN5Uvv5Sfwkb6qGd10bWXyfFSGKHAl3qH1kYYe+fD/X4B+w/uzk qnqNAqGxJN0xFdFa+8wcjwLR7CQI6MbXrFZ03hYo5L10RXQHwUEKqpHTYYe3NfNFqy6GZA4H9+1f W5opDXc7K/gX4tyUN6AV+Yqc3ZpfiS3U68qFZAomrEJaLZmvex7iMhmmg02Ux6DooCz7h8mZNO7P RFOBJ7swj6EPgvXXWRzJJxBhXH87JIqLlaXYHyJwPOeuyO0mCtNtLTjxEuCXq1yqYtnIoDDJTEXD mt3HNg4gdqBE4SRDW23k7Con3LnBaKBSswI12DMHKWWw1yzFv4bBqEv5LfWBhsZUF3nbXu0Ahwbv DyMUw8LBrIu3XinIaREKHk63udjPHXrKvXGB4aZZu+/CY78hxTFaQtQauLfFEWE6pe8YoIjPKndb bVYNjSJlWzrPPlvL0yvnlmBzVjdJ/dBSc+Go6LLDSOyaCcc+B5zJYNVzpBOFJEIPlMiaqpJgVHCh LhYUlP+f1l2x70Kn/sJa0Ff/jkWlGIL52tZ7ARTdCzCkOQamncnGG8nFiD3yvODIAqv0Ae57frZi IL7NjBc1haAafeNCtyePODOZGU4NCmVuZHN0cmVhbQ1lbmRvYmoNMTQxIDAgb2JqL1BhdHRlcm4N ZW5kb2JqDTE0MiAwIG9iajw8L0ZpcnN0IDg4Ni9MZW5ndGggMTY4MS9GaWx0ZXIvRmxhdGVEZWNv ZGUvTiAxMDAvVHlwZS9PYmpTdG0+PnN0cmVhbQ0KSf5XRrzfC7FeHWVtee+UHs09OdICth+ruWet LFdrdES95S1mS9wpmE7aJNl/g7XODor2Wnmey8AFRTZ8vWRYBb9erIW0l0Tw1yDC7RnBFPapdzDb bmyDwiZXV0uwV9rj1fpKKu6rSmjbvXMl4dY0yAmJOyX9wWh6NThQwxruJpCK7VvaUyWtqrpv/kyO rUQ1J6SDqewpJu0J11OJgpp0TYgfw7l35nJgYMZ/EcaJblZ1hHVyZIed5MZIUxPlNRAc4WezIODu GVR4tjj5S0OV39Y0wgH5jef83JWxGc+ou3ARnFDL8vXXAUtaeYxy+ixBuhdNlNwDHgHo6Xw0qYpe HU4TlIN3EOpkb3bm1de/QyjKtorBL4eRonIt5VjAz07j5ZWg+9YY8HSgl6+tsdi9875ItB3jvTZm Fb0fwYjt1HI+Uv5Cb9DMfzaG+AOaKUrvrJhII5jMMsHJdCu3MxgC6FGp3soHZo7xTqxsWUhF30mJ uCQ0zEfuO+qFXigwlOP0xp4XdPa3DF4QAtZyxW3d18RkSHMAhirZ7lKYlzM9R+Wuvr0abYXUOYyc xn1zbbDvgRaDqj3FZ0aZg5GQPD/qda4Rg2DyWOzbaXVh7VJBOi1v1cR0BkxjfzJOm6+A3fkeugVl 9+gmknGAcXhhCgxIC8MGrdUvMP4DT/BT2aXvBsUNhxScanQgybHATWHdoQ/MJb2E5z2MJ1V4etuJ ryRnE2j23MQF8kxJi57YfP8SttpO5IaX/mJVcjYEu0AsZi6rA7UXmLB4Y1WUNnwBFtVTdlfoQ6mq c7g0RWSL3/kArh9Ms9YZAeAjf++iCCDjWk0XtxUwqxRk+cxygobg/q+XT9zUHgMu8qEu5R+puXku sgR78JBnIUKR3tUa+mTJ03JcnP6JgzprIZsIvra8iMA5pGJsXc/aE+MSwC0ME882+wrvN4Ig2B8z P6venPBvFjVJjXFq6Cwfy1/PrKOtEENXYjpsEBaM3shw282SmQOvaCwrX2xDc801r+SNxiS9d/mF 4xA57XsXF+IVT0b0H9aYHTvSElFZXMHyv/DxbPWufPjDwksN9tEtGQLriBwTLFSzJYU0Iu+9z8c7 LQTm7dEuCnfFnzNJyT2LDmoaSjgfLv2zoZD6/ynXEyJMPkMeBFtN5tIlnRQRauaQA54nAFOU+bVt +jrWwWdv0gEiBsr4DpoBS2Ig3ZdoZ4jXkRe2UiphGO+YM1ACCVkWKFkeHA/HHfYpfYQN2nHRl+9E 8jCDJOApZpvqyioCkV/B0MmoBXCHE6Q9E6JC+XgNdQweZVp+ps3dCKc3gc8qDPGazjcK/ou7xd9h wpsCbvMjcez2/s0SzsHekPuY3lxhiuEQFa6oktLmXi6K4aGdCc9XyGW74atDWOWo5P5la23LUOr4 wPa0tXSsicaRjduWRNm67JdxbLqB+fglptRQ4NeuKw4bRremIDNPTEgke5vob9+bobmPkuJfIudt LaVuNIjuJD7wXuNcjR9mB3iFw12jpnlh3FpmWhRWDZaT8FaF/LhIp8CY3PwU9Yn80KcTd3b84k0f dZh+OsHZoU1dNqdtu+sQ8ODiL2yNLD3NLE1quOjl7YvLqvMGaWHkmWqC7BF1gxbQmXH58e0bSqfh /uet711+kmn0+Tg1tOiC4p4o5rvGRJKs9UBPpcibT7PicUifKgBvPklW4df/Bi5HIfNlWYWGuChz YQIXlIIjY/MarRebIgp8zpES5vEnhfDzi8EfTCcz+oqSZrbgnWMwPqvSc5QknZh9/FUCx72z0wDv KZYuLYW8NQzpVSYfhyVeB28e+PgGJg2X7J3uQKmOrhWVudLlT0xoFbx02ylPSxt7uY76k7l2d/Xr Cc1YKzgnzq52OQxavDli5daaTr5l6wlQnb5nphQGldbLlbPCDD4fM9+tEPz6TYNfd09La/iwwuFd r3CQEONEISzzEfw9uj2dADaiG5UMUEJ4coBX75FpVvz3msbuO8Odtt+IeYGPaufNkCOCCmVFOmGx A9oC/CU5Ctdtk17YE3EfmOUf5S5RcH0cM6ktUcoDZq/L9neA715GGfi/KNqauW6tNzURiO2CZBdq LdiL3CHl9ip9No+9rpe8DhM58FSvPSSUJYTcSnJrdrB0wx7KPi2SELbxsZFWD7wzV4vb4eQQzFD0 1v1f6qX641RLQ1qvMJm52FLUyVjiMQ6njc3bEZet4DkacF70T+2gXkUgK9nnkTrVh5mCWdQt9fwV 8w0KZW5kc3RyZWFtDWVuZG9iag0xNDMgMCBvYmo8PC9GaXJzdCA4NjgvTGVuZ3RoIDExNzEvRmls dGVyL0ZsYXRlRGVjb2RlL04gMTAwL1R5cGUvT2JqU3RtL0V4dGVuZHMgMTQyIDAgUj4+c3RyZWFt DQoF96rwf8rBKGOM+ZJN/4Es4XHcDTh7cGy5XWpt6VWn6i+uZsu8OljXk/kHagL5q2iKmi+2yiky 1ix2sYDrMwvoWTHkx3+OcskhXWM54NNm90p6dwAIp+b6m+ITgq76tjS8NHnDjF6+R5ZfUHZ3D8xO aAX3WKP9mZhaTMMR9qgNhxaZbRa5+5k64XOcFUEOhrIqu4L01HHiXQMqEcE0OznaO7J0D1YZ/Ww5 rg6pReOr/3CWuaM4Vf8LDlB7no+N733Lc6u5voRllDbyNfFtUyip2PhdETzfJt2XNCJDtCsHJMTX gMrvfS9WCcfQuvU9nx+LExsDmriKi0gIDkAzzKfRRtWtJ8AlLDsXicfI+vZYwzwieqknRsAjcitB EKJt9Wl4Fzr+HnScqe9OhR/QqGCGdjk2v2/bfeZwXdvUudKB1/g3MhOX8ycN9OmCMSWmc9rcPpQD m0EkuQoFhjSpShcw4u09XQ10eDOxeafvqoy7HU+Nh4DNljx9bPbXQmotQE/Rm4URQNC0wsGYjIrI d22gcvp+p/E+itdxgIojCEzieusdaGwy0hlrG3B0YxklHxTzr1KNKh91a4WdHlRLdjd9HIqNWVSF cF7H2FGwxTGGNcn+HWamaGHR3N3HCYWXhdPhOAlQkpPMvDpKPSL4LtV4dapL/RvKM3QiskXWjYyk AbSpEFUfwl+ms51PGvUNPp9ONrLqJYSKB9hKYRhMWwhR12+vVg1LF7ucDUh++rcoK+dfyfEBt7J+ M1rXiSp28xPwAqizasX816dkfH3Xc1b4Vbzs4OTI6IhkuDsY6ByyK4Q7GL2WcQwP4AOR4lQsA13S bYTbw60Gfhl66gOc9p2sWby+TYcZkGDEgCBK1r/56u65/VBO+r6k+JpP81+rtjn0wvNjk+XOQhc0 2VCcts2xw40hKRBDEY1lqK8ob+J8i4a9cypHLLDIuxFy4mtbG1ecodh6TAj4bTQ+gNhppy08GwL4 VgOSdQ25uHdsUEhstCTVBkhA4iSbi16NyDz5ChmQW05ucca0CXMU2w2Ia+Sol6g9eKqRuC3jriVQ sopfTkAxDhfuVZ/rOJHu9bYapOIq1qICDFl0guq78darBBUOqzox/1p5PfQnwmzlf1UdRM9eZGQP 7UY+bvfSr4rpFKzYPyjlotTtHtzoEC5VKSnE18wTqRnoekJN7yDSTzBSnI3XZT2tYq/o07OXY0Eo fMu7i9nO9sRlWKaUzzQCjHVovjwxX0gJ8vmwRJyDlbkBMQr34G+UeRv5TC+q/PgE1G5uvfxkQqjG Jo5LNfyI9RNIpg+mIbPKuZom7clHMTab2e5RvKOTyV9aKSXDNpxLjxN8bBXBmOR7NSr5GMHwYvnW yx3+iBAgAYigM3tLFe3sOyDvJaWN3mi3uWZDuA6+AXOd1HFQ4WwKyEW7dmaGVR7ljalyyAKWA0ch IOGRe4mX0ETd1ZP7nRr16AjmIEHUu0e8nVUVGrn9XX1lfOozAEe3+blqng/dsHO3ejKTbDhD51NW wWOgjGlm52fjDtRrWjKab8f0gVvJHJDU0zEtd0wA4G2IDQplbmRzdHJlYW0NZW5kb2JqDTE0NCAw IG9iajw8L0ZpcnN0IDEwMS9MZW5ndGggMzI5L0ZpbHRlci9GbGF0ZURlY29kZS9OIDEzL1R5cGUv T2JqU3RtL0V4dGVuZHMgMTQyIDAgUj4+c3RyZWFtDQoft0hDwckmFKl8F4eoygcqIwvMTolZmw04 W441FLYNR1zc6SYDdPjjMlIxiEfYaYp/kpUTbraD6/ryCCj1a3LfIV0K4kBMg36wllJq1jtnY4BF nt87aTNzYOf50QFTDvO/x2HwrCNI4FtBCidbW3U0d142dfU2Dwg9IukD/9DXGrL9q2mDVkES1/l6 OP/Y8XPqd7xT8DmIgDH2fUOxsIvsAs2BxPh+up74iK2omIu/C6Re/YbbZgS8xGv50VGhO++l5ULb BeAHUk+73l6o/VBLv0unwQCHQfE/lSL5MpPzQy0bt1CST4P0SJqaNbtCXX15zUJOcV0o15HQmnao aBfO8FlQ/FCuTBKDynrPjVfYNhP5GJfHF3GHAY0Vrp2ZExiO4YudIHZ7CFXa/u+WIXCOv1hmpXqX ZnRY2kWK61X+HmeZ2az9A0aK4w0KZW5kc3RyZWFtDWVuZG9iag0xNDUgMCBvYmo8PC9MZW5ndGgg NDE0Mi9GaWx0ZXIvRmxhdGVEZWNvZGUvVHlwZS9FbWJlZGRlZEZpbGU+PnN0cmVhbQ0KX/8luV/r fl2cuzb6xo6u+iotQiqChcYP9BC2HgxVmNtVWOut0hmvFY3v6fduswzQi3toG4s+hTFVdt5srS8K 1D7p6aJyetsojDDKBlgiNQgk9dQMefpCJ/nXRUGWLpcDDC2+to0Wki1P/ce1fm0nm4B7IoqZBpWZ UX2mIwTuwJPxkdtjz/BDqCHbSXaMAYyGBOCxGwjHwnYzVDBdZWuQ4GrZFCLDD/IEYwkGDryTb4RF aNt3F95HEjhdzwtyYbVi3kdYsmGRcBGkaysT9L2w0UTwQPyLFM9qXYa4jGMxTQ+4LEL5vNZrarXd 7dRC11feFDAM7b1/503IQrtEgEVoYW3eL+3Hlb8rkYz7EucRo/EsdZnRyGN1bgYzFKf+6XcPBXtu 1Z7EgGW9JnXkOpnJCd2jcJr3PBnC/8yeVruF/5flN04ooT83Dw2Fh7ibtMzSb4lP9zyJFDu19uOd e0MZqUfMg6OEqmeTll3zxrAmDTBHj/EBYBitHt+s+ODo/X6pWl+sdY4jDw8yMlYY/xRVkDDqBhg+ 5EVZ29JuJZw3ODnnQeOn6AkilaOCcTn+4o/BB5eV+meljMxhevp5Vbfm2pk0J2CPR8C87rq1ijym 6kYktJes5SW4kxyZxwcrBJZZZiqGv4uNi2MYsDcMz6j1PzpogR0stc3XVLxB4hwKt6JdZXBGLCHE eiupoH7cvBQJQcb0SRBSO1LGLy4l6XQ5TwSmNuOVQ731biAf+5Y/WfNb9O+nyn7P+wCcsgM2M0DU t9ft8zSkeMvA36oYgCvuGCu8joI05l+4SbstmMZ+QJbZeO1MT0DhNefivitJB6JiHQAE99p928ls 1dljdhLgvQZ8HGQhXLDCANh0Iki0PBfuPDmdb5gd3ymhIH7YEt8KXh6Ux/AV91xV+xBL5P3szjC3 85k+Y4zB20jhWrBOd/PFvOMqSXHEmwA9vZrGeaisGSgQdQCBojNc0+FXhTMyjeT3oIDkUrQQ/eZs Hpsqelu109pwmlRxq3RInpAV2xXC9t/H5DWmZFl9orEIJ2yy4ZKBFuJRxtjS52Iqr6zXUuGn8/Wi DiLXFban4LyZGPgo6wAl/E2Eoy2CDug7hseJFVav/yoRC5S6ri/I2iGzNIHN3rX8F0Sh69m0xxHM 6/FRq2ofPb9MHZRVNdFxupxAu6Ap0EYQ0pj9MofjKqWqW0beYRU6wFsC+WXMZlQq4RHYZODgMS4G J6KBZWBFF7J3a/yX7WAURbJNlNNdT4BEuzozu9Li4YwzT8L5i7Zc2o91MqOJ7+hYX7JwiAMOiqvP KCHthuaBqQbCFctg2pRYWIKUHRvONYxGLqwWmTGfQFjc/3ex9NzhNeBGk6UBmoKPAwPS89A5mzD1 IOMPhwrIZrKrxnWsLOmIP/rZtwpfnak/5QH51+GCdqYHL+r2c2o0FFvO0cXioPxP6H2kZsQ0Bak2 aIjpH0KGbFw/JxzlLPV1UnC8Gmp0kb46xxB7KhD9cbSLzdyJctMKWucf5Qt7+VRBgiWb98+4+Diz lzYJcZIl3VsW59ZxJMYf7DTQxMTppwDkSMKTE/hQvzj9PhQMAGOoenVcsjaMI8kP1yfEhkLoLs6V xaX2HeU9RiGfY5+hXV3uWM5zpQTGM0B69u3Ssg0LxlymXr6BC8xjgsj7gT3un5+wd5yqVziNljDu MDBkNBygKdRarbwbq5aimvHbEDBRu3CUpTk2KxaKutWMNJN2+seq5q2laIbpII1GzsQb6GIWEpnF gHq8EF9HEerhtiEpNKqjMwvQpqIzrIx0UtRZCfNGX2VM7pZ+c3EwGLMzmRiWHu15SeEW2k4hV4yH DiNojEIP1oMToJAs9oJ9biG1UBTghcc+qiA5QuUNuGagGTf6RvjdIin4LlVQtHUBaX1odkVAb+41 Y4eX4zbhkfjpU12EYopJp6AkcxhAyIeYeLlTRYsqiModoVAmMCb/sRMNIO2DEq1I0FFBgA2JiffI fcagwA3W80VG6WTFqy7QJjyL3Co9vXu2pj0520u7sy0J2YShJGC9OMwETksSKWnFlOiyltypTeV8 oh0p+z64v7EKD10lFsY8SnTa+jZzsTjGc4jS4FB3FdJLXapgwTqbmAixRguq9v0v0OMhjZg7NuFR ycx3cjJao/wERBYWmiDbfXQjHVq54GCbyQ7gZg7npQ6OCssPe90LuMDDBvEIjdLIw9A5P5vEltWy hFRh0IU0r5pYbIzYCS/yNQ4OBmlke0EMbo3wfH1rCVdxE6ugUyY5m0YpKs7xyT00U1t075dSoWGQ 8Bfdfvud54bChMFjszIFSx/1Dh/jSXPCOfwhvCgVy8khJTtEvBCTwHMuAI7/GYpNxEKFdrKFE/s+ MXs9Hptg6RExQdZlkZdu359IMg1gdKmeOOldOk9JuvYVTAsU4Oy6B/cGG25Mg26/8yqzpA/vs3qI q89TUC/cbaTSlecwBuqM1nQEXOdh9E9FUJvfkn0oO8fv8cySTwqHjU58KgIQ6q/i2uZL9vZJBEkI LbPXHW48xuD2q8dMXQAPDL0EsHdLG2ytHojI8YserBlOAgs7QN8xC1faSwKv9RN+JuZ/4qi6tJkQ N0uTMTS/Kyzt7kmnfHIKhKyUkrKQUo4Arnwhk7ObCiFDcvrgRIHPnsDgOwqcw32uQHKCiKLzuQsO 5MHvXWjrezFXBH6YmtMRtMUAc4LTdnifNgLtQMrhR7CiLnC6wVnNHU6PHyzPI3RqdIb+xy9kiAIP 5EUjx5fEWj34mZOuCAkF390LOlrMVTpdqgZKpFe24xudOVdQY/xMBeBUBw5MrPp5A1+/y2p5H2rS F4QgdMkPVOte2cY9NNOulTFZr6t8vFHrLQI4OwHxQ/4H3UcTT/mxChsTuuAOFNBBEf2e1DZLc5RJ f3aLZ0lnrhMXNU26KAqQ0XJWNvaigD5fss9g92rY+IiD7K5jDrPv7VHtFJmaEdilCB4FD3HAqPpv yL7ZjBm+8TE1gdnkAm6RK5exCSgCRHf2EtmtDPfgBKtr+W+03UCSGIvt/GcpV25NW9HpQfjaGNZI gb2CJy2Crl1QekxnU7f3uThL5+ZBXTNsFmkxcOKfhk3oSP6q+0bNRsIX2bBZakDE8qtvDuyJkA+F sFUOGIljHQjmmdrkkQoxiW3OvVm/p0DX05XGD3QlFcd7fA5s7qwmS759L25LS9X2wdEVjZH0A8HU gZePCuJkHy4jBdlr0Rvyw0b4fHvaweHfnQXrzExzhgaezYVmCzRUC2JaGk6/anCVlZiQvZ2eDFKy VhBXcmh700a3PInYuMH0kj2zbb7QZKSTjY0xZ+U6qyj9KAMto+UecgxaiSzQsllP7IBLbbKCwLSr j00aCzGVt1rT0rB906ZPAbPqIrCwPhbIxrNDqKpRDNowtSfz9Q7Lisv7asO3AHOPPLKMt0bupwjd ButHA0fNkPhgQ16qju0mpjx7EK3o5nAQiKzq8fRlKnRvERx3HkmNmLOg0f1r9BPeym4WKzAcjfT4 X7LNLsN5REYrjNTxRK/8U8OwP6GS3NPW1K373eU754pAz3ep0Tm+rfSOf9FNy+kyCIsETFoSFtFa YsgxWgaxr33I8H5qwHlt2eNvCoC3arNX8oPr5Oje/Ox1ME2c59kQtQ8dMOK8K1PfGaHUsb5PaZZH RoBx2cZ3ZuJ3z4O8eKgPcqWZ+Gjy18Avv7g1qvRk6y79rSbKfGlogviRdXTWq14Xz67Tn72nubbq l/hVXdrd7v5GoEYUYf13JsvQgTCw0YzegS/sn25S4pnO8engXJ1Vhw0lGx+OR0fZaMya3ishJd8v OGt+AEwQeQzjV0JwIT4z2cEc34v4PA6oKJkpMIs/t32QQOieknXnRLNNYS2fOMtbm6CzWICJUHQ0 Y/XneX098uHE8lejJdANk+JLSnSWa27ml79sCxx4XO/Tm30/gGry/KA3AOCW9KIyLY2MHhv/1ZQd uziixB/J9qBPrOpmih6DWLJUEKIulDth2ba2oMwVcWT/zldlrW0+O83s/XBRCwJTmfOabCq6V9/+ /9zP7uLIhnfbh+fzCpEtj9F+/LCTyDwmynfU6/HfA/OV3MqQMWv1bVzBo1b41ULMfKiuwBmiHQSs nvFs+3JvafCVjpSEXLqs0xuSMFiL4VnBTTHfgmTmhbSnhBuoie6ATVrpwDtEBPyCKYC3KHDHaQ/7 78MxbzQMDgzpKQSqoQ7U0A3q+68Nrj5I/dnbp6B3h3RiNoxi2QtnXs91jxITNDetkyB3uO/bBUuM g2/1Cr9M48YPCC1Z+K1f1EXV035e9UBXMCXD0gmkR+63T6UBgnC0xGnQWklUd5ZAO1ts7pBpRnyl b16OIfmTlJQ2qAsjf26+wUsKMCqvU8lsWBiVFQaigJ2yhtFKXrBH7R9ma3HBPrxFNCVaRczSLXs8 lD0Rl5O8ovrr/iNA4Zui3hPEjW3MFji9B5/O9CyNghN0mWrBADfd2sC+6Mu4j3vpVqnygY8QJRhy ZZPK5S9951RQ6/NALw/yO6ZVarbnWHB0fbQ9fxBtjwfBFixeDGYsYgwwoYTEDNaaKnrhtstS+pFQ 1rh9XtioC8LzUN31S6BGRXrHlRiAh6sMdRUaTPGu5w6VIm8o7EquxrU6npY3NtCL7xcRX9x6Imsq euO+KzoAqTLG69tpuvOA1DHtZURbeYnRJ5HTW1poGJDS0Y5OteVu9jH7KyjS5vBp3FTXxX1ImveU DoqDa0vbgMl8VVJssdqWP+zpwmSn+s7opePHTGuDtIPVUC74Ftsuw8MvDepD1w1qIVE+cVF3lxZ7 fd+QSoun8foMF6g1HaFqYaSRC7h0XBI8RFo5rVLDL/fD1QZH1h7uMRW9peQPVtQFohiVDgzRDbNu pV3OlvsYl0Y2YHSjIWI769Cu72VuBXFFXgy7Ek0s6vPb2ryLYuP9xjeTY+7KgfrQx4oLiT/PN4j2 eCyoI0h7XV1lHqAgLipX3pMBr6c/pJoxYDr6mV0PVwk6FI4RS2uUwJLyqDBuMV1186YH6PcAnCWj Mglexjl8wwqv7+CcDNddDMuqURR7Nf7h4jzHG4QTH8TJaYuxycT3JVaKn96fo5JVVDs0Za1Tu3Vq r2oTexKxr9Y85mxz4gOcA+FPvBYxHo0ik1fTglu95ut7jIm0/ExtB/d2/PxOZ2skRUHNF5MoUftJ KEdFPPX/8YrypjuxLiGKBNGKpNiQ23Y+D+EyA9fWwnFasFYRnEPuNegE8yX3oluIAcTV0tpq91V2 gzsNpXFFzLUqxxGALxpuJGWEUJoYruinM+0oS/G410RtAjUAmQx0U+AvCWSj9zzLGh8nsmyZjLaS GgHLOqcc+uJn2myZg2eOMczGzxSPZ3CVgDCjebcQOKFWPqsQFQy94OgS/1atqQKNv38Odu0EAFVH 11NKTSglpg7fnVxCvULuxdK20cAWynYPQSI8fBQNJ+a6oq6m0jKOFumI+PKpJsrGDD531054JqIw 3Bva39JpUrIvBXQdtgPZflA7ORwPXD799q4/9hUtNe8NCmVuZHN0cmVhbQ1lbmRvYmoNMTQ2IDAg b2JqPDwvTnVtc1swIDE0NyAwIFJdPj4NZW5kb2JqDTE0NyAwIG9iajw8L1MvRD4+DWVuZG9iag0x NDggMCBvYmo8PC9Db3VudCA5L1R5cGUvUGFnZXMvS2lkc1szNzIgMCBSIDEgMCBSIDQ1IDAgUiA0 NyAwIFIgNDkgMCBSIDUxIDAgUiA1MyAwIFIgMTM3IDAgUiAxMzkgMCBSXT4+DWVuZG9iag0xNDkg MCBvYmo8PC9LaWRzWzE1MCAwIFJdPj4NZW5kb2JqDTE1MCAwIG9iajw8L0xpbWl0c1soMTOD+TYa VHe4LDKeYbJSRjDYiBHSR4n3VdndT83CPMV/sOy78qdcbhQE5O1S6CPfr9TKPAJKRgfe+T9+wyko MTOD+TYaVHe4LDKeYbJSRjDYiBHSR4n3VdndT83CPMV/sOy78qdcbhQE5O1S6CPfr9TKPAJKRgfe +T9+wyldL05hbWVzWygxM4P5NhpUd7gsMp5hslJGMNiIEdJHifdV2d1PzcI8xX+w7Lvyp1xuFATk 7VLoI9+v1Mo8AkpGB975P37DKTE1MSAwIFJdPj4NZW5kb2JqDTE1MSAwIG9iajw8L0YoYIFF2XuH TW2cGvoEI4e7npDxlAizEWB059MQEH0pL0VGPDwvRiAxNDUgMCBSPj4vVHlwZS9GaWxlc3BlYz4+ DWVuZG9iag0xNTIgMCBvYmo8PC9TdWJ0eXBlL1hNTC9MZW5ndGggNDczMS9UeXBlL01ldGFkYXRh Pj5zdHJlYW0NCsj4MCL4hEuMXZvnXtzK0+Ac7Olmiyu9aIMFqxWvC9KX5C3czSW6IsTxbEO45l4O ZznAU6M2Oi2LWzMREV+XIwZHrecE15A/AibRwIZRgJE12c8snBn3BL1TIw3yASAN5fHu17OViicr u9E88MxZg4n2hwYlgiydB8L4qjfqSEuKitI7e/R4VK+5mVCxu0yApVpUGQA73FtufE/UdEuREG/k 4Tg03or/lTq9JPWfvO22cEougFEKkD8rRVq5gwOBCA1rVOCNNCxaSltvnA3JduUJabvRbwnLtY0q FqWh3884tNLn902QBSR9J4CbOaKsx2l0oMfWmJqkaADzrJQ7m6niH2EFAYjHuhgOahbasj8WV24J dq+DJBkkznj9KsS52p3lOqfHPC5fKSXJHikxeSXs0NwsE4TatNJSSAthbzBBXeRc469jTmRrbVon GPAYNtRhD55sF/J2o+ZIb19zXCnKkwQlppJdtxRL1T6OGG+Zc8JnLkz5fLEt/PvDddWE1JRlgAZK gfkcx4OpZt5EKQdg6daR8xZBHRjc0YykDXu/nKOkQOae3wUAMBk+c6DlJFxWhirvu9up2rhdZiGj Js9z1HcftP+NSUzvk9wKdQdw28pDrU8c/Hn+FDUGECDvEZo4OgNAcWm1m3iX7KgsDKkTVH+h47I4 AdbRH6voHkZZuMa2ilhzVcncGYZi2GmpK/KyBVUSlFon5LKURpQ1/nvKZ7PgkxQulnCUKESI93MS gRe6KWz25j8JJTMitoQl2c69EraujBNWkcYaa8OCrezfs9j93ZSROhelORUpRMsenbV/+w1ggjVU rTBXWL7jW5jtxNBWKvuP7a+xLLGYz+pNg2gnHm8W+kGbpPFOSwv/VIE9JMeOnWeAWW4q4hMefwcr yiDOje56WoZNPc23V/fgSItcMhk6MQc4b6w8kuWt3j412nd6vgScPxVM4qVjWuB5uc8sIOhkzL/0 JZyPuxAgGGZHnSthbVUtiIsBTM7RA0bvZVRWhUAa1bWseKY6lypODH2XSsuK90lvA95jdlUr/5hw Ut5z0P/nvVQzaMqdIkjGS9esBAIv88n3ku/918bwi0tVmmeZM5ScF9/nOGWADDh7X5gfcmcJKkEe Fsb9C/zI+I+P356xHGcmlJYH1CoQM0FcEj/VczQxtXD+At1TxON1NhzV2kbECF8UzWbFBEeelY+r pF6E/WkLJsNADOs8NAfj4ZNtVt0d+W5iDOdYibB6qxW00ZFrohdzbICYYFGwG62zImAid5P37qlT oALt6BkAkBHOewIDif7NSLSharljJ7KG0cjqJKiRNh1UA3nSLvA9RAz0F+NyVtuv/W0NHqcEXD3v EYgmXQ1PcDodLEoiYOZdKoICLztUvngIlPi+Z7hc78sZVQ0vV7eO4tLUg+VSIvPqf3bL83VAhOn9 mAFWCAIAIHkVdV7J8hiaS4wAiKsgXRFO2wUEwxkUhBBrdOFvUacVX+hV4erXOKFwiPlWg63GJ5+m LJvlsFJGiMIO5xiKFx1hsn5oCgWpz5DIiKZs5j4JY3NWOAk9FoOt8LF7YVAJo2nC/xdqWX8S1blc w8RU/pOmSWGrrmcO+jyECfwv/11hqregkwd9SeQhSqdcLA4YusQnb+v52xuBlkAoAwEUfuiPRxzC V+evOBIDkruNJWQRLqozariMNlIK4PRaPA6VvdJPh7smBIgYfSoO/LQ1cIq8eDTakVHLMScIJUz+ 7T5AxwF4Na8rF6wMeBvS53fzEp06VMNPHKz4aUdqHdAwTOyixVzicu8q4o0JQvX9YpIH8VRSOEhQ leA+93yP7ozjfF3RAMF6JnRanoOvduFvSHcsodO93N73zZ5y2ZDl0ddrmWlzBpjup41SURmCTkNj ZedoB/btxx2IjOzvZGY3piHuPWK2Fs83pwBoLZPZgCTsUuXBYY0rDHc1JEAcI8M/rrzyY2yTaoqa iH5iuEvHbZEM8XhDoPXIDvheptNywx8prl+JifMuKSjd1Mh8J6ueE9GMxx/+k6YJTbN9wVBiuJXK keWW82dRJ0S8c92mmEr/y5xqBgdajeT0DX3vjzAw+Ihd/rwpLm/LjJtgbvyDh671pgh1ytg5bnpw 73d7pFmF3uIBJIvvRFbzX8xx7a9iay64x59CXooDJgSdEZ6igQtura/Cg/lIQ1pXY5LSAhXYS3HY x13rXJq0bHrDhLZ0nOz5LSHo5YBU0roBOJ9z7qn7WZmkCPZzvXv5sA77L9uB38WsKg+WrdJXcSaO UgBp5yeBzBKbUJQvu4iC6fnUTBbbB5rBhjIvcsbstcfkKkhBNq9rGLEXtUe1ZOx3jbQ4W/nusmXs wQcTq9jcqWdMU1oVDtGA4wghIE8VqUnC2im0dWvTLW/KOiHIM0YL3H+F5SVvwt/g4rfmecfjz/R3 InJAJyTLk8GKEgpcG05DKuF+n/ih5z60e9Qh20PK8IqEB+4fcYgjMh9AKoKxtrVRJg9xArsSwgKi WAsEVKZDL8GIp2rphCRJlD1EBcBrcBygt1sbQadWWwwhSgcfcKuw/pLDOZKHLnEz0gwWliQNP0/N Pl3sMP/MtdVJKrJ+RiUVFLftBTNVKuZOW30bmgVf7JYWo5Np6zgDabHN2EIoziQ/5dB34WElofJk xqj5J/GKH/rNYEckzp/JcMlcpVUG7f83iGQi8QxO++SX/qwm2clGbP/LglYmTfUgWprMoM9tbxWJ xLdUn2vf3BPQZ5hHYSc82obSpvU1+0iEmA2wIwrzlQCPZL/rUpVpEXyzMKJiHVABJ8bM9jbSbW0t NbjVb7tZ8dy+2ImTMa0O6uQ7zCx01e4DCVkCZVGZRH3kfRCBTe2c+XfIxP5WiiQeOMiH8B0lu9gt SiqnLv4mAtaDXct8JQh+dYq2OcljJ7C03x5U2T9fT7TYZDTsarflS1+o073cA0pr5Ma9sU2F2vGk OwIkChHX1r8TiM26VOCfKpuiGlkhilzyovxV1FhWYysERzQRFkKCW+tlsuJTQVFqOvgAyMVNWsd1 kTF0tgV39h3BQwKjrzI2gTA9iFiOE5RnrbqSF/65hpDr1WQpOAQKLVQWY3rL08dMmBVrzpWtjY48 J3jhRN/dyIZMSVZ357MpuEUT8NL2bx7GAg4CEKys3gYeskHY46+ANCe85+CPGbBpH7eamNjuBjBu MH+JRZiCT58iHBgNOPT1ZzMiXaehOGzj/vKoAII8+qxSY4mi221PR+gVN3iHuWaYPOAEsNOp5p0C +mtjsr8GvhzRBSrOoLbb/E9TW/NxM/LlQWnaAU/bCQfBysMl5gWGUA5uyKF5TjiGlmYnR9EQVNMr 9ueXdxzMJtiWUMpw7Nts0CSsv4w5bqOHpAWSkCW2fjYd1TEngg4RBqe6PkH/QzKw3G1NUn2DUNrr bf99grArXJFZVeSD6pyzcASDPSPKDZ5oQkwMHlBBo2BsZvCbwh9bqSK1/XCJ9gndviQMQu3xdxD3 vYYT6vKvD/ApzGp7+acFSeYOQAqYpO0XDJGLbJ22xYi3e/riOkJkuMwekn/s2DCzxcFmzi27ATsz HKghFwriGl9nhs0P6deqpkDYwEYybBGu4jDsLDXuJdKra9KCDJHd9a6/yhb6KhlR2BXhbBhWPj7x i4OP4bwvbKqyGO5c9nKhZi8ppeaxbGGDBns4kMHvPDvN95yQZmHUO4xID6/lNAByNhSGFZO/BfB5 JDt/W1+p7KIDbuGox5D4hhDdFfF72ch0X0pm0psI3r2k643ebohk9Ew8/0VGHEcHkp3eE4JCpsIv fWYZFiII05T3UamLGO33NF7RGYWARDN/fFqaXpBj12hlXnP4h7y8a9vKjOg/WV8TxIWfRk6bUzmu g8i0YM/jCRIeG+QNxun5Anqzfc71SKfX+g+HqfjvG6T80Td8pbCZcI7K9nMTC70I/dp7Q8I2dKGB tuQIo0yk3sGnYyIp2RHjw9hqrIX55+WNLpyAc90+jOmt7+F3M9OosL8QorWJymOmlZNSumTuSMsY amH9XLDLmRmPbh3nZep/GgzE/+10hodylDlLxiXNvpbJqITV9sqWKtYxbchxS6aTf20aHM2+yfFK XkdX715LHrjxePkcUrMIJmCG1SjuLygU55ObJZRE2Xc11aVcojnrQMCaGISCQGX3M1wholxh6m1x USR3kf73Tj1SdQVlF+hq8f18wCNtL1khGxGS4tlSa5gttQqXEvPEVGIlmnEOS7ceApDFV4MxSc2P 9eeUPn716WXN1Q8bWggEcvx/6WujFyqAaXeMMlr62Y94vXFh83d/VtT1cotWIJs91gfHfLWjcfx8 b+vGl2ySfZxV6MNlXxCKotDOLJ5vPfOV6O/MxE5DXvYrYz3XZhwEN1xqT9IuqSGDsdIKSVlLMenF gYQVCnRKPQlUeIS+FK5pzePE1ub+oNZ5KF+m0AOcw0MOyCD9Ys+BeM0Bdckjef5qQe6uNTVRNdeR cerUG1eLHKeuMQLHh49U4YtozePNLAcCSphoVZA6SUGA1LEixU+x0gBJRKLz9HusnAJQlGiMHrMi auYic97YbnpkmVCtv8UbKQuW2lYL7uoOk2OnxRp7u/qOIJxJ2bD01RdOF/D2wyn9BI74JGoBzbBN rFlE+/QdueGD9/CbMV1Pe4lj9jSHkjnxPM6YJFcXXpo/5j6z7Ly33EHeRWNK+RbfwCufFXMwyKwJ crEWCa6wuz2pW39ObZ3P+dKiHjT3f12fCLOQwaWCiWQETQd1BuCWI82EVUOeAoWkuo5cQyFK+Roz swOI4AkeX43gLzxTVVbgGdCOG5sJBMLMfG96zLKsY6CR0EvS47Gar/A9mdWfnhjeo7qvnVTw4Wll 6yoH3QGhU6ytFr9Le2ljW+0SBJvJa5T20vEZRXjZvOaQuqQ1zspUEzfzoUnjkZJhACVeoN4XVqcO ewTOJJpXAhf+x8WM8loWHZ3X8Gl5juMuFitKDkBXPU9ETQpkbdw0t5FBYjlzOF2qJ3BtqpUgNyN5 iOkr9cESwt+JX/Rvf0lqLT/lgmV2Mpa0e0+e4HwNvhngL35JPxImiry1yzziUTPMvJlxj31H4cc1 1C9LUG3J5C7MisBsivheXM3S+HiVqIDBCkNzmADwPy3Qj95U2oE1W15/N4nTfbO32sSTNGAvfxLO oc4pQwjSfQcjolyqysByyx96aEMg8DNsPu+F/A/RuAN2zYhXBr6S9A33e+vSZa8YccqLgdHyW+b+ DdxPWlaxtTa4CR7at+S/fLXgG/2tUYfvOHgNyJG4UIeHlazcz5BNfK7luMjfTyl2nvWY3vE/rgFY SOmPj5dw4KMbTk6wS4I2amQx453+9aDozkLbCM62l7VFi1JZ2wn5Ajhfg0Rp2RCwNBiNd1rzLKzg XAUSrpTCYj7lE6rmFOrSZ7Ep3+EA//JHBBtEqF/EiPXMFfNVgg85q9LR+4QFQAmgolNKDCh7uHMl WfQdlZu73ZDFKWQwKV9mXrp9k0GgfpSSON1X8L3ONXj4uiXyZcB1subzyyrnViKhqteasKdEKL24 DeoI6A+/LyNFQ0x/JPkZSHYjqed8EGLelC+9+F0GYY6e8cLJeNKEc/sysodhK1ySRU2DI9ck1pcE 58r21TTPZmsLTkTVnjMDf4hf1Cr1UWDhPX8O8EvG2fFmBG5vx0QmfOJKwgtPWnSJaoz5MDsZr3+N WCrLOuvMUh/n05WY60qm7AsCF73KqDE+UNR5jPxzbni/EVNhcGaE1KEPsr99Q7FRb3mcCPmH6cC8 3nhDESXed0pZfzKhqk6j1xLOLp6EgNAbRLmoBVqAWw40n3hi+stZH5LG3c3Oo730v1iwwKf7JzLt WUk7FCxslq6LAEbMePzJqUXBjOG3DGaJgp8PE8rZK6d53Z3kJPxSnYSbduvNVK3IUawyFeMW5RIw DG0rFNG70G3WTCt2EnDN12lIbdbSqiDWQnde2R6IleZ73D2Nj6XK5ViKvTRcflEgdLT2XE11TmiV wfV/XPn10OAn4XHHWS7IkJxftVD7ZK21qN54ZKskbQMaGsBySt0MaYQelzvd58VlXYnMOr+sq5NB 6ql2u9/b7hOOyQZRsAjeadqcJ5IB5QdB91cJ0mdsFYviVLkk1IoxhK2R5yqB3NkIDbzrY4N1tpRN yKqqIiIze4960daGHX/uX0UFuSs9bF+w6feIClvODG2mywY022RB6KT3Pk2BoKczRKKuKATWYWSY 9ZWwojKEZ0PvepOTiKGmoa9hbOkS7Q5FfCpeQdcSozIy3YP5bDtq4tFprrPR04vfOmVMoEJobDVK Q6Rnm9TtwWPHOQ0KZW5kc3RyZWFtDWVuZG9iag0xNTMgMCBvYmo8PC9DcmVhdGlvbkRhdGUoyO3l AY8U2Nd50TrUilo1C11eRG6+gi4pL1N1YmplY3QoxYf3fNBAgY8hlnIpL0F1dGhvcijIorsdn2OG hzqHfMH0KS9DcmVhdG9yKM20pV7dQ5zDGKZNrNkHYkBQWV15oIVcKUxsLE2nzJLeKS9LZXl3b3Jk cyjlp/tcXNBAgY8hlnLN2gt3HgIBBj3n3G4pL1Byb2R1Y2VyKM20pV7dQ5zDDIt4ldEAa1cCTkRn vpw8XG4rCQSex4/N4KgpL01vZERhdGUoyO3lAY8U2Nd50TrUils1BV1eRG6+gi4pL0NvbXBhbnko 2L+yEf1NjYomhSui1wF3Ux4XKS9NYW5hZ2VyKMm2pEXYQ5qHZMJfjtUpL1NvdXJjZU1vZGlmaWVk KMjt5QGPFNjXedE504paNgMpL1RpdGxlKMu7uFPeTsiqGMJFhMwbaEAbTj4m7NtlQ3cnTYXQidT0 oeB0prM2/p97x8apB3E+mIB2k24XQXAYKT4+DWVuZG9iag14cmVmDQowIDM2Nw0KMDAwMDAwMDAw MCA2NTUzNSBmDQowMDAwMTY2NDQ3IDAwMDAwIG4NCjAwMDAxNjcxNzQgMDAwMDAgbg0KMDAwMDE3 OTE0NiAwMDAwMCBuDQowMDAwMTc5ODE5IDAwMDAwIG4NCjAwMDAxODAxNzAgMDAwMDAgbg0KMDAw MDE4MDg2NiAwMDAwMCBuDQowMDAwMTgxMTA1IDAwMDAwIG4NCjAwMDAxODE1OTYgMDAwMDAgbg0K MDAwMDE4MTkxMiAwMDAwMCBuDQowMDAwMTgyNTUyIDAwMDAwIG4NCjAwMDAxODMwOTggMDAwMDAg bg0KMDAwMDE4Mzc5OSAwMDAwMCBuDQowMDAwMTg0Mzk3IDAwMDAwIG4NCjAwMDAxODQ5MjcgMDAw MDAgbg0KMDAwMDE4NTMxNyAwMDAwMCBuDQowMDAwMTk4MDk1IDAwMDAwIG4NCjAwMDAxOTgzMTIg MDAwMDAgbg0KMDAwMDE5ODkwMiAwMDAwMCBuDQowMDAwMTk5NDA4IDAwMDAwIG4NCjAwMDAxOTk3 MjQgMDAwMDAgbg0KMDAwMDIwMDQyNSAwMDAwMCBuDQowMDAwMjAxMDczIDAwMDAwIG4NCjAwMDAy MDE1MzIgMDAwMDAgbg0KMDAwMDIwMjM4MSAwMDAwMCBuDQowMDAwMjAyODExIDAwMDAwIG4NCjAw MDAyMDM2MTQgMDAwMDAgbg0KMDAwMDIwMzg3MiAwMDAwMCBuDQowMDAwMjA0NjEzIDAwMDAwIG4N CjAwMDAyMDUxMDAgMDAwMDAgbg0KMDAwMDIwNTQ3NSAwMDAwMCBuDQowMDAwMjA2MTI1IDAwMDAw IG4NCjAwMDAyMDY0OTkgMDAwMDAgbg0KMDAwMDIwNzE0NiAwMDAwMCBuDQowMDAwMjA3ODAyIDAw MDAwIG4NCjAwMDAyMDgxMTMgMDAwMDAgbg0KMDAwMDIwODI2OSAwMDAwMCBuDQowMDAwMjA4MzE1 IDAwMDAwIG4NCjAwMDAyMDg1MzcgMDAwMDAgbg0KMDAwMDIwODY2MSAwMDAwMCBuDQowMDAwMjA4 OTMyIDAwMDAwIG4NCjAwMDAyMDkxMTIgMDAwMDAgbg0KMDAwMDIwOTE1OCAwMDAwMCBuDQowMDAw MjA5MzgwIDAwMDAwIG4NCjAwMDAyMDk1MDQgMDAwMDAgbg0KMDAwMDIwOTc4NSAwMDAwMCBuDQow MDAwMjEwMDI5IDAwMDAwIG4NCjAwMDAyMTU4MjcgMDAwMDAgbg0KMDAwMDIxNjA5OSAwMDAwMCBu DQowMDAwMjI5NjY4IDAwMDAwIG4NCjAwMDAyMjk5MTIgMDAwMDAgbg0KMDAwMDIzNjQxMCAwMDAw MCBuDQowMDAwMjM2NjY2IDAwMDAwIG4NCjAwMDAyNDI3NTIgMDAwMDAgbg0KMDAwMDI0NDA0MiAw MDAwMCBuDQowMDAwMjYxMjM1IDAwMDAwIG4NCjAwMDAyNjIwMDcgMDAwMDAgbg0KMDAwMDI2MjQ1 NCAwMDAwMCBuDQowMDAwMjYzMjk0IDAwMDAwIG4NCjAwMDAyNjM2MTEgMDAwMDAgbg0KMDAwMDI2 NDUzMiAwMDAwMCBuDQowMDAwMjY0OTYzIDAwMDAwIG4NCjAwMDAyNjU4NDcgMDAwMDAgbg0KMDAw MDI2NjE3MyAwMDAwMCBuDQowMDAwMjY2OTk5IDAwMDAwIG4NCjAwMDAyNjc0MTggMDAwMDAgbg0K MDAwMDI2Nzg3OCAwMDAwMCBuDQowMDAwMjY4MDU1IDAwMDAwIG4NCjAwMDAyNjg1ODEgMDAwMDAg bg0KMDAwMDI2ODg3MSAwMDAwMCBuDQowMDAwMjY5MDY2IDAwMDAwIG4NCjAwMDAyNjk0MDcgMDAw MDAgbg0KMDAwMDI3MDA2MyAwMDAwMCBuDQowMDAwMjcwNzY2IDAwMDAwIG4NCjAwMDAyNzE0OTUg MDAwMDAgbg0KMDAwMDI3MTY4NiAwMDAwMCBuDQowMDAwMjcyMDM2IDAwMDAwIG4NCjAwMDAyNzI1 NjMgMDAwMDAgbg0KMDAwMDI3MzIxOSAwMDAwMCBuDQowMDAwMjczOTE5IDAwMDAwIG4NCjAwMDAy NzQ1MjQgMDAwMDAgbg0KMDAwMDI3NDkxOSAwMDAwMCBuDQowMDAwMjc1NjA3IDAwMDAwIG4NCjAw MDAyNzU4OTUgMDAwMDAgbg0KMDAwMDI3Njc3OSAwMDAwMCBuDQowMDAwMjc3MTA1IDAwMDAwIG4N CjAwMDAyNzc0NjggMDAwMDAgbg0KMDAwMDI3NzYzMyAwMDAwMCBuDQowMDAwMjc3OTYxIDAwMDAw IG4NCjAwMDAyNzgyNDcgMDAwMDAgbg0KMDAwMDI3ODU4NiAwMDAwMCBuDQowMDAwMjc4ODM0IDAw MDAwIG4NCjAwMDAyNzkyNzYgMDAwMDAgbg0KMDAwMDI3OTkzNiAwMDAwMCBuDQowMDAwMjgwMjAz IDAwMDAwIG4NCjAwMDAyODEyMjUgMDAwMDAgbg0KMDAwMDI4MTU4NiAwMDAwMCBuDQowMDAwMjgy MDM0IDAwMDAwIG4NCjAwMDAyODI0OTEgMDAwMDAgbg0KMDAwMDI4MzA5NSAwMDAwMCBuDQowMDAw MjgzNjk1IDAwMDAwIG4NCjAwMDAyODM4NjMgMDAwMDAgbg0KMDAwMDI4NDQzNiAwMDAwMCBuDQow MDAwMjg1MjYzIDAwMDAwIG4NCjAwMDAyODU3NjkgMDAwMDAgbg0KMDAwMDI4NjA2MSAwMDAwMCBu DQowMDAwMjg2NTAwIDAwMDAwIG4NCjAwMDAyODc0NTkgMDAwMDAgbg0KMDAwMDI4NzkxOCAwMDAw MCBuDQowMDAwMjg4MjI4IDAwMDAwIG4NCjAwMDAyODg2OTAgMDAwMDAgbg0KMDAwMDI4OTUyMCAw MDAwMCBuDQowMDAwMjkwMTM4IDAwMDAwIG4NCjAwMDAyOTA0ODcgMDAwMDAgbg0KMDAwMDI5MDk2 MyAwMDAwMCBuDQowMDAwMjkxMzYwIDAwMDAwIG4NCjAwMDAyOTE1NDUgMDAwMDAgbg0KMDAwMDI5 MTc0OSAwMDAwMCBuDQowMDAwMjkyMDU1IDAwMDAwIG4NCjAwMDAyOTI0MTkgMDAwMDAgbg0KMDAw MDI5Mjc2MCAwMDAwMCBuDQowMDAwMjkzMzE3IDAwMDAwIG4NCjAwMDAyOTM5MTcgMDAwMDAgbg0K MDAwMDI5NDc0MyAwMDAwMCBuDQowMDAwMjk1NjQ2IDAwMDAwIG4NCjAwMDAyOTYxOTkgMDAwMDAg bg0KMDAwMDI5NzEyMiAwMDAwMCBuDQowMDAwMjk3NTU1IDAwMDAwIG4NCjAwMDAyOTc5MzggMDAw MDAgbg0KMDAwMDI5ODM1MSAwMDAwMCBuDQowMDAwMjk4NzM5IDAwMDAwIG4NCjAwMDAyOTkxMTgg MDAwMDAgbg0KMDAwMDI5OTQ5NyAwMDAwMCBuDQowMDAwMjk5ODYxIDAwMDAwIG4NCjAwMDAzMDAx MzMgMDAwMDAgbg0KMDAwMDMwMDMxNCAwMDAwMCBuDQowMDAwMzAxMDQwIDAwMDAwIG4NCjAwMDAz MDEyMzMgMDAwMDAgbg0KMDAwMDMwMTUwNyAwMDAwMCBuDQowMDAwMzE3Mjk4IDAwMDAwIG4NCjAw MDAzMTc1NDQgMDAwMDAgbg0KMDAwMDMyMzI4NiAwMDAwMCBuDQowMDAwMzIzMzExIDAwMDAwIG4N CjAwMDAzMjUwOTEgMDAwMDAgbg0KMDAwMDMyNjM3NyAwMDAwMCBuDQowMDAwMzI2ODE5IDAwMDAw IG4NCjAwMDAzMzEwNTAgMDAwMDAgbg0KMDAwMDMzMTA4NyAwMDAwMCBuDQowMDAwMzMxMTEyIDAw MDAwIG4NCjAwMDAzMzEyMjMgMDAwMDAgbg0KMDAwMDMzMTI1OCAwMDAwMCBuDQowMDAwMzMxNDky IDAwMDAwIG4NCjAwMDAzMzE1NzcgMDAwMDAgbg0KMDAwMDMzNjM4NiAwMDAwMCBuDQowMDAwMDAw MDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAw MDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQow MDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBm DQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUz NSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2 NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAw MCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAw MDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAw MDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0K MDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUg Zg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1 MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAg NjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAw MDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAw MDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAw MDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYN CjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1 IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1 NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAw IDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAw MDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAw MDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQow MDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBm DQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUz NSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2 NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAw MCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAw MDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAw MDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0K MDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUg Zg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1 MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAg NjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAw MDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAw MDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAw MDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYN CjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1 IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1 NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAw IDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAw MDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAw MDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQow MDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBm DQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUz NSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2 NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAw MCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAw MDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAw MDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0K MDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUg Zg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1 MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAg NjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAw MDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAw MDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAw MDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYN CjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1 IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1 NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAw IDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAw MDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAw MDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQow MDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBm DQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUz NSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2 NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAw MCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAwMDAw MDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0KMDAw MDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUgZg0K MDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1MzUg Zg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAgNjU1 MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAwMDAg NjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAwMDAw MDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQowMDAwMDAwMDAwIDY1NTM1IGYNCjAwMDAw MDAwMDAgNjU1MzUgZg0KMDAwMDAwMDAwMCA2NTUzNSBmDQp0cmFpbGVyDQo8PC9TaXplIDM2Ny9F bmNyeXB0IDM2OCAwIFI+Pg0Kc3RhcnR4cmVmDQoxMTYNCiUlRU9GDQo= ------_=_NextPart_001_01C65FF9.DF73011A-- From nemo-bounces@ietf.org Sun Apr 16 04:06:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FV2HQ-0003uB-VQ; Sun, 16 Apr 2006 04:06:48 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FV2HP-0003u6-S9 for nemo@ietf.org; Sun, 16 Apr 2006 04:06:47 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FV2HP-0001ai-Ac for nemo@ietf.org; Sun, 16 Apr 2006 04:06:47 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 16 Apr 2006 10:06:47 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3G86jUE008867; Sun, 16 Apr 2006 10:06:45 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Sun, 16 Apr 2006 10:06:45 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About route aggregation in global HAHA(was Re:draft-thubert-nemo-global-haha-01.txt (was:[nemo] Charter proposal) Date: Sun, 16 Apr 2006 10:06:29 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC020EA5E2@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About route aggregation in global HAHA(was Re:draft-thubert-nemo-global-haha-01.txt (was:[nemo] Charter proposal) Thread-Index: AcZcbiA+h592IJ3RRjqcHSMAEKyQ5ADeKtsgAFCtm+A= From: "Pascal Thubert \(pthubert\)" To: "Dul, Andrew L" , "marcelo bagnulo braun" , "Alexandru Petrescu" X-OriginalArrivalTime: 16 Apr 2006 08:06:45.0779 (UTC) FILETIME=[B4493630:01C6612C] X-Spam-Score: 0.0 (/) X-Scan-Signature: 36c793b20164cfe75332aa66ddb21196 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >> well, we still need to inject one bgp route per distributed >> home network, right? > >I think I'm getting confused on the terminology. I tried going back to the nemo-terminology draft >but that didn't help... :( [Pascal] This one comes from HAHA. HAHA removes the home link centric behaviour which attaches home to a link. Some home is now distributed, but it is still one Home (one SP say) as opposed to a consortium formed by the aggregation of ISPs. > >We need to inject at least 1 BGP route. Theoretically all mobile nodes could be aggregated into a >single route, however that isn't likely. The best case that I see happening is an aggregated prefix >per "mobile-network operator". Where a "mobile network operator" is responsible for the provisioning >an providing service to a large number of mobile platforms. > >Make sense? > [Pascal] Well, the hope is to find a place somewhere between the 2, closer to the one route model, for a large consortium of SPs. In the best case, that single route is advertised ubiquitously, from all accesses to a second, parallel Internet, dedicated to mobility. In that parallel Internet, you find BGP between SPs, and maybe between sites of a distributed Home, though the latter can also be IGP. Those routes are injected in the traditional sense, and refer to Homes, which are not mobile. But they do not leak to the main Internet. And you find NEMO for route projection. This happens between a proxy HA and either a HA or another proxy. In that model, the route is just sent to where is it is needed, over a tunnel that is built on demand. NEMO basic already uses route projection, and HAHA is widening the scope. Projected routes do not leak at all (a proxy does not redistribute a projected route it got from another proxy).=20 >> >> > >> ... >> > >> > Third, if the two ISPs are tightly linked together somehow (because >> > they advertise same /32) then it would be sufficient for >> them to run >> > route updates between them to point exactly where the /48 is, thus >> > avoiding core-BGP injections _and_ Mobile IP tunnelling. >> > > >Advertising BGP routes between providers certainly is an option, but the complexity can really grow >when there are more than two providers. My concern about that is that you are now seperating the >mobility signalling between the NEMO and BGP. It seems like it would be easier operationally if your >mobility signalling was in one protocol not two. > [Pascal] Agreed. It is difficult, almost impossible, to configure similar rules in 2 protocols, so we can not expect to see parallel sessions in BGP and NEMO. But maybe, if they happen between different set of peers having different set of policies, it could happen; and establishing BGP sessions on the fly between proxies might be a bit of overkill.=20 Anyway, global HAHA describes a model, it does not force you to use NEMO signaling, though it proposes one because sometimes it can be mush simpler and more efficient that way. >> >> good point, >> >> basically if the aircraft already knows what ISPs it will be >> visiting and it has some commercial arrangement with it, it >> is likely that the ISP can accept routes from the aircraft >> when it is visiting, so no mobile is required whatsoever >> >> I guess that the bottom line is that the trust/commercial >> model in the case of an aircraft visiting an ISP is different >> from the one assumed in mobile IP. I mean, i guess we can >> assume a much more higher level of trust between the aircraft >> and the visited ISP, so that route injection is an option (we >> can see that from boeing presentation, i guess) >> > >I can see there being a higher level of trust as you describe. Route injection could be an option. >I don't necessarily think that is the best option though. Route injection similar to what we are >doing today requires a high level of integration between ISPs. > > >Andrew > [Pascal] Note that route projection is a bit different.=20 Initially, proxy trusts HA (and other proxies), and MR's HA trusts MR. MR binds to proxy which binds to MR's HA. As part of that flow, MR's credentials and ownership on MNP can be verified by proxy as part of the binding to HA. Now proxy trusts MR for its ownership of MNP. If the same thing happens to that other MR and the proxies can find and trust each other, then proxies can project to each other the routes to the MNPs of their respective MRs. Those routes are never injected anywhere. This makes the trust model significantly simpler than if the route had to we injected and accepted all the way through between the proxies. Also route projection does not have the scalability issues that we see with route injection. Pascal From nemo-bounces@ietf.org Mon Apr 17 06:05:07 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVQbH-0006Zb-Er; Mon, 17 Apr 2006 06:04:55 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVQbG-0006ZW-BU for nemo@ietf.org; Mon, 17 Apr 2006 06:04:54 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVQbD-0004OU-W1 for nemo@ietf.org; Mon, 17 Apr 2006 06:04:54 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 17 Apr 2006 12:04:46 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3HA4iUE008547; Mon, 17 Apr 2006 12:04:44 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Mon, 17 Apr 2006 12:04:44 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Mon, 17 Apr 2006 12:04:28 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC020EA650@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZfYYYU1yCeLSrTR362hoH6m+iXSQCo/w7g From: "Pascal Thubert \(pthubert\)" To: "Masafumi Watari" X-OriginalArrivalTime: 17 Apr 2006 10:04:44.0057 (UTC) FILETIME=[59AE7890:01C66206] X-Spam-Score: 0.0 (/) X-Scan-Signature: 69a74e02bbee44ab4f8eafdbcedd94a1 Cc: marcelo bagnulo braun , nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Masafumi-san >>> What worries me is that i don't see any mobile ISP in the list claiming >>> that they need global HAHA to deploy their business, but there are end >>> users (air companies) that claim that want PI addressing for deploying >>> this. >>> >> If I understand well Tony's draft, NEMO is not the right forum for >> Provider Independent (PI) allocation policy. And that work is still >> ongoing, somewhere else... >> >> As for ISP side, I'm copying Masafumi-san who co authored the NEMO RO >> PS. > >We are now migrating our network based on MMD and NGN, and within the >next two years all of our mobile and fixed network would be based on IP. >Once that's done, I would imagine that something like global HAHA would >be needed. But if global HAHA is for NEMO and not MIP6, I don't see >a strong need at this point in time. > [Pascal] I guess it can work for both MIP6 and NEMO though the main benefits are for NEMO.=20 I've seen interest for NEMO in that space because of mobile Personal Area Networks (medical applications, this type of stuff) and in-Car Mobile Networks. In a same fashion, I've seen interest for IPv6 for Home Gateways because that brings a whole global network for a very long time instead of a temporary NAT address as we see today. So I'm a bit surprised by this comment... Pascal From nemo-bounces@ietf.org Tue Apr 18 06:07:12 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVn6y-0007mV-9i; Tue, 18 Apr 2006 06:07:08 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVn6x-0007mQ-PW for nemo@ietf.org; Tue, 18 Apr 2006 06:07:07 -0400 Received: from omgo.iij.ad.jp ([202.232.30.157]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVn6w-0003s0-9F for nemo@ietf.org; Tue, 18 Apr 2006 06:07:07 -0400 Received: OTM-MO id k3IA73WY008429; Tue, 18 Apr 2006 19:07:03 +0900 (JST) Received: OTM-MIX(otm-mix01) id k3IA72Q4095762; Tue, 18 Apr 2006 19:07:02 +0900 (JST) Received: from localhost (keiichi00.osaka.iij.ad.jp [192.168.64.45]) by jc-smtp.iij.ad.jp (JC-SMTP/jc-smtp) id k3IA722T009331; Tue, 18 Apr 2006 19:07:02 +0900 (JST) Date: Tue, 18 Apr 2006 19:09:29 +0900 (JST) Message-Id: <20060418.190929.109565634.keiichi@iijlab.net> To: pthubert@cisco.com Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA From: Keiichi SHIMA In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC020EA005@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC020EA005@xmb-ams-337.emea.cisco.com> X-Mailer: Mew version 4.1.50 on Emacs 21.3 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Spam-Score: 0.0 (/) X-Scan-Signature: 8abaac9e10c826e8252866cbe6766464 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hello Pascal, From: "Pascal Thubert \(pthubert\)" Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Wed, 12 Apr 2006 18:17:56 +0200 > Proxy HAs can be widely distributed. A MR binds to the closest proxy > (proxy binding) which succeeds if the proxy can bind to the MR HA > (primary binding). When two MRs need to establish a RO path, this is > actually performed by the proxies (secondary binding) between each > other, on behalf of the MRs. Proxies discover each other using a form of > NHRP, and establish a tunnel. At that point, since there is a trust > model within the infrastructure, a proxy HA can inject a route to the > MNP of its MR over the proxy to proxy tunnel, enabling RO. OK, I understand the above. And when we consider the MNN-CN case, will the traffic be like following? MNN-> MR -> proxy HA -> CN CN -> nearest HA -> parent HA -> proxy HA -> MR -> MNN Is my understanding correct? Regards, --- Keiichi SHIMA IIJ Research Laboratory WIDE Project From nemo-bounces@ietf.org Tue Apr 18 06:51:16 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVnng-0000CG-7U; Tue, 18 Apr 2006 06:51:16 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVnnf-0000CB-JC for nemo@ietf.org; Tue, 18 Apr 2006 06:51:15 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVnne-0006xn-3C for nemo@ietf.org; Tue, 18 Apr 2006 06:51:15 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id E1AC6212C59; Tue, 18 Apr 2006 13:51:12 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 7EEA2212C44; Tue, 18 Apr 2006 13:51:12 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 48A32BDC40; Tue, 18 Apr 2006 13:51:12 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 12575BDC38; Tue, 18 Apr 2006 13:51:12 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC020E9FF8@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC020E9FF8@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Date: Tue, 18 Apr 2006 13:51:11 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 4d87d2aa806f79fed918a62e834505ca Cc: ml-nemo WG , Ryuji Wakikawa Subject: [nemo] Re: draft-thubert-nemo-global-haha-01.txt X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Pascal, El 12/04/2006, a las 19:05, Pascal Thubert (pthubert) escribi=F3: > Hi again Marcelo > >> section 5.1.1 states >> >> In return, each site advertises the aggregation that encompasses > the >> Distributed Home Network aggregation back into the service = provider >> network, to the outside world (the internet). >> >> this seems to me to imply that the home network prefix is announced >> through BGP to the rest of the internet... i mean if you want to >> announce something to the rest of the internet it needs to be done=20 >> with >> BGP right? >> > [Pascal] I understand that this wording is disturbing you and why. I > agree I need to fix the text; I will try to reflect the mobile > aggregation we discussed based on our long yet positive thread :) > > Yes that would be good Bottom line is that as i see it, global HAHA enables two deployment=20 models: - one that is the one that you are suggesting that is the mobile ISP=20 idea, which seems to preserve the PA model of aggregation - the other one, in which each air company (and ship company) gets its=20= own PI prefix to use as for distributed Home network. Such model breaks=20= aggregation since they would be end sites injecting their own prefixes.=20= Enabling such possibility is a liability imho since it implies a cost=20 of the common resource. i would better with this approach if there was some wording about=20 discouraging the second approach regards, marcelo > Pascal > From nemo-bounces@ietf.org Tue Apr 18 07:07:02 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVo2s-0006EM-DV; Tue, 18 Apr 2006 07:06:58 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVo2r-0006EH-FC for nemo@ietf.org; Tue, 18 Apr 2006 07:06:57 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVo2q-0007eD-Uw for nemo@ietf.org; Tue, 18 Apr 2006 07:06:57 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id EB156212C59; Tue, 18 Apr 2006 14:06:55 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 83925212C44; Tue, 18 Apr 2006 14:06:55 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 177B8BDC40; Tue, 18 Apr 2006 14:06:55 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id D48F0BDC38; Tue, 18 Apr 2006 14:06:54 +0300 (EEST) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AADF@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AADF@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <8b90781eec34fae8388df3d1044bc374@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About route aggregation in global HAHA (was Re:draft-thubert-nemo-global-haha-01.txt (was: [nemo] Charter proposal) Date: Tue, 18 Apr 2006 14:06:54 +0300 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: a7d2e37451f7f22841e3b6f40c67db0f Cc: ml-nemo WG , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 14/04/2006, a las 20:49, Dul, Andrew L escribi=F3: >> >> well, we still need to inject one bgp route per distributed >> home network, right? > > I think I'm getting confused on the terminology. I tried going back=20= > to the nemo-terminology draft but that didn't help... :( > > We need to inject at least 1 BGP route. Theoretically all mobile=20 > nodes could be aggregated into a single route, however that isn't=20 > likely. The best case that I see happening is an aggregated prefix=20 > per "mobile-network operator". Where a "mobile network operator" is=20= > responsible for the provisioning an providing service to a large=20 > number of mobile platforms. > > Make sense? > I guess that the in order to evaluate a solution that consumes common=20 resources (like this one that burns global bgp table slots) we need to=20= evaluate: - number of resource consumed and - need for the particular solution So far we have discussed different deployment models that have impact=20 on the resources consumed. We have the mobile ISP model, where there seems to be a few global=20 consortiums that provide internet connectivity for aircrafts. In this=20 model, the number of routing slots burned seem to be quite limited the other option is that local ISPs host home agents for different=20 airlines, in which case each airline would have its own distributed=20 home network prefix and the number of routing table slots burned would=20= increase if internet access becomes a comodity (and more when they=20 start to use such apporach for other globally moving devices, like=20 ships ) So, i guess we really need to asses the cost that this apporach will=20 imply for the community in term of routing slots The second issue is why this solution and not other alternatives, that=20= may not have this cost (and that may have another cost for the=20 particular end site) The justification for this would be that this solution fits best the=20 requiremetns which unfortunatelly are not documented afaict, which=20 makes it very hard to figure out if another solution would solve this=20 problem.... regards, marcelo >> >>> >> ... >>> >>> Third, if the two ISPs are tightly linked together somehow (because >>> they advertise same /32) then it would be sufficient for >> them to run >>> route updates between them to point exactly where the /48 is, thus >>> avoiding core-BGP injections _and_ Mobile IP tunnelling. >>> > > Advertising BGP routes between providers certainly is an option, but=20= > the complexity can really grow when there are more than two providers.=20= > My concern about that is that you are now seperating the mobility=20 > signalling between the NEMO and BGP. It seems like it would be easier=20= > operationally if your mobility signalling was in one protocol not two. > >> >> good point, >> >> basically if the aircraft already knows what ISPs it will be >> visiting and it has some commercial arrangement with it, it >> is likely that the ISP can accept routes from the aircraft >> when it is visiting, so no mobile is required whatsoever >> >> I guess that the bottom line is that the trust/commercial >> model in the case of an aircraft visiting an ISP is different >> from the one assumed in mobile IP. I mean, i guess we can >> assume a much more higher level of trust between the aircraft >> and the visited ISP, so that route injection is an option (we >> can see that from boeing presentation, i guess) >> > > I can see there being a higher level of trust as you describe. Route=20= > injection could be an option. I don't necessarily think that is the=20= > best option though. Route injection similar to what we are doing=20 > today requires a high level of integration between ISPs. > > > Andrew > > > From nemo-bounces@ietf.org Tue Apr 18 07:13:25 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVo96-0007wu-6S; Tue, 18 Apr 2006 07:13:24 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVo95-0007wj-02 for nemo@ietf.org; Tue, 18 Apr 2006 07:13:23 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVo93-0007xl-Gm for nemo@ietf.org; Tue, 18 Apr 2006 07:13:22 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id AB605212C59; Tue, 18 Apr 2006 14:13:20 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id F00F7212C44; Tue, 18 Apr 2006 14:13:19 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id B0A65BDC40; Tue, 18 Apr 2006 14:13:19 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 7B040BDC38; Tue, 18 Apr 2006 14:13:19 +0300 (EEST) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AAE2@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAE2@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Date: Tue, 18 Apr 2006 14:13:18 +0300 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 4b800b1eab964a31702fa68f1ff0e955 Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 14/04/2006, a las 22:30, Dul, Andrew L escribi=F3: > >> -----Original Message----- >> From: Pascal Thubert (pthubert) [mailto:pthubert@cisco.com] > >> >>> >>>> it has been supported up to the plenary, >>> >>> what plenary? ietf plenary? since when IETf plenary supports >> solutions? >>> I guess this is the wg to decide, not the plenary... >> >> [Pascal] It was presented at the IETF plenary as something >> wanted. Was that IETF 62 or 63? >> >> > > I think people are referring to the presentation I made at IETF 62. A > copy of the presentation was unfortunately omitted from the=20 > proceedings. > I've posted a copy here for people who are interested. > http://www.quark.net/docs/Global_IP_Mobility_IETF62.pdf > but this is not about globalHAHA at all, right? I mean, i agree that providing support fro globally moving networks is=20= a problem that needs to be addressed, but what i am missing here is the=20= part when we actually define the problems, the requirements for the=20 solution and consider different alternatives and then pick up a=20 solution, which seems to me the way to solve problems... however here it seems that we have a problem (so far i agree) and then=20= just a single solution that can do the trick, without considering other=20= alternatives, and not even properly defining the requirements for the=20 solutions.... BTW, i think your paper and your presentation is a very good starting=20 point for describing the problems and also the requirements, just that=20= i would like to have this or other wg chartered to properly analyze the=20= problem and anlyze different approaches to the problem at hand... > The point of the presentation was to present the solution we had > deployed for ipv4 network mobility for our service and to spark > conversations about how network mobility might be done in the future. > exactly this is the discussion that i would like to have regards, marcelo > I've also recently written a whitepaper which describes our current > solution as well as some additional information. I've attached a copy > to this message. > > Andrew > > = From nemo-bounces@ietf.org Tue Apr 18 07:40:55 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVoZb-000617-FG; Tue, 18 Apr 2006 07:40:47 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVoZa-000612-8Z for nemo@ietf.org; Tue, 18 Apr 2006 07:40:46 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVoZY-0000sx-QP for nemo@ietf.org; Tue, 18 Apr 2006 07:40:46 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id D2EA2212C59; Tue, 18 Apr 2006 14:40:43 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 9B544212C44; Tue, 18 Apr 2006 14:40:43 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 62EBCBDC40; Tue, 18 Apr 2006 14:40:43 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 2D82CBDC38; Tue, 18 Apr 2006 14:40:43 +0300 (EEST) In-Reply-To: <443EF89D.2020508@kddilabs.jp> References: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> <443EF89D.2020508@kddilabs.jp> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: marcelo bagnulo braun Subject: RO approahces and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 18 Apr 2006 14:40:42 +0300 To: Masafumi Watari X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 0bc60ec82efc80c84b8d02f4b0e4de22 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > > The charter is still open for discussion :) The only reason for Global > HAHA being mentioned in the charter I believe is because it's been > presented several times in different WGs, though originally it wasn't > about RO but more on HA redundancy I believe(local HAHA?). > My question is then: what are the RO approaches that we want to include in the charter as work items? By this, i don't mean what solutions we want to include, but what problems that require RO are we going to work on... I can see the RO for globally moving networks is an important issue that seems to have a lot for interest, so i would suggest to include two items in the proposed charter: - Defining the requirements for a RO approach for globally moving networks - Propose a solution for globally moving networks But are there any other RO scenarios that people have a need for? Regards, marcelo From nemo-bounces@ietf.org Tue Apr 18 14:30:02 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVuxR-0001Rq-Bw; Tue, 18 Apr 2006 14:29:49 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVuxQ-0001Rl-Pl for nemo@ietf.org; Tue, 18 Apr 2006 14:29:48 -0400 Received: from stl-smtpout-01.boeing.com ([130.76.96.56]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVuxP-0007VW-Bx for nemo@ietf.org; Tue, 18 Apr 2006 14:29:48 -0400 Received: from stl-av-01.boeing.com ([192.76.190.6]) by stl-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id NAA09691; Tue, 18 Apr 2006 13:29:46 -0500 (CDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by stl-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k3IITjN01232; Tue, 18 Apr 2006 13:29:45 -0500 (CDT) Received: from XCH-NW-7V1.nw.nos.boeing.com ([130.247.54.34]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Tue, 18 Apr 2006 11:29:42 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Date: Tue, 18 Apr 2006 11:29:42 -0700 Message-ID: <108F5BE9DC629140A6E119C8BCDC28FC0109AAF6@XCH-NW-7V1.nw.nos.boeing.com> In-Reply-To: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Thread-Index: AcZi2Rtwl9dQxw8QS7akeYuvpaNRCQAOx0Tg From: "Dul, Andrew L" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 18 Apr 2006 18:29:42.0697 (UTC) FILETIME=[0F7DC990:01C66316] X-Spam-Score: 0.0 (/) X-Scan-Signature: e8a67952aa972b528dd04570d58ad8fe Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > -----Original Message----- > From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 > > > > I think people are referring to the presentation I made at=20 > IETF 62. A=20 > > copy of the presentation was unfortunately omitted from the=20 > > proceedings. > > I've posted a copy here for people who are interested. > > http://www.quark.net/docs/Global_IP_Mobility_IETF62.pdf > > >=20 > but this is not about globalHAHA at all, right? yes, it is about our IPv4 current method of BGP route-injection. =20 > I mean, i agree that providing support fro globally moving=20 > networks is a problem that needs to be addressed, but what i=20 > am missing here is the part when we actually define the=20 > problems, the requirements for the solution and consider=20 > different alternatives and then pick up a solution, which=20 > seems to me the way to solve problems... I guess there is a debate here about if we should be writing full requirements documents before we actually start discussing and designing. I know there are two groups of people in the IETF regarding this... Some believe that writing full requirements documents first is the best way, others think that they slow things down and you don't necessarily come out with better results. I think I am somewhere in the middle. =20 I know Terry was working on a basic set of requirements for aircraft at one point. I'll have to ask him how that is going. >=20 > however here it seems that we have a problem (so far i agree)=20 > and then just a single solution that can do the trick,=20 > without considering other alternatives, and not even properly=20 > defining the requirements for the solutions.... >=20 > BTW, i think your paper and your presentation is a very good=20 > starting point for describing the problems and also the=20 > requirements, just that i would like to have this or other wg=20 > chartered to properly analyze the problem and anlyze=20 > different approaches to the problem at hand... >=20 >=20 > > The point of the presentation was to present the solution we had=20 > > deployed for ipv4 network mobility for our service and to spark=20 > > conversations about how network mobility might be done in=20 > the future. > > >=20 > exactly this is the discussion that i would like to have >=20 I think we are already having this discussion. I think we have already talked about the limitations of route-injection as well as clarifying the operating models of global-HAHA. =20 Andrew From nemo-bounces@ietf.org Tue Apr 18 17:15:14 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVxXU-0000L8-2X; Tue, 18 Apr 2006 17:15:12 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FVxXT-0000L3-8y for nemo@ietf.org; Tue, 18 Apr 2006 17:15:11 -0400 Received: from smtp01.uc3m.es ([163.117.136.121]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FVxXQ-0007Ii-LK for nemo@ietf.org; Tue, 18 Apr 2006 17:15:11 -0400 Received: from smtp01.uc3m.es (localhost [127.0.0.1]) by localhost.uc3m.es (Postfix) with ESMTP id 658F4B1DE5; Tue, 18 Apr 2006 23:15:07 +0200 (CEST) Received: from [163.117.203.112] (unknown [163.117.203.112]) by smtp01.uc3m.es (Postfix) with ESMTP id 2ED81B1DEE; Tue, 18 Apr 2006 23:15:06 +0200 (CEST) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AAF6@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAF6@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <7707e369554752ab97a6b561636ed19b@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Date: Wed, 19 Apr 2006 00:14:45 +0300 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.623) X-Spam-Score: 0.0 (/) X-Scan-Signature: a8a20a483a84f747e56475e290ee868e Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 18/04/2006, a las 21:29, Dul, Andrew L escribi=F3: > >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>> >>> I think people are referring to the presentation I made at >> IETF 62. A >>> copy of the presentation was unfortunately omitted from the >>> proceedings. >>> I've posted a copy here for people who are interested. >>> http://www.quark.net/docs/Global_IP_Mobility_IETF62.pdf >>> >> >> but this is not about globalHAHA at all, right? > > yes, it is about our IPv4 current method of BGP route-injection. > >> I mean, i agree that providing support fro globally moving >> networks is a problem that needs to be addressed, but what i >> am missing here is the part when we actually define the >> problems, the requirements for the solution and consider >> different alternatives and then pick up a solution, which >> seems to me the way to solve problems... > > I guess there is a debate here about if we should be writing full > requirements documents before we actually start discussing and > designing. I know there are two groups of people in the IETF = regarding > this... Some believe that writing full requirements documents first = is > the best way, others think that they slow things down and you don't > necessarily come out with better results. > > I think I am somewhere in the middle. > i feel somehow the same way, but in this particular case, i think that=20= we are dealing with a delicate and critical issues i.e. the scaling of=20= the global routing table If we are going to design a solution that contribute with additional=20 routes in the dfz i would like to make sure that we are doing this=20 because there is no other (more scalable) way of doing this. The only way i can see to determine that this is the case is to figure=20= out the requirements, look for alternative solutions and come to the=20 conclusion that injection additional routes in the dfz is the way to go=20= to provide an acceptable solutions. I don't want to delay the process, and i think you have pretty much=20 clear ideas about the requirements. After all you have a solution=20 working right now and what the features and the problems are in this=20 solution i am willing to work on this if you need help > I know Terry was working on a basic set of requirements for aircraft = at > one point. I'll have to ask him how that is going. > that would be interesting indeed regards, marcelo >> >> however here it seems that we have a problem (so far i agree) >> and then just a single solution that can do the trick, >> without considering other alternatives, and not even properly >> defining the requirements for the solutions.... >> >> BTW, i think your paper and your presentation is a very good >> starting point for describing the problems and also the >> requirements, just that i would like to have this or other wg >> chartered to properly analyze the problem and anlyze >> different approaches to the problem at hand... >> >> >>> The point of the presentation was to present the solution we had >>> deployed for ipv4 network mobility for our service and to spark >>> conversations about how network mobility might be done in >> the future. >>> >> >> exactly this is the discussion that i would like to have >> > > I think we are already having this discussion. I think we have = already > talked about the limitations of route-injection as well as clarifying > the operating models of global-HAHA. > Andrew > From nemo-bounces@ietf.org Wed Apr 19 17:59:34 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWKhp-000322-T9; Wed, 19 Apr 2006 17:59:25 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWKhn-00031x-SF for nemo@ietf.org; Wed, 19 Apr 2006 17:59:23 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWKhn-00027D-Im for nemo@ietf.org; Wed, 19 Apr 2006 17:59:23 -0400 Received: from [10.1.201.19] ([10.1.201.19]) by bart.corp.azairenet.com with Microsoft SMTPSVC(6.0.3790.1830); Wed, 19 Apr 2006 14:59:22 -0700 Message-ID: <4446B2BA.5060102@azairenet.com> Date: Wed, 19 Apr 2006 14:59:22 -0700 From: Vijay Devarapalli User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> In-Reply-To: <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 19 Apr 2006 21:59:22.0084 (UTC) FILETIME=[83CDAA40:01C663FC] X-Spam-Score: 0.0 (/) X-Scan-Signature: f4c2cf0bccc868e4cc88dace71fb3f44 Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org hi Marcelo, first, I apologize for disappearing from the list for a while. marcelo bagnulo braun wrote: > However, the proposed charter only has Global HAHA as a RO solution for > nemo, so it seems to imply that this is the RO that we have selected. > I would propose to open the general discussion about which RO approaches > are we going to work on first rather than focussing in if we are going > to work on global haha or not... there is some misunderstanding. there was never a proposal to make Global HAHA as _the_ RO solution for NEMO. there was going to be a separate work item on a general RO solution for NEMO that does not involve multiple home agents. perhaps TJ or Thierry could comment on this. Vijay > > > Second, a more focussed issue with respect to routing system scalability > and business model > > you stated that: > >> Let me give you an example. Say that 3-4 ISPs from different continents >> form a mobile ISP consortium. >> >> That consortium obtains a short prefix from one of the NICs, call it the >> mobile prefix. Internally, the consortium shares that addressing space >> between members, leaving big chunks free for new comers. The rule is >> that only the big aggregation is redistributed over the Internet BGP, >> from most POP of members ISPs. But internally, the consortium of service >> providers runs a second BGP which deals with the subnetting of mobile >> prefix, over a mesh of tunnels and private links. > > but this seems to be complettelly different that what other people have > been claiming here on the list. > Past comments have strongly argued that global mobile network need PI > addressing and global HAHA seemed a mechanisms to provide access to this > PI prefix that the airline company gets. > > This is completelly different than the mobile ISP model that you are > suggesting. > > What worries me is that i don't see any mobile ISP in the list claiming > that they need global HAHA to deploy their business, but there are end > users (air companies) that claim that want PI addressing for deploying > this. > > Regards, marcelo > > From nemo-bounces@ietf.org Wed Apr 19 18:03:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWKmF-0004Ls-6Z; Wed, 19 Apr 2006 18:03:59 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWKmD-0004Ln-GD for nemo@ietf.org; Wed, 19 Apr 2006 18:03:57 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWKmC-0002JP-7U for nemo@ietf.org; Wed, 19 Apr 2006 18:03:57 -0400 Received: from [10.1.201.19] ([10.1.201.19]) by bart.corp.azairenet.com with Microsoft SMTPSVC(6.0.3790.1830); Wed, 19 Apr 2006 15:03:54 -0700 Message-ID: <4446B3CA.3070302@azairenet.com> Date: Wed, 19 Apr 2006 15:03:54 -0700 From: Vijay Devarapalli User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> In-Reply-To: <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 19 Apr 2006 22:03:54.0377 (UTC) FILETIME=[261A4790:01C663FD] X-Spam-Score: 0.0 (/) X-Scan-Signature: ffa9dfbbe7cc58b3fa6b8ae3e57b0aa3 Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: >> Let me give you an example. Say that 3-4 ISPs from different continents >> form a mobile ISP consortium. >> >> That consortium obtains a short prefix from one of the NICs, call it the >> mobile prefix. Internally, the consortium shares that addressing space >> between members, leaving big chunks free for new comers. The rule is >> that only the big aggregation is redistributed over the Internet BGP, >> from most POP of members ISPs. But internally, the consortium of service >> providers runs a second BGP which deals with the subnetting of mobile >> prefix, over a mesh of tunnels and private links. > > but this seems to be complettelly different that what other people have > been claiming here on the list. > Past comments have strongly argued that global mobile network need PI > addressing and global HAHA seemed a mechanisms to provide access to this > PI prefix that the airline company gets. well, your home prefix, along with the set of mobile network prefixes is a set of PI addresses. as long as the distributed set of home agents let the mobile network keep the same home address and mobile network prefix while providing mobility, you have a solution. Vijay From nemo-bounces@ietf.org Thu Apr 20 02:32:49 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWSiP-00080l-8k; Thu, 20 Apr 2006 02:32:33 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWSiO-00080g-6x for nemo@ietf.org; Thu, 20 Apr 2006 02:32:32 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWSiM-0002qL-Oh for nemo@ietf.org; Thu, 20 Apr 2006 02:32:32 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 7CBDB212C8D; Thu, 20 Apr 2006 09:32:21 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 15EF0212C61; Thu, 20 Apr 2006 09:32:21 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id D4585BDC40; Thu, 20 Apr 2006 09:32:20 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 9F675BDC38; Thu, 20 Apr 2006 09:32:20 +0300 (EEST) In-Reply-To: <4446B3CA.3070302@azairenet.com> References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <4446B3CA.3070302@azairenet.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Thu, 20 Apr 2006 09:32:19 +0300 To: Vijay Devarapalli X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 769a46790fb42fbb0b0cc700c82f7081 Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 20/04/2006, a las 1:03, Vijay Devarapalli escribi=F3: > marcelo bagnulo braun wrote: > >>> Let me give you an example. Say that 3-4 ISPs from different=20 >>> continents >>> form a mobile ISP consortium. >>> >>> That consortium obtains a short prefix from one of the NICs, call it=20= >>> the >>> mobile prefix. Internally, the consortium shares that addressing=20 >>> space >>> between members, leaving big chunks free for new comers. The rule is >>> that only the big aggregation is redistributed over the Internet = BGP, >>> from most POP of members ISPs. But internally, the consortium of=20 >>> service >>> providers runs a second BGP which deals with the subnetting of = mobile >>> prefix, over a mesh of tunnels and private links. >> but this seems to be complettelly different that what other people=20 >> have been claiming here on the list. >> Past comments have strongly argued that global mobile network need PI=20= >> addressing and global HAHA seemed a mechanisms to provide access to=20= >> this PI prefix that the airline company gets. > > well, your home prefix, along with the set of mobile > network prefixes is a set of PI addresses. as long > as the distributed set of home agents let the mobile > network keep the same home address and mobile network > prefix while providing mobility, you have a solution. > yes, so the contribution of this solution to the global routing table=20 would be one entry per home prefix Regards, marcelo > Vijay > From nemo-bounces@ietf.org Thu Apr 20 08:40:23 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWYSK-0006Rc-SS; Thu, 20 Apr 2006 08:40:20 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWYSJ-0006RX-Hf for nemo@ietf.org; Thu, 20 Apr 2006 08:40:19 -0400 Received: from mandala.kddilabs.jp ([192.26.91.6]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWYSJ-00055i-62 for nemo@ietf.org; Thu, 20 Apr 2006 08:40:19 -0400 Received: from localhost (localhost [127.0.0.1]) by mandala.kddilabs.jp (Postfix) with ESMTP id 04169EC86F; Thu, 20 Apr 2006 21:40:10 +0900 (JST) Received: from neutrino.hsc.kddilabs.jp (neutrino.hsc.kddilabs.jp [2001:200:601:200::60]) by mandala.kddilabs.jp (Postfix) with ESMTP id B238BEC860; Thu, 20 Apr 2006 21:40:09 +0900 (JST) Received: from [IPv6:2001:200:601:200:f469:a049:88ab:e504] (unknown [IPv6:2001:200:601:200:f469:a049:88ab:e504]) by neutrino.hsc.kddilabs.jp (Postfix) with ESMTP id 5DCC02E02E; Thu, 20 Apr 2006 21:34:53 +0900 (JST) Message-ID: <4447810F.7000003@kddilabs.jp> Date: Thu, 20 Apr 2006 21:39:43 +0900 From: Masafumi Watari Organization: KDDI R&D Laboratories Inc. User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: RO approahces and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> <443EF89D.2020508@kddilabs.jp> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new X-Spam-Score: 0.0 (/) X-Scan-Signature: cab78e1e39c4b328567edb48482b6a69 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Marcelo, On 2006/04/18 20:40, marcelo bagnulo braun wrote: >> >> The charter is still open for discussion :) The only reason for Global >> HAHA being mentioned in the charter I believe is because it's been >> presented several times in different WGs, though originally it wasn't >> about RO but more on HA redundancy I believe(local HAHA?). >> > > My question is then: > > what are the RO approaches that we want to include in the charter as > work items? > > By this, i don't mean what solutions we want to include, but what > problems that require RO are we going to work on... > I can see the RO for globally moving networks is an important issue that > seems to have a lot for interest, so i would suggest to include two > items in the proposed charter: > > - Defining the requirements for a RO approach for globally moving networks When you say "Globally moving networks", it would potentially include any NEMO, so I suggest we rephrase this, but I agree. > - Propose a solution for globally moving networks Just curious. would this be HAHA based approach or something different? BTW, I personally support the HAHA based approach, but I also agree that draft-thubert has a BGP/routing scalability issue that it needs to solve. > But are there any other RO scenarios that people have a need for? Well, there hasn't been much discussion on this and I don't know if people have interest in it either. I've seen this question a couple of times, but nobody has spoken up. One of the problem is that if you take a look at the analysis draft, the problem of NEMO RO is that it is quite complex. There were comments that NEMO WG isn't the right place to handle some solution, like MANEMO, while others think that we are too early in stage to work on other scenarios. watari From nemo-bounces@ietf.org Thu Apr 20 08:58:56 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWYkJ-0006yX-6R; Thu, 20 Apr 2006 08:58:55 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWYkI-0006yS-1c for nemo@ietf.org; Thu, 20 Apr 2006 08:58:54 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWYkH-0005pf-Iu for nemo@ietf.org; Thu, 20 Apr 2006 08:58:54 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 302F1212C59; Thu, 20 Apr 2006 15:58:52 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id F2D10212C44; Thu, 20 Apr 2006 15:58:51 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id BDD9EBDC40; Thu, 20 Apr 2006 15:58:51 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 8AC14BDC38; Thu, 20 Apr 2006 15:58:51 +0300 (EEST) In-Reply-To: <4447810F.7000003@kddilabs.jp> References: <7892795E1A87F04CADFCCF41FADD00FC020E9F22@xmb-ams-337.emea.cisco.com> <443EF89D.2020508@kddilabs.jp> <4447810F.7000003@kddilabs.jp> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: RO approahces and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Thu, 20 Apr 2006 15:58:51 +0300 To: Masafumi Watari X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: d8ae4fd88fcaf47c1a71c804d04f413d Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 20/04/2006, a las 15:39, Masafumi Watari escribi=F3: > Hi Marcelo, > > On 2006/04/18 20:40, marcelo bagnulo braun wrote: >>> >>> The charter is still open for discussion :) The only reason for=20 >>> Global >>> HAHA being mentioned in the charter I believe is because it's been >>> presented several times in different WGs, though originally it = wasn't >>> about RO but more on HA redundancy I believe(local HAHA?). >>> >> My question is then: >> what are the RO approaches that we want to include in the charter as=20= >> work items? > > >> By this, i don't mean what solutions we want to include, but what=20 >> problems that require RO are we going to work on... >> I can see the RO for globally moving networks is an important issue=20= >> that seems to have a lot for interest, so i would suggest to include=20= >> two items in the proposed charter: >> - Defining the requirements for a RO approach for globally moving=20 >> networks > > When you say "Globally moving networks", it would potentially include > any NEMO, so I suggest we rephrase this, but I agree. > > > - Propose a solution for globally moving networks > > Just curious. would this be HAHA based approach or something = different? > depends on the requirements we agree on i guess and whether alternative=20= solutions can fulfill the requirements without burden on the global bgp=20= table, i guess > BTW, I personally support the HAHA based approach, but I also agree=20 > that > draft-thubert has a BGP/routing scalability issue that it needs to > solve. > >> But are there any other RO scenarios that people have a need for? > > Well, there hasn't been much discussion on this and I don't know if > people have interest in it either. I've seen this question a couple > of times, but nobody has spoken up. agree, nobody seems to care for the other cases so far regards, marcelo > One of the problem is that if you > take a look at the analysis draft, the problem of NEMO RO is that it = is > quite complex. There were comments that NEMO WG isn't the right place > to handle some solution, like MANEMO, while others think that we are=20= > too > early in stage to work on other scenarios. > > watari > From nemo-bounces@ietf.org Thu Apr 20 10:17:42 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWZyW-0007WW-AW; Thu, 20 Apr 2006 10:17:40 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWZyU-0007WQ-9U for nemo@ietf.org; Thu, 20 Apr 2006 10:17:38 -0400 Received: from pigwidgeon.lancs.ac.uk ([148.88.0.67]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWZyT-0001qr-Rr for nemo@ietf.org; Thu, 20 Apr 2006 10:17:38 -0400 Received: from marl.lancs.ac.uk ([148.88.1.10]) by pigwidgeon.lancs.ac.uk with esmtp (Exim 4.60) (envelope-from ) id 1FWZyS-0006iG-RD for nemo@ietf.org; Thu, 20 Apr 2006 15:17:36 +0100 Received: from ind031000002.lancs.ac.uk ([148.88.224.214] helo=IND031000002) by marl.lancs.ac.uk with esmtp (Exim 4.60) (envelope-from ) id 1FWZyS-0000Ko-SH; Thu, 20 Apr 2006 15:17:36 +0100 From: "Ben McCarthy" To: "'marcelo bagnulo braun'" , "'Masafumi Watari'" Subject: RE: RO approahces and the proposed charter (was Re: About theproposed charter and global HAHA (was Re: [nemo] RE: Aboutroute aggregation in global HAHA Date: Thu, 20 Apr 2006 15:17:36 +0100 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable X-Mailer: Microsoft Office Outlook, Build 11.0.6353 Thread-Index: AcZkejLvQbXfRmbtT8GonOMqFlNqawACERrA X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 In-Reply-To: X-Spam-Score: 0.0 (/) X-Scan-Signature: a7d2e37451f7f22841e3b6f40c67db0f Cc: 'ml-nemo WG' X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Message-Id: Hi all, In relation to the topic of re-chartering, has the WG considered the possibility of splitting the charter in two? With one charter = concentrating on researching the single roaming NEMO scenario and the other = concentrating on the Nested NEMO scenario? It seems to me from the work I've done that = the scenarios which require a single NEMO model (such as ITS) and those that = may require a Nested NEMO model are often (but admittedly not always) quite distinct.=20 With two separate charters the WG could focus on topics such as RO and multihoming that are specifically suited to the two different scenarios, rather than attempt to produce a one-fits-all solution. The later = charter could also conceptually house the MANEMO work as well possibly? I am not particularly pushing my own agenda here, I just had this idea = and thought I'd push it out on the ML to see what other people think about = it.=20 Regards, Ben -----Original Message----- From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es]=20 Sent: 20 April 2006 13:59 To: Masafumi Watari Cc: ml-nemo WG Subject: Re: RO approahces and the proposed charter (was Re: About theproposed charter and global HAHA (was Re: [nemo] RE: Aboutroute aggregation in global HAHA El 20/04/2006, a las 15:39, Masafumi Watari escribi=F3: > Hi Marcelo, > > On 2006/04/18 20:40, marcelo bagnulo braun wrote: >>> >>> The charter is still open for discussion :) The only reason for=20 >>> Global >>> HAHA being mentioned in the charter I believe is because it's been >>> presented several times in different WGs, though originally it = wasn't >>> about RO but more on HA redundancy I believe(local HAHA?). >>> >> My question is then: >> what are the RO approaches that we want to include in the charter as=20 >> work items? > > >> By this, i don't mean what solutions we want to include, but what=20 >> problems that require RO are we going to work on... >> I can see the RO for globally moving networks is an important issue=20 >> that seems to have a lot for interest, so i would suggest to include=20 >> two items in the proposed charter: >> - Defining the requirements for a RO approach for globally moving=20 >> networks > > When you say "Globally moving networks", it would potentially include > any NEMO, so I suggest we rephrase this, but I agree. > > > - Propose a solution for globally moving networks > > Just curious. would this be HAHA based approach or something = different? > depends on the requirements we agree on i guess and whether alternative=20 solutions can fulfill the requirements without burden on the global bgp=20 table, i guess > BTW, I personally support the HAHA based approach, but I also agree=20 > that > draft-thubert has a BGP/routing scalability issue that it needs to > solve. > >> But are there any other RO scenarios that people have a need for? > > Well, there hasn't been much discussion on this and I don't know if > people have interest in it either. I've seen this question a couple > of times, but nobody has spoken up. agree, nobody seems to care for the other cases so far regards, marcelo > One of the problem is that if you > take a look at the analysis draft, the problem of NEMO RO is that it = is > quite complex. There were comments that NEMO WG isn't the right place > to handle some solution, like MANEMO, while others think that we are=20 > too > early in stage to work on other scenarios. > > watari > From nemo-bounces@ietf.org Thu Apr 20 13:26:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWcvB-0006Ll-LW; Thu, 20 Apr 2006 13:26:25 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWcv9-0006Ld-NH for nemo@ietf.org; Thu, 20 Apr 2006 13:26:23 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWcv8-0003cC-Dy for nemo@ietf.org; Thu, 20 Apr 2006 13:26:23 -0400 Received: from [10.1.201.19] ([10.1.201.19]) by bart.corp.azairenet.com with Microsoft SMTPSVC(6.0.3790.1830); Thu, 20 Apr 2006 10:26:16 -0700 Message-ID: <4447C434.6050703@azairenet.com> Date: Thu, 20 Apr 2006 10:26:12 -0700 From: Vijay Devarapalli User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <4446B3CA.3070302@azairenet.com> <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> In-Reply-To: <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-OriginalArrivalTime: 20 Apr 2006 17:26:16.0502 (UTC) FILETIME=[87A60560:01C6649F] X-Spam-Score: 0.0 (/) X-Scan-Signature: 8b431ad66d60be2d47c7bfeb879db82c Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: > > El 20/04/2006, a las 1:03, Vijay Devarapalli escribió: > >> marcelo bagnulo braun wrote: >> >>>> Let me give you an example. Say that 3-4 ISPs from different continents >>>> form a mobile ISP consortium. >>>> >>>> That consortium obtains a short prefix from one of the NICs, call it >>>> the >>>> mobile prefix. Internally, the consortium shares that addressing space >>>> between members, leaving big chunks free for new comers. The rule is >>>> that only the big aggregation is redistributed over the Internet BGP, >>>> from most POP of members ISPs. But internally, the consortium of >>>> service >>>> providers runs a second BGP which deals with the subnetting of mobile >>>> prefix, over a mesh of tunnels and private links. >>> but this seems to be complettelly different that what other people >>> have been claiming here on the list. >>> Past comments have strongly argued that global mobile network need PI >>> addressing and global HAHA seemed a mechanisms to provide access to >>> this PI prefix that the airline company gets. >> >> well, your home prefix, along with the set of mobile >> network prefixes is a set of PI addresses. as long >> as the distributed set of home agents let the mobile >> network keep the same home address and mobile network >> prefix while providing mobility, you have a solution. >> > > yes, so the contribution of this solution to the global routing table > would be one entry per home prefix since global HAHA is not a generic route optimization solution, and not many deployments can actually have distributed home agents, it shouldn't be a concern for now. Vijay > > Regards, marcelo > > > > >> Vijay >> > From nemo-bounces@ietf.org Thu Apr 20 15:38:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWezP-0001Uu-JH; Thu, 20 Apr 2006 15:38:55 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWezO-0001Up-Ux for nemo@ietf.org; Thu, 20 Apr 2006 15:38:54 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWezO-0001uT-TY for nemo@ietf.org; Thu, 20 Apr 2006 15:38:54 -0400 Received: from numenor.qualcomm.com ([129.46.51.58]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FWezO-0005Gx-5D for nemo@ietf.org; Thu, 20 Apr 2006 15:38:54 -0400 Received: from sabrina.qualcomm.com (sabrina.qualcomm.com [129.46.61.150]) by numenor.qualcomm.com (8.13.6/8.12.5/1.0) with ESMTP id k3KJcoKB025566 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=FAIL); Thu, 20 Apr 2006 12:38:51 -0700 Received: from NAEXBR02.na.qualcomm.com (naexbr02.qualcomm.com [10.46.92.109]) by sabrina.qualcomm.com (8.13.6/8.12.5/1.0) with ESMTP id k3KJcn65007098; Thu, 20 Apr 2006 12:38:50 -0700 (PDT) Received: from NAEX06.na.qualcomm.com ([129.46.135.160]) by NAEXBR02.na.qualcomm.com with Microsoft SMTPSVC(6.0.3790.1830); Thu, 20 Apr 2006 12:38:49 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5.7226.0 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Date: Thu, 20 Apr 2006 12:38:47 -0700 Message-ID: <1487A357FD2ED544B8AD29E528FF9DF00233CEB8@NAEX06.na.qualcomm.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Thread-Index: AcZkoEssXVgMGEQ4SyuQvuwbigGZOQAEXWfA From: "Soliman, Hesham" To: "Vijay Devarapalli" , "marcelo bagnulo braun" X-OriginalArrivalTime: 20 Apr 2006 19:38:49.0339 (UTC) FILETIME=[0BE8A4B0:01C664B2] X-Spam-Score: -2.6 (--) X-Scan-Signature: bb8f917bb6b8da28fc948aeffb74aa17 Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > > yes, so the contribution of this solution to the global=20 > routing table=20 > > would be one entry per home prefix >=20 > since global HAHA is not a generic route optimization > solution, and not many deployments can actually have > distributed home agents, it shouldn't be a concern for > now. =3D> I must say that this is not a good assumption to go by.=20 We've seen many examples of protocols and features that are used in unintended ways with unintended consequences. I don't think we can keep making these assumptions given that we have no=20 control over how deployments will evolve.=20 Hesham >=20 > Vijay >=20 > >=20 > > Regards, marcelo > >=20 > >=20 > >=20 > >=20 > >> Vijay > >> > >=20 >=20 >=20 >=20 From nemo-bounces@ietf.org Thu Apr 20 21:46:37 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWkjA-0008LW-Gk; Thu, 20 Apr 2006 21:46:32 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWkj6-0008Ks-SX for nemo@ietf.org; Thu, 20 Apr 2006 21:46:28 -0400 Received: from f.csce.kyushu-u.ac.jp ([133.5.33.1]) by ietf-mx.ietf.org with smtp (Exim 4.43) id 1FWkj1-0006Iq-0u for nemo@ietf.org; Thu, 20 Apr 2006 21:46:28 -0400 Received: (qmail 14203 invoked from network); 21 Apr 2006 01:46:13 -0000 Received: from unknown (HELO ?172.17.3.211?) (192.168.77.254) by 192.168.77.101 with SMTP; 21 Apr 2006 01:46:13 -0000 Date: Fri, 21 Apr 2006 10:46:13 +0900 From: Teruaki Kitasuka To: manet@ietf.org, mip4@ietf.org, mip6@ietf.org, mipshop@ietf.org, monami6@ietf.org, nemo@ietf.org Message-Id: <20060421104311.4A52.KITASUKA@f.csce.kyushu-u.ac.jp> MIME-Version: 1.0 Content-Type: text/plain; charset="ISO-2022-JP" Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver. 2.24.02 [ja] X-Spam-Score: 0.0 (/) X-Scan-Signature: 202a3ece0492a8c7e7c8672d5214398f Cc: Subject: [nemo] Extended Deadline - CfP - 3rd Int. Conf. on Mobile Computing and Ubiquitous Networking (ICMU2006) at London, UK X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org (Apologies if you receive multiple copies of this message) !!!! Extended Deadline to May 9, ICMU2006 !!!! ======================================================================== C A L L F O R P A P E R S ======================================================================== The Third International Conference on Mobile Computing and Ubiquitous Networking (ICMU2006) October 11-13, 2006 BCS London Office, London, UK http://www.icmu.org/icmu2006/ Sponsored by: IPSJ SIG-MBL (Information Processing Society of Japan, Special Interest Group of Mobile Computing and Ubiquitous Networking) Technically co-sponsored by: BCS (British Computer Society), IEE (Institution of Electrical Engineers), and IEEE ITS Society, Technical Committee on Mobile Communications & Applications Supported by: ICF (International Communications Foundation) and SCAT (Support Center for Advanced Telecommunications Technology Research, Foundation) SCOPE: Mobile computing aims at ubiquitous user access to computer application. To make it possible, networking technology which enables Internet access anytime and everywhere, i.e. ubiquitous networking is necessary. It is obvious that with the accelerating trends towards mobile networks, ad-hoc networks, and wireless networks, mobile computing and ubiquitous networking will play an important role in future network. This conference is aimed at providing researchers and practitioners in this hot research area a forum for discussion and collaboration. Authors are invited to submit papers addressing, but not limited to, the following topics: - Network management for ubiquitous networking - Data management for mobile computing - Mobile device management - Performance evaluation for mobile computing and ubiquitous networking systems - Broadcast communications - Network architectures, protocols, or service models for ubiquitous networking - Security in mobile computing and ubiquitous networking - Mobile OS - Wireless and mobile communications - 4G wireless communications - Mobile Internetworking - Ad-hoc networks - Network mobility - Mesh networks - Sensor networks - Location-based services SUBMISSION INSTRUCTIONS: Papers are solicited as full papers of no more than 8 pages, each of which will be subject to a full review process. Submission should already follow the author guidelines as specified in the web site below. An electronic, PDF-based submission of papers is mandatory. Please check the web site of the conference http://www.icmu.org/icmu2006/ for further submission instructions. After the conference, a selected number of papers will be published as special issues in IPSJ journal and IPSJ Digital Courier. Please check the web site of IPSJ http://www.ipsj.or.jp/08editt/dc/index.html for the details of IPSJ Digital Coulier. IMPORTANT DATES: ########################################################## # Deadline for submissions : May 9, 2006 # # Notification of acceptance: July 8, 2006 # # Camera ready due: July 31, 2006 # ########################################################## ORGANIZING COMMITTEE: General Chair: Chai-Keong Toh (Queen Mary University of London, UK) Technical Program Committee Co-Chairs: Susumu Ishihara (Shizuoka University, Japan) Eiji Kamioka (National Institute of Informatics, Japan) Publication Co-Chairs: Gen Kitagata (Tohoku University, Japan) Yang Yang (University College London, UK) Publicity Co-Chairs: Teruaki Kitasuka (Kyushu University, Japan) Kun Yang (University of Essex, UK) Local Arrangement Chair: David Martland (Kingston University London, UK) Hiroshi Mineno (Shizuoka University, Japan) Treasurers: David Martland (Kingston University London, UK) Tomohiko Yagyu (NEC, Japan) Registration Co-Chairs: Jindong Hou (France Telecom, UK) Homare Murakami (NICT, Japan) ICMU STEERING COMMITTEE: Tadanori Mizuno (Shizuoka University, Japan) Osamu Takahashi (Future University-Hakodate, Japan) Takashi Watanabe (Shizuoka University, Japan) Miki Yamamoto (Kansai University, Japan) TECHNICAL PROGRAM COMMITTEE: Mohammed Atiquzzaman (University of Oklahoma, USA) Xiuzhen Cheng (George Washington University, USA) We Duke Cho (Ajou University, Korea) John Gardiner (Bradford University, UK) Teruo Higashino (Osaka University, Japan) Choong Seon Hong (Kyung Hee University, Korea) Russell Hsing (Telcordia Technologies, USA) Chung Ming Huang (National Cheng Kung University, Taiwan) Joseph Hui (Arizona State University, USA) Masugi Inoue (NICT, Japan) Masahiro Ishiyama (Toshiba Corporation, Japan) Holger Karl (University of Paderborn, Germany) Hughes Kester (Qinetiq Ltd, UK) Dongkyun Kim (Kyungpook National University, Korea) Peter Langendoerfer (IHP Microelectronics, Germany) Madjid Merabti (Liverpool John Moores University, UK) Isabelle Moreau (Mitsubishi Electric ITE, France) Hiroyuki Morikawa (University of Tokyo, Japan) Yasuto Nakanishi (Keio University, Japan) Ken Ohta (NTT DoCoMo, Japan) Ryouji Ono (Mitsubishi Electric, Japan) Mohamed Ould-Khaoua (University of Glasgow, UK) Antonio Pescape (University of Napoli, Italy) Petar Popovski (Aalborg University, Denmark) Jeremy Randles (British Telecom, UK) Ichiro Satoh (National Institute of Informatics, Japan) Winston Seah (Institute for Infocomm Research, Singapore) Hiroshi Shigeno (Keio University, Japan) Biplab Sikdar (Rensselaer Polytechnic Institute, USA) David Simplot-Ryl (University of Lille 1, France) Tatsuya Suda (University of California Irvine, USA) Shinta Sugimoto (Nippon Ericsson, Japan) Keisuke Suwa (Musashi Institute of Technology, Japan) Kevin W. Tsai (University of California, Irvine, USA) Xin Wang (Fudan University, China) Neil Williams (ERA Technology Ltd, UK) Nik Van den Wingaert (University of Antwerp, Belgium) Hirozumi Yamaguchi (Osaka University, Japan) Kun Yang (Essex University, UK) Yang Yang (University College London, UK) Masashi Yano (Hitachi Ltd, Japan) Hidetoshi Yokota (KDDI R&D Laboratories, Japan) Ekio Yoneki (Cambridge University, UK) Qing-An Zeng (University of Cincinnati, USA) ---- $BKL?\2l(B $B51L@(B TEL/FAX: 092-583-7625($B8&5f<<(B), FAX2: 092-583-1338($B>pJs;vL3<<(B) $B6e=#Bg3X(B $BBg3X1!(B $B%7%9%F%`>pJs2J3X8&5f1!(B $B>pJs9)3XItLg(B($BC^;gCO6h(B) $BJ!ED!&Cf@>8&5f<<(B $B=u References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <4446B3CA.3070302@azairenet.com> <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> <4447C434.6050703@azairenet.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <39532c9d98fff9951fab56f313de4fb5@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Fri, 21 Apr 2006 09:59:22 +0300 To: Vijay Devarapalli X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: b19722fc8d3865b147c75ae2495625f2 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 20/04/2006, a las 20:26, Vijay Devarapalli escribi=F3: > since global HAHA is not a generic route optimization > solution, and not many deployments can actually have > distributed home agents, it shouldn't be a concern for > now. > but if inflight internet access becomes a comodity (and i think it may)=20= then all airlines and ships may want to have access to this technology,=20= i am afraid, so i am not ready to dismiss the effect of this in the=20 global routing table so easily... I am more convinced by the argument that there are some deployment=20 models that achieve route aggregation (Pascal's mobile ISP argument)=20 than by the expected reduced number of users, but still, i am far from=20= easy with this... regards, marcelo > Vijay > >> Regards, marcelo >>> Vijay >>> > From nemo-bounces@ietf.org Fri Apr 21 03:19:31 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWpvN-0005fJ-EJ; Fri, 21 Apr 2006 03:19:29 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWpvL-0005fE-U6 for nemo@ietf.org; Fri, 21 Apr 2006 03:19:27 -0400 Received: from mail.sfc.wide.ad.jp ([203.178.142.146]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWpvK-0005oz-I1 for nemo@ietf.org; Fri, 21 Apr 2006 03:19:27 -0400 Received: from dhcp-rocq-97.inria.fr (guest-rocq-135223.inria.fr [128.93.135.223]) by mail.sfc.wide.ad.jp (Postfix) with ESMTP id 156F94D867 for ; Fri, 21 Apr 2006 16:19:15 +0900 (JST) Date: Fri, 21 Apr 2006 09:19:58 +0200 From: Thierry Ernst To: ml-nemo Message-Id: <20060421091958.1613ead0.ernst@sfc.wide.ad.jp> Organization: Keio University X-Mailer: Sylpheed version 0.9.12 (GTK+ 1.2.10; powerpc-apple-darwin7.8.0) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Spam-Score: 0.0 (/) X-Scan-Signature: 7baded97d9887f7a0c7e8a33c2e3ea1b Subject: [nemo] CFP - WNEPT - Waterloo, August 2006. X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Dea all, As chair of the NEMO working group, I've refrained myself to post CFP (due to past critics, etc), but other people do it, and it is acceptable I think when the CFP does include network mobility. So, here is one that people on this list will have interest in: Workshop on Networking in Public Transport (WNEPT, http://www.wnept.org) to be held in Waterloo, Ontario, Canada, on August 10, 2006. (in conjunction with the Third International Conference on Quality of Service in Heterogeneous Wired/Wireless Networks [QShine] 2006, http://www.qshine.org) The workshop will provide a highly selective venue to present work related to communication networks on public transport, ranging from theoretical research to practical systems deployment. The full CFP is available on the web site http://www.wnept.org Important Dates Full Papers due: May 15, 2006 Notification of Acceptance: June 16, 2006 Camera-ready Manuscripts due: June 30, 2006 Workshop Date: August 10, 2006 Regards, Thierry. -- Thierry ERNST, PhD INRIA Rocquencourt Projet IMARA +33 1 39 63 59 30 From nemo-bounces@ietf.org Fri Apr 21 03:26:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWq2d-00009I-0w; Fri, 21 Apr 2006 03:26:59 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWq2b-00009D-EZ for nemo@ietf.org; Fri, 21 Apr 2006 03:26:57 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWq2Y-00062K-2h for nemo@ietf.org; Fri, 21 Apr 2006 03:26:57 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 21 Apr 2006 09:26:55 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3L7QoUG011087; Fri, 21 Apr 2006 09:26:52 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Fri, 21 Apr 2006 09:26:50 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: RO approaches and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Fri, 21 Apr 2006 09:26:32 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021512A9@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: RO approaches and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZkd8QlHQ3dzCJDQTyX03wuy9BC9QAmMNrA From: "Pascal Thubert \(pthubert\)" To: "Masafumi Watari" , "marcelo bagnulo braun" , "Hesham Soliman" , X-OriginalArrivalTime: 21 Apr 2006 07:26:50.0943 (UTC) FILETIME=[F4EAFCF0:01C66514] X-Spam-Score: 0.0 (/) X-Scan-Signature: cab78e1e39c4b328567edb48482b6a69 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >Just curious. would this be HAHA based approach or something different? > >BTW, I personally support the HAHA based approach, but I also agree that >draft-thubert has a BGP/routing scalability issue that it needs to >solve. > [Pascal]=20 This discussion has started with the assumption that global HAHA would be deployed against all common sense. Something similar could have been=20 said in the early IPv6 days. Such a vast address space, BGP core will=20 collapse! Global HAHA does not mean that every single distributed Home will end up in the DFZ. If the draft gives that impression, I'll fix the draft.=20 In fact, there could be a very limited set of mobile ISP consortiums,=20 and a single prefix in the DFZ for each of those. GTP has proven that=20 this can happen in the real world, hasn't it? Global HAHA introduces a mobile Plane that is accessible from many POPs=20 around the Internet. It's like a subway where the stations are proxy HAs.=20 Only the aggregation of that mobile Plane (one prefix per consortium)=20 has to be injected in the DFZ, like a metro sign indicating the nearest=20 station.=20 Fixed routes in the mobile Plane are handled using IGP and EGP as usual. That would be the routes to the Homes. What's missing in the draft is that Description with ISPs interacting at BGP level. Inside the mobile Plane, Proxies also could inject mobile prefixes if that was a wanted feature. But global HAHA also introduces route projection; Proxies at the end of a conversation discover each other (NHRP) and=20 then exchange routes over a tunnel they set up dynamically.=20 So yes, like anything else, global HAHA could be deployed stupidly,=20 but guess what, people will know better. Pascal From nemo-bounces@ietf.org Fri Apr 21 04:32:52 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWr4L-0000h0-TS; Fri, 21 Apr 2006 04:32:49 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWr4J-0000gu-EA for nemo@ietf.org; Fri, 21 Apr 2006 04:32:47 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWr4H-0000Ui-Un for nemo@ietf.org; Fri, 21 Apr 2006 04:32:47 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 1A749212C59; Fri, 21 Apr 2006 11:32:45 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id D6662212C44; Fri, 21 Apr 2006 11:32:44 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 874F7BDC40; Fri, 21 Apr 2006 11:32:44 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 4FF7FBDC38; Fri, 21 Apr 2006 11:32:44 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021512A9@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021512A9@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <0da912b0ea840b7bf37244ac1d384d6b@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: RO approaches and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Fri, 21 Apr 2006 11:32:43 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: d0bdc596f8dd1c226c458f0b4df27a88 Cc: Hesham Soliman , Vijay.Devarapalli@AzaireNet.com, ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 21/04/2006, a las 10:26, Pascal Thubert (pthubert) escribi=F3: .... > In fact, there could be a very limited set of mobile ISP consortiums, > and a single prefix in the DFZ for each of those. Interesting.... However, this is NOT presented anywhere in the draft actually what it is actually in the draft is 5.1 Initial routing 5.1.1 External routing Sites are expected to be connected locally to the internet, via the network of one or more service provider. Each site has a default route to the internet via that connection. . .. / --------- ........ ..;. --------- | | .. / ..... | | | ::/0 -> .... ; ... <- ::/0 | | =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3DHAHA=3DTUNNEL=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D | | | .... ; .... | | | | .<- Home::/32 / Home::/32 ->..| | --------- ... ; ... --------- ..... / .. . ;.... ...... / .......... In return, each site advertises the aggregation that encompasses the Distributed Home Network aggregation back into the service provider network, to the outside world (the internet). Please note that we have sites and providers here and the prefix=20 announced belongs to the SITE and NOT to the provider. This i read as=20 PI for end sites being injected in the DFZ. I couldn't find the concept of mobile ISP consortiums anywhere in the=20 draft, which seems to be a key role player in your supposed deployment=20= model (but maybe i missed that...) OTOH, i did see several emails from potential users that claimed that=20 mobile platforms needed PI addressing now these are the users that will be deploying this stuff and they said=20= they want PI for this... > So yes, like anything else, global HAHA could be deployed stupidly, > but guess what, people will know better. > Now maybe may idea about how to deploy this is indeed stupid, but what=20= can i say, this is what i could infer from what i read in the draft and=20= the comments in the mailing list.... probably i am too short... regards, marcelo > Pascal > From nemo-bounces@ietf.org Fri Apr 21 05:04:31 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWrYw-0005CH-Ub; Fri, 21 Apr 2006 05:04:26 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWrYu-0005CC-Uz for nemo@ietf.org; Fri, 21 Apr 2006 05:04:24 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWrYu-0001Yl-Kp for nemo@ietf.org; Fri, 21 Apr 2006 05:04:24 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 21 Apr 2006 11:04:25 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3L94MUE010214; Fri, 21 Apr 2006 11:04:22 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Fri, 21 Apr 2006 11:04:22 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: RO approaches and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Fri, 21 Apr 2006 11:03:31 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC02151340@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: RO approaches and the proposed charter (was Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZlHjyS/yMDFJvDQ4e6zSCEj8itNQAAutUg From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 21 Apr 2006 09:04:22.0628 (UTC) FILETIME=[94CB4240:01C66522] X-Spam-Score: 0.0 (/) X-Scan-Signature: 0ddefe323dd869ab027dbfff7eff0465 Cc: Hesham Soliman , Vijay.Devarapalli@AzaireNet.com, ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > >5.1 Initial routing > >5.1.1 External routing > > Sites are expected to be connected locally to the internet, via the > network of one or more service provider. Each site has a default > route to the internet via that connection. > > . .. / > --------- ........ ..;. --------- > | | .. / ..... | | > | ::/0 -> .... ; ... <- ::/0 | > | = =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3DHAHA=3DTUNNEL=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D | > | | .... ; .... | | > | | .<- Home::/32 / Home::/32 ->..| | > --------- ... ; ... --------- > ..... / .. > . ;.... ...... > / .......... > > > In return, each site advertises the aggregation that encompasses the > Distributed Home Network aggregation back into the service provider > network, to the outside world (the internet). > >Please note that we have sites and providers here and the prefix >announced belongs to the SITE and NOT to the provider. This i read as >PI for end sites being injected in the DFZ. > [Pascal] I did not have time yet to change the text since you raised the point. I'll have to word that that the distributed home network can be shared among ISPs, and maybe place a shorter prefix in the examples. That's a good comment to improve the draft, but that does not change the model at all. >Now maybe may idea about how to deploy this is indeed stupid, but what >can i say, this is what i could infer from what i read in the draft and >the comments in the mailing list.... probably i am too short... I never understood that you thought it should be deployed that way, quite the contrary... And I agreed already to change the draft according to your comments... I expect only minor impacts on the text, though... Pascal From nemo-bounces@ietf.org Fri Apr 21 05:51:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWsIv-00014v-6R; Fri, 21 Apr 2006 05:51:57 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWsIu-00014q-Fq for nemo@ietf.org; Fri, 21 Apr 2006 05:51:56 -0400 Received: from h-74-0-36-190.snfccasy.covad.net ([74.0.36.190] helo=multihop.net) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWsIs-0004VR-5J for nemo@ietf.org; Fri, 21 Apr 2006 05:51:56 -0400 Received: by deimos.multihop.net (Postfix, from userid 1013) id 526768AD149; Fri, 21 Apr 2006 02:51:38 -0700 (PDT) X-Spam-Score: -4.3 X-Spam-Checker-Version: SpamAssassin 3.1.0 (2005-09-13) on deimos.multihop.net X-Spam-Level: X-Spam-Status: No, score=-4.3 required=5.0 tests=ALL_TRUSTED,AWL,BAYES_00 autolearn=ham version=3.1.0 Received: from [192.168.4.16] (wideload-eth.tehama.multihop.net [192.168.4.16]) by deimos.multihop.net (Postfix) with ESMTP id F14C08ABA9A; Fri, 21 Apr 2006 02:51:37 -0700 (PDT) In-Reply-To: <108F5BE9DC629140A6E119C8BCDC28FC0109AAE2@XCH-NW-7V1.nw.nos.boeing.com> References: <108F5BE9DC629140A6E119C8BCDC28FC0109AAE2@XCH-NW-7V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <2A8C776A-38A9-4ECD-BD14-A95846A937A7@kniveton.com> Content-Transfer-Encoding: 7bit From: "T.J. Kniveton" Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE:About route aggregation in global HAHA Date: Fri, 21 Apr 2006 02:51:50 -0700 To: "Dul, Andrew L" X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.1 (/) X-Scan-Signature: 7a6398bf8aaeabc7a7bb696b6b0a2aad Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org On Apr 14, 2006, at 12:30 PM, Dul, Andrew L wrote: > I think people are referring to the presentation I made at IETF 62. A > copy of the presentation was unfortunately omitted from the > proceedings. Sorry it never made it to the proceedings, and I couldn't find a copy of it locally, so thanks for reposting. I put it on the NEMO IETF62 slides web page at: http://www.mobilenetworks.org/nemo/ietf62/nemo-ietf62-global-mobility- bgp.pdf From nemo-bounces@ietf.org Fri Apr 21 06:12:43 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWscy-00020l-Lz; Fri, 21 Apr 2006 06:12:40 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWscw-00020f-L8 for nemo@ietf.org; Fri, 21 Apr 2006 06:12:38 -0400 Received: from motgate5.mot.com ([144.189.100.105]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWscw-0006SO-9x for nemo@ietf.org; Fri, 21 Apr 2006 06:12:38 -0400 Received: from az33exr03.mot.com (az33exr03.mot.com [10.64.251.233]) by motgate5.mot.com (8.12.11/Motgate5) with ESMTP id k3LAOH77020488; Fri, 21 Apr 2006 03:24:18 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by az33exr03.mot.com (8.13.1/8.13.0) with ESMTP id k3LAT008014430; Fri, 21 Apr 2006 05:29:01 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id B04BC865980; Fri, 21 Apr 2006 12:12:28 +0200 (CEST) Message-ID: <4448B00C.2080200@motorola.com> Date: Fri, 21 Apr 2006 12:12:28 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Vijay Devarapalli Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <4446B3CA.3070302@azairenet.com> <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> <4447C434.6050703@azairenet.com> In-Reply-To: <4447C434.6050703@azairenet.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by motgate5.mot.com id k3LAOH77020488 X-Spam-Score: 0.0 (/) X-Scan-Signature: 769a46790fb42fbb0b0cc700c82f7081 Cc: marcelo bagnulo braun , nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Vijay Devarapalli wrote: > marcelo bagnulo braun wrote: >> >> El 20/04/2006, a las 1:03, Vijay Devarapalli escribi=F3: >> >>> marcelo bagnulo braun wrote: >>> >>>>> Let me give you an example. Say that 3-4 ISPs from different=20 >>>>> continents >>>>> form a mobile ISP consortium. >>>>> >>>>> That consortium obtains a short prefix from one of the NICs, call=20 >>>>> it the >>>>> mobile prefix. Internally, the consortium shares that addressing sp= ace >>>>> between members, leaving big chunks free for new comers. The rule i= s >>>>> that only the big aggregation is redistributed over the Internet BG= P, >>>>> from most POP of members ISPs. But internally, the consortium of=20 >>>>> service >>>>> providers runs a second BGP which deals with the subnetting of mobi= le >>>>> prefix, over a mesh of tunnels and private links. >>>> but this seems to be complettelly different that what other people=20 >>>> have been claiming here on the list. >>>> Past comments have strongly argued that global mobile network need=20 >>>> PI addressing and global HAHA seemed a mechanisms to provide access=20 >>>> to this PI prefix that the airline company gets. >>> >>> well, your home prefix, along with the set of mobile >>> network prefixes is a set of PI addresses. as long >>> as the distributed set of home agents let the mobile >>> network keep the same home address and mobile network >>> prefix while providing mobility, you have a solution. >>> >> >> yes, so the contribution of this solution to the global routing table=20 >> would be one entry per home prefix >=20 > since global HAHA is not a generic route optimization > solution, What is global HAHA for if not for shortening paths? Isn't its main=20 goal to "use a HA that is closer to MR"? Alex From nemo-bounces@ietf.org Fri Apr 21 06:12:48 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWsd5-000224-VU; Fri, 21 Apr 2006 06:12:47 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWsd5-00021z-1g for nemo@ietf.org; Fri, 21 Apr 2006 06:12:47 -0400 Received: from h-74-0-36-190.snfccasy.covad.net ([74.0.36.190] helo=multihop.net) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWsd4-0006Sk-Ee for nemo@ietf.org; Fri, 21 Apr 2006 06:12:47 -0400 Received: by deimos.multihop.net (Postfix, from userid 1013) id 91EBB8AD162; Fri, 21 Apr 2006 03:12:30 -0700 (PDT) X-Spam-Score: -4.3 X-Spam-Checker-Version: SpamAssassin 3.1.0 (2005-09-13) on deimos.multihop.net X-Spam-Level: X-Spam-Status: No, score=-4.3 required=5.0 tests=ALL_TRUSTED,AWL,BAYES_00 autolearn=ham version=3.1.0 Received: from [192.168.4.16] (wideload-eth.tehama.multihop.net [192.168.4.16]) by deimos.multihop.net (Postfix) with ESMTP id 5526A8ABBC0 for ; Fri, 21 Apr 2006 03:12:27 -0700 (PDT) Mime-Version: 1.0 (Apple Message framework v749.3) Content-Transfer-Encoding: 7bit Message-Id: Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: ml-nemo WG From: "T.J. Kniveton" Date: Fri, 21 Apr 2006 03:12:40 -0700 X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.1 (/) X-Scan-Signature: b4a0a5f5992e2a4954405484e7717d8c Subject: [nemo] Rewording of RO work in the charter X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Folks, I have been watching the conversation started by Marcelo a couple of weeks ago, regarding the rewording of the (n,n,n) item in the charter. (Sorry I haven't been commenting lately). We need to figure out how to massage the concepts being batted around so that we have something tractable that we can work on and complete. From the discussion so far, I tend to think: 1) We need somewhat more concrete idea of the requirements for the type of situation we want to solve. I agree with Andrew that we don't want to get bogged down in this, and I do believe Terry gave us a list for the Conexion scenario quite a while ago. Perhaps he or Andrew could give an updated list. 2) We need a bit more analysis still on the RO problem space, based on some idea of requirements for the solution to be worked out (according to charter items). Specifically, the effects of having a "Mobile ISP" vs. "ISP-independent mobile home networks", the implications of that for the global BGP tables, how likely it is to get custom agreements between a mobility provider and ISPs it (or its customers) uses, etc. 3) Route optimization - "Performance" as Marcelo called it - avoiding long tunneling (geographically) - or avoiding expensive links. This is what we have discussed so far, and we need a stronger boundary so that we can define what is to be solved in these specific circumstances. Otherwise this walks down the road toward many other long-standing issues. My view is that having a good list of requirements for the RO situation of interest would be helpful here. 4) Aviation - question of "if" to use Mobile IP... NEMO base does use it because it made sense. It's possible we have another deployment case where it does not make sense, but we still want a solution for Network Mobility. However, we have to be careful here. There are some general network mobility problems.. such as multiple ISPs, having high-cost links, and how/whether to inject routes into the BGP tables. I believe Terry already gave us a list of requirements for this type of situation - can we review that? 5) Regarding HAHA - As Pascal mentioned, this wasn't supposed to be "the only" or even "the" solution for RO. It is simply something that addressed a shown need, and so I considered it to be something that could be added for future work. But according to the list items above, the group is not including or excluding any certain solution at this point. We need more definition of what and how to solve, and the implications of various approaches first. I would like to hear the RO-analysis authors chime in on this one. Besides Aviation, I have asked numerous times for anyone who is actually deploying NEMO or NEMO-like technology to give us industry input for requirements. Aside from aviation, I have heard a little bit regarding Japanese auto makers, but not much else. I also used to know things about some large cell phone makers that might be doing these things, but never enough to make a good list of requirements. Remember, our goal here is to reshape the charter to solve problems so that NEMO can be deployed. If there are deployments that are facing these issues, they can be brought forward to be studied at this phase. In conclusion, I would like to have a bit more discussion, and then maybe people can offer suggestions on how specifically to reword charter language and/or change milestones so that these things can be accomplished. From nemo-bounces@ietf.org Fri Apr 21 06:27:20 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWsr8-000864-V1; Fri, 21 Apr 2006 06:27:18 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWsr6-00085z-QC for nemo@ietf.org; Fri, 21 Apr 2006 06:27:16 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWsr6-0007ae-I8 for nemo@ietf.org; Fri, 21 Apr 2006 06:27:16 -0400 Received: from az33exr04.mot.com ([10.64.251.234]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k3LAhoDL015751; Fri, 21 Apr 2006 03:43:51 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by az33exr04.mot.com (8.13.1/8.13.0) with ESMTP id k3LAcnOG029104; Fri, 21 Apr 2006 05:38:50 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 7A257865980; Fri, 21 Apr 2006 12:27:09 +0200 (CEST) Message-ID: <4448B37D.80105@motorola.com> Date: Fri, 21 Apr 2006 12:27:09 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "Pascal Thubert (pthubert)" References: <7892795E1A87F04CADFCCF41FADD00FC02151340@xmb-ams-337.emea.cisco.com> In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC02151340@xmb-ams-337.emea.cisco.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: f60d0f7806b0c40781eee6b9cd0b2135 Cc: marcelo bagnulo braun , Hesham Soliman , Vijay.Devarapalli@AzaireNet.com, ml-nemo WG Subject: [nemo] Re: globalHAHA can not guarantee shorter path benefits; can't predict placement of CN X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Pascal Thubert (pthubert) wrote: >> 5.1 Initial routing >> >> 5.1.1 External routing >> >> Sites are expected to be connected locally to the internet, via the >> network of one or more service provider. Each site has a default >> route to the internet via that connection. >> >> . .. / >> --------- ........ ..;. --------- >> | | .. / ..... | | >> | ::/0 -> .... ; ... <- ::/0 | >> | ============HAHA=TUNNEL=========== | >> | | .... ; .... | | >> | | .<- Home::/32 / Home::/32 ->..| | >> --------- ... ; ... --------- >> ..... / .. >> . ;.... ...... >> / .......... >> >> >> In return, each site advertises the aggregation that encompasses > the >> Distributed Home Network aggregation back into the service provider >> network, to the outside world (the internet). >> >> Please note that we have sites and providers here and the prefix >> announced belongs to the SITE and NOT to the provider. This i read as >> PI for end sites being injected in the DFZ. >> > [Pascal] I did not have time yet to change the text since you raised the > point. I'll have to word that that the distributed home network can be > shared among ISPs, and maybe place a shorter prefix in the examples. > That's a good comment to improve the draft, but that does not change the > model at all. Pascal, do you agree that even if you change the draft, and even if HAHA home networks don't distribute BGP routes in the core Internet, there is no guaranteed globalHAHA benefit in terms of shorter paths? I think there can be many cases where the MR is closer to HA2 (than to HA1) but still the path CN-HA1-MR is much shorter than CN-HA2-MR. GlobalHAHA benefits make strong assumptions not only on the placements of the HAs, but also on the placement of the CNs. It assumes that _if_ CN is placed at a certain position then it's better for MN to use the HA closer to it. But placement of CN is really unpredictable. In some unfortunate case CN can be placed such that even if MR uses what it thinks to be its closest HA, the whole path CN-HA-MR would be shorter if the MR used the HA _farthest_ from it. >> Now maybe may idea about how to deploy this is indeed stupid, but what >> can i say, this is what i could infer from what i read in the draft and >> the comments in the mailing list.... probably i am too short... > > I never understood that you thought it should be deployed that way, > quite the contrary... And I agreed already to change the draft according > to your comments... I expect only minor impacts on the text, though... Would it be possible to insert a section in the draft with an illustration of the deployment case where globalHAHA forces MR to use the HA closest to it and CN is placed such that MR had better used the HA farthest from it? Alex From nemo-bounces@ietf.org Fri Apr 21 06:49:05 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWtC7-0000IA-Px; Fri, 21 Apr 2006 06:48:59 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWtC5-0000I2-TF for nemo@ietf.org; Fri, 21 Apr 2006 06:48:57 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWtC5-0000JN-Bw for nemo@ietf.org; Fri, 21 Apr 2006 06:48:57 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id B9C5C212C61; Fri, 21 Apr 2006 13:48:55 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 7FA72212C59; Fri, 21 Apr 2006 13:48:55 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 15135BDC42; Fri, 21 Apr 2006 13:48:55 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id D1946BDC40; Fri, 21 Apr 2006 13:48:54 +0300 (EEST) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: [nemo] Rewording of RO work in the charter Date: Fri, 21 Apr 2006 13:48:54 +0300 To: "T.J. Kniveton" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 2beba50d0fcdeee5f091c59f204d4365 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi T.J. thanks for the summary as i see it, w.r.t. RO problem i think we need three steps: - First identify potential scenarios where RO is interesting. In this part, i think we have clearly identified that there is a very strong use case for globally moving networks. It would be nice, as you point out, if other folks from other use cases come up and speak up their needs - Second, for each of the identified use cases, define a requirements set for the RO operation/solutions for that specific use case In particular, for the only identified use case, the gloablly moving netwroks, a reqs docuemtns is needed imho. As you point out, there seems to be preliminary work in this space, so i guess we have somewhere to start from - Third, propose solutions for the use cases identified that are compliant with the identified scenarios I guess the first step needs to be done here and now, in order to=20 determine in which problems we will be working on, and the second and=20 third steps should be reflected in the charter as specific charter=20 items, for each of the use cases we identify during this charter=20 disucssion. So, again, this is the moment imho for people with other use cases to=20 bring them to the list regards, marcelo El 21/04/2006, a las 13:12, T.J. Kniveton escribi=F3: > Folks, > > I have been watching the conversation started by Marcelo a couple of=20= > weeks ago, regarding the rewording of the (n,n,n) item in the charter.=20= > (Sorry I haven't been commenting lately). > > We need to figure out how to massage the concepts being batted around=20= > so that we have something tractable that we can work on and complete.=20= > =46rom the discussion so far, I tend to think: > > 1) We need somewhat more concrete idea of the requirements for the=20 > type of situation we want to solve. I agree with Andrew that we don't=20= > want to get bogged down in this, and I do believe Terry gave us a list=20= > for the Conexion scenario quite a while ago. Perhaps he or Andrew=20 > could give an updated list. > > 2) We need a bit more analysis still on the RO problem space, based on=20= > some idea of requirements for the solution to be worked out (according=20= > to charter items). Specifically, the effects of having a "Mobile ISP"=20= > vs. "ISP-independent mobile home networks", the implications of that=20= > for the global BGP tables, how likely it is to get custom agreements=20= > between a mobility provider and ISPs it (or its customers) uses, etc. > > 3) Route optimization - "Performance" as Marcelo called it - avoiding=20= > long tunneling (geographically) - or avoiding expensive links. This is=20= > what we have discussed so far, and we need a stronger boundary so that=20= > we can define what is to be solved in these specific circumstances.=20 > Otherwise this walks down the road toward many other long-standing=20 > issues. My view is that having a good list of requirements for the RO=20= > situation of interest would be helpful here. > > 4) Aviation - question of "if" to use Mobile IP... NEMO base does use=20= > it because it made sense. It's possible we have another deployment=20 > case where it does not make sense, but we still want a solution for=20 > Network Mobility. However, we have to be careful here. > There are some general network mobility problems.. such as multiple=20= > ISPs, having high-cost links, and how/whether to inject routes into=20 > the BGP tables. I believe Terry already gave us a list of requirements=20= > for this type of situation - can we review that? > > 5) Regarding HAHA - As Pascal mentioned, this wasn't supposed to be=20 > "the only" or even "the" solution for RO. It is simply something that=20= > addressed a shown need, and so I considered it to be something that=20 > could be added for future work. But according to the list items above,=20= > the group is not including or excluding any certain solution at this=20= > point. We need more definition of what and how to solve, and the=20 > implications of various approaches first. I would like to hear the=20 > RO-analysis authors chime in on this one. > > Besides Aviation, I have asked numerous times for anyone who is=20 > actually deploying NEMO or NEMO-like technology to give us industry=20 > input for requirements. Aside from aviation, I have heard a little bit=20= > regarding Japanese auto makers, but not much else. I also used to know=20= > things about some large cell phone makers that might be doing these=20 > things, but never enough to make a good list of requirements. > > Remember, our goal here is to reshape the charter to solve problems so=20= > that NEMO can be deployed. If there are deployments that are facing=20 > these issues, they can be brought forward to be studied at this phase. > > In conclusion, I would like to have a bit more discussion, and then=20 > maybe people can offer suggestions on how specifically to reword=20 > charter language and/or change milestones so that these things can be=20= > accomplished. > From nemo-bounces@ietf.org Fri Apr 21 07:37:38 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWtx3-0002PI-BQ; Fri, 21 Apr 2006 07:37:29 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWtx2-0002PD-Au for nemo@ietf.org; Fri, 21 Apr 2006 07:37:28 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWtx1-0001xG-0A for nemo@ietf.org; Fri, 21 Apr 2006 07:37:28 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 21 Apr 2006 13:37:26 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3LBbQUE026014; Fri, 21 Apr 2006 13:37:26 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Fri, 21 Apr 2006 13:37:26 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Fri, 21 Apr 2006 13:37:18 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC02151418@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZlLC/BDIglMEbMT9SWZDZR4gNAcwAC29qQ From: "Pascal Thubert \(pthubert\)" To: "Alexandru Petrescu" , "Vijay Devarapalli" X-OriginalArrivalTime: 21 Apr 2006 11:37:26.0139 (UTC) FILETIME=[F697D4B0:01C66537] X-Spam-Score: 0.0 (/) X-Scan-Signature: 52f7a77164458f8c7b36b66787c853da Cc: marcelo bagnulo braun , nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org One of the motivations is to get rid of the issues associated with the = Home Link (single site, single point of failure) and enable distributing = Home and L3. RO comes when we introduce the satellites and route = projection, which is another aspect of the draft. Pascal >-----Original Message----- >From: Alexandru Petrescu [mailto:alexandru.petrescu@motorola.com] >Sent: Friday, April 21, 2006 12:12 PM >To: Vijay Devarapalli >Cc: marcelo bagnulo braun; nemo@ietf.org; Pascal Thubert (pthubert) >Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] = RE: About route aggregation >in global HAHA > >Vijay Devarapalli wrote: >> marcelo bagnulo braun wrote: >>> >>> El 20/04/2006, a las 1:03, Vijay Devarapalli escribi=F3: >>> >>>> marcelo bagnulo braun wrote: >>>> >>>>>> Let me give you an example. Say that 3-4 ISPs from different >>>>>> continents >>>>>> form a mobile ISP consortium. >>>>>> >>>>>> That consortium obtains a short prefix from one of the NICs, call >>>>>> it the >>>>>> mobile prefix. Internally, the consortium shares that addressing = space >>>>>> between members, leaving big chunks free for new comers. The rule = is >>>>>> that only the big aggregation is redistributed over the Internet = BGP, >>>>>> from most POP of members ISPs. But internally, the consortium of >>>>>> service >>>>>> providers runs a second BGP which deals with the subnetting of = mobile >>>>>> prefix, over a mesh of tunnels and private links. >>>>> but this seems to be complettelly different that what other people >>>>> have been claiming here on the list. >>>>> Past comments have strongly argued that global mobile network need >>>>> PI addressing and global HAHA seemed a mechanisms to provide = access >>>>> to this PI prefix that the airline company gets. >>>> >>>> well, your home prefix, along with the set of mobile >>>> network prefixes is a set of PI addresses. as long >>>> as the distributed set of home agents let the mobile >>>> network keep the same home address and mobile network >>>> prefix while providing mobility, you have a solution. >>>> >>> >>> yes, so the contribution of this solution to the global routing = table >>> would be one entry per home prefix >> >> since global HAHA is not a generic route optimization >> solution, > >What is global HAHA for if not for shortening paths? Isn't its main >goal to "use a HA that is closer to MR"? > >Alex From nemo-bounces@ietf.org Fri Apr 21 08:26:31 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWuiO-0005IH-6q; Fri, 21 Apr 2006 08:26:24 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWuiM-0005IC-NL for nemo@ietf.org; Fri, 21 Apr 2006 08:26:22 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWuiL-0003bR-GL for nemo@ietf.org; Fri, 21 Apr 2006 08:26:22 -0400 Received: from il06exr04.mot.com (il06exr04.mot.com [129.188.137.134]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k3LCgwRD002459; Fri, 21 Apr 2006 05:42:58 -0700 (MST) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by il06exr04.mot.com (8.13.1/8.13.0) with ESMTP id k3LCe9qV024860; Fri, 21 Apr 2006 07:40:09 -0500 (CDT) Message-ID: <4448CF68.60609@motorola.com> Date: Fri, 21 Apr 2006 14:26:16 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "Pascal Thubert (pthubert)" Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC02151418@xmb-ams-337.emea.cisco.com> In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC02151418@xmb-ams-337.emea.cisco.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 2409bba43e9c8d580670fda8b695204a Cc: marcelo bagnulo braun , nemo@ietf.org, Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Pascal Thubert (pthubert) wrote: > One of the motivations is to get rid of the issues associated with > the Home Link (single site, single point of failure) Address "single site - single point of failure"???? I could understand addressing single HA - single point of failure (in which case multiple HAs can co-exist on the home link), but "single site - single point of failure" can't be addressed by any protocol, should better be addressed by some administrative organizations. Or by shim6. Good food for thought but no protocol to be designed here. Imagine I want to test that protocol; how could I inflict some failure to some site to illustrate the benefits of my protocol? > and enable distributing Home and L3. I am afraid the network is made in such a way that IP addresses can't be distributed. Have you looked at shim6? > RO comes when we introduce the satellites and route projection, which > is another aspect of the draft. The draft seems to have too many aspects it tries to deal with. Alex From nemo-bounces@ietf.org Fri Apr 21 09:15:15 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWvTe-0004bl-CJ; Fri, 21 Apr 2006 09:15:14 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWvTd-0004Yn-Bh for nemo@ietf.org; Fri, 21 Apr 2006 09:15:13 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWtlj-0001mL-1R for nemo@ietf.org; Fri, 21 Apr 2006 07:25:47 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FWtin-0006dT-Dg for nemo@ietf.org; Fri, 21 Apr 2006 07:22:46 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 21 Apr 2006 13:22:44 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3LBMfUE021940; Fri, 21 Apr 2006 13:22:41 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Fri, 21 Apr 2006 13:22:41 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: RO approaches and the proposed charter (was Re: About theproposed charter and global HAHA (was Re: [nemo] RE: Aboutroute aggregation in global HAHA Date: Fri, 21 Apr 2006 13:21:13 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021513FB@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: RO approaches and the proposed charter (was Re: About theproposed charter and global HAHA (was Re: [nemo] RE: Aboutroute aggregation in global HAHA Thread-Index: AcZlHjyS/yMDFJvDQ4e6zSCEj8itNQAAutUgAAHMVoA= From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 21 Apr 2006 11:22:41.0596 (UTC) FILETIME=[E75D47C0:01C66535] X-Spam-Score: -0.8 (/) X-Scan-Signature: 0a7aa2e6e558383d84476dc338324fab Cc: Hesham Soliman , ml-nemo WG , Vijay.Devarapalli@AzaireNet.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >> >>Please note that we have sites and providers here and the prefix >>announced belongs to the SITE and NOT to the provider. This i read as >>PI for end sites being injected in the DFZ. >> [Pascal] Hi Marcelo:=20 Again, thanks for your comments on global HAHA. What do you think of the following changes: Before: In return, each site advertises the aggregation that encompasses the Distributed Home Network aggregation back into the service provider network, to the outside world (the internet). Proposed text: In return, each site advertises a Home aggregation to the internet. The Home aggregation has a very short prefix which should be partitioned amongst a number of Service Providers and subnetted to serve as Distributed Home Networks for their customers. One could visualize this aggregation as a subway for Mobile Nodes. Before: In each area, border HAs are elected to mesh with peers in other sites. Again, they discover each other, mesh using tunnels, and form a backbone. It might be preferable to form a fully meshed backbone, in order to limit the cost of routing between sites. Also, making the backbone a transit area enables the use of a specific HA for the HAHA protocol that acts as an OSPF Designated Router or a BGP Route Reflector. Proposed text: In each site, border HAs are elected to mesh with peers in other sites. Sites are interconnected over a mesh tunnels and private links. Routing between sites obeys the traditional rules of the Internet, using an Exterior Gateway Protocol (like BGP) between sites of different service providers, and an IGP within a Distributed Home Network. Between sites of a given Distributed Home Network, it might be preferable to form a fully meshed backbone, in order to limit the cost of routing and optimize the paths. Also: replace Home://32 by Home::/16 which might make more sense in this space. What do you think? Pascal From nemo-bounces@ietf.org Fri Apr 21 10:39:09 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWwmm-0002Ir-Kl; Fri, 21 Apr 2006 10:39:04 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWwml-0002Im-Jh for nemo@ietf.org; Fri, 21 Apr 2006 10:39:03 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWwmk-00045U-Aa for nemo@ietf.org; Fri, 21 Apr 2006 10:39:03 -0400 Received: from il06exr04.mot.com (il06exr04.mot.com [129.188.137.134]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k3LEtb4a000341; Fri, 21 Apr 2006 07:55:41 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by il06exr04.mot.com (8.13.1/8.13.0) with ESMTP id k3LEqkbf026438; Fri, 21 Apr 2006 09:52:47 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 968B2865980; Fri, 21 Apr 2006 16:38:56 +0200 (CEST) Message-ID: <4448EE80.3030809@motorola.com> Date: Fri, 21 Apr 2006 16:38:56 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "T.J. Kniveton" Subject: Re: [nemo] Rewording of RO work in the charter References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: f60d0f7806b0c40781eee6b9cd0b2135 Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org T.J. Kniveton wrote: > 4) Aviation - question of "if" to use Mobile IP... NEMO base does use > it because it made sense. It's possible we have another deployment > case where it does not make sense, but we still want a solution for > Network Mobility. However, we have to be careful here. I agree we need to be careful. The WG was founded on a Mobile IP basis. However, the WG has somehow monopolized the word "Mobile Router" too. I'd say that use of Mobile IP on the Mobile Router makes sense in many circumstances. But it's not the only means that people have tested and deployed Mobile Routers. A hub equipment manufacturer sells a "Mobile Router" that doesn't do Mobile IP, but does IP and 2 wireless interfaces and an Ethernet. It's "Mobile" in that it attaches to wireless networks that allow it to move. If one is ready to accept that the session continuity and permanent reachability at a Home Address is not necessarily a strong requirement in some cases (at least not in terms of permanent _IP_ address) then a different framework for Mobile Router protocols can be planned. Another point is that when we started the WG we were saying it makes no sense to put a NAT on the MR. However, experience shows that some testbeds with MRs do just that - run NAT and DHCP on the Mobile Router. Other problems can come from the PAN world. There's really a growing number of small devices with multiple wireless interfaces running IP. But in only a few cases do they communicate at IP level. They mostly use the IP capability to talk straight to the infrastructure. If they could go through a common Mobile Router (which itself could change from Mobile Router to simple LFN) then the IP routing will be involved. > There are some general network mobility problems.. such as multiple > ISPs, having high-cost links, and how/whether to inject routes into > the BGP tables. I believe Terry already gave us a list of > requirements for this type of situation - can we review that? > > 5) Regarding HAHA - As Pascal mentioned, this wasn't supposed to be > "the only" or even "the" solution for RO. It is simply something that > addressed a shown need, and so I considered it to be something that > could be added for future work. But according to the list items > above, the group is not including or excluding any certain solution > at this point. We need more definition of what and how to solve, and > the implications of various approaches first. I would like to hear > the RO-analysis authors chime in on this one. > > Besides Aviation, I have asked numerous times for anyone who is > actually deploying NEMO or NEMO-like technology to give us industry > input for requirements. Aside from aviation, I have heard a little > bit regarding Japanese auto makers, but not much else. I also used to > know things about some large cell phone makers that might be doing > these things, but never enough to make a good list of requirements. > > Remember, our goal here is to reshape the charter to solve problems > so that NEMO can be deployed. I would be interested in working on topics solving how "Mobile Routers" are deployed, not necessarily "NEMO". > If there are deployments that are facing these issues, they can be > brought forward to be studied at this phase. Some issues I see about deployments is that session continuity and reachability at a permanent address is not seen as an equivalent to Mobile IP, and different technologies offer it sometimes. Alex From nemo-bounces@ietf.org Fri Apr 21 13:16:15 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWzEn-0004fR-KP; Fri, 21 Apr 2006 13:16:09 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FWzEm-0004fJ-QX for nemo@ietf.org; Fri, 21 Apr 2006 13:16:08 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FWxYA-0006fR-Cx for nemo@ietf.org; Fri, 21 Apr 2006 11:28:02 -0400 Received: from slb-smtpout-01.boeing.com ([130.76.64.48]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FWxNs-0000pw-Lh for nemo@ietf.org; Fri, 21 Apr 2006 11:17:29 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by slb-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id IAA08369; Fri, 21 Apr 2006 08:17:11 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k3LFHAw24514; Fri, 21 Apr 2006 08:17:10 -0700 (PDT) Received: from XCH-NW-8V1.nw.nos.boeing.com ([130.247.55.69]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 21 Apr 2006 08:17:05 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Rewording of RO work in the charter Date: Fri, 21 Apr 2006 08:17:04 -0700 Message-ID: <0D090F1E0F5536449C7E6527AFFA280A21BF57@XCH-NW-8V1.nw.nos.boeing.com> In-Reply-To: <4448EE80.3030809@motorola.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Rewording of RO work in the charter Thread-Index: AcZlUf3hvjOSdceCTZWrk+ymeRYviQABJBsA From: "Davis, Terry L" To: "Alexandru Petrescu" , "T.J. Kniveton" X-OriginalArrivalTime: 21 Apr 2006 15:17:05.0446 (UTC) FILETIME=[A6127C60:01C66556] X-Spam-Score: -2.6 (--) X-Scan-Signature: 34d35111647d654d033d58d318c0d21a Cc: ml-nemo WG X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org TJ I'll try to put out an update next week on aviation considerations. Take care Terry > -----Original Message----- > From: Alexandru Petrescu [mailto:alexandru.petrescu@motorola.com] > Sent: Friday, April 21, 2006 7:39 AM > To: T.J. Kniveton > Cc: ml-nemo WG > Subject: Re: [nemo] Rewording of RO work in the charter >=20 > T.J. Kniveton wrote: > > 4) Aviation - question of "if" to use Mobile IP... NEMO base does use > > it because it made sense. It's possible we have another deployment > > case where it does not make sense, but we still want a solution for > > Network Mobility. However, we have to be careful here. >=20 > I agree we need to be careful. The WG was founded on a Mobile IP basis. > However, the WG has somehow monopolized the word "Mobile Router" too. >=20 > I'd say that use of Mobile IP on the Mobile Router makes sense in many > circumstances. But it's not the only means that people have tested and > deployed Mobile Routers. A hub equipment manufacturer sells a "Mobile > Router" that doesn't do Mobile IP, but does IP and 2 wireless interfaces > and an Ethernet. It's "Mobile" in that it attaches to wireless networks > that allow it to move. >=20 > If one is ready to accept that the session continuity and permanent > reachability at a Home Address is not necessarily a strong requirement > in some cases (at least not in terms of permanent _IP_ address) then a > different framework for Mobile Router protocols can be planned. >=20 > Another point is that when we started the WG we were saying it makes no > sense to put a NAT on the MR. However, experience shows that some > testbeds with MRs do just that - run NAT and DHCP on the Mobile Router. >=20 > Other problems can come from the PAN world. There's really a growing > number of small devices with multiple wireless interfaces running IP. > But in only a few cases do they communicate at IP level. They mostly > use the IP capability to talk straight to the infrastructure. If they > could go through a common Mobile Router (which itself could change from > Mobile Router to simple LFN) then the IP routing will be involved. >=20 > > There are some general network mobility problems.. such as multiple > > ISPs, having high-cost links, and how/whether to inject routes into > > the BGP tables. I believe Terry already gave us a list of > > requirements for this type of situation - can we review that? > > > > 5) Regarding HAHA - As Pascal mentioned, this wasn't supposed to be > > "the only" or even "the" solution for RO. It is simply something that > > addressed a shown need, and so I considered it to be something that > > could be added for future work. But according to the list items > > above, the group is not including or excluding any certain solution > > at this point. We need more definition of what and how to solve, and > > the implications of various approaches first. I would like to hear > > the RO-analysis authors chime in on this one. > > > > Besides Aviation, I have asked numerous times for anyone who is > > actually deploying NEMO or NEMO-like technology to give us industry > > input for requirements. Aside from aviation, I have heard a little > > bit regarding Japanese auto makers, but not much else. I also used to > > know things about some large cell phone makers that might be doing > > these things, but never enough to make a good list of requirements. > > > > Remember, our goal here is to reshape the charter to solve problems > > so that NEMO can be deployed. >=20 > I would be interested in working on topics solving how "Mobile Routers" > are deployed, not necessarily "NEMO". >=20 > > If there are deployments that are facing these issues, they can be > > brought forward to be studied at this phase. >=20 > Some issues I see about deployments is that session continuity and > reachability at a permanent address is not seen as an equivalent to > Mobile IP, and different technologies offer it sometimes. >=20 > Alex >=20 From nemo-bounces@ietf.org Fri Apr 21 15:02:33 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FX0tV-0006Qm-PO; Fri, 21 Apr 2006 15:02:17 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FX0tU-0006LT-JT for nemo@ietf.org; Fri, 21 Apr 2006 15:02:16 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FX0tU-0000w4-AT for nemo@ietf.org; Fri, 21 Apr 2006 15:02:16 -0400 Received: from [10.1.201.19] ([10.1.201.19]) by bart.corp.azairenet.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 21 Apr 2006 12:02:14 -0700 Message-ID: <44492C35.2020509@azairenet.com> Date: Fri, 21 Apr 2006 12:02:13 -0700 From: Vijay Devarapalli User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Alexandru Petrescu Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <4446B3CA.3070302@azairenet.com> <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> <4447C434.6050703@azairenet.com> <4448B00C.2080200@motorola.com> In-Reply-To: <4448B00C.2080200@motorola.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 21 Apr 2006 19:02:14.0714 (UTC) FILETIME=[1A3931A0:01C66576] X-Spam-Score: 0.0 (/) X-Scan-Signature: 7baded97d9887f7a0c7e8a33c2e3ea1b Cc: marcelo bagnulo braun , nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Alexandru Petrescu wrote: >> since global HAHA is not a generic route optimization >> solution, > > What is global HAHA for if not for shortening paths? Isn't its main > goal to "use a HA that is closer to MR"? global HAHA can do some sort of route optimization. but we didn't start working on it to solve the route optimization problem. the history goes some thing like this 1. we needed a solution where multiple home agents on the same home link can share the binding cache. this was first mentioned four years ago when the 'S' bit in the binding update was being discussed in the Mobile IP working group. 2. then there was a need for load balancing, HA reliability, etc... so we came up with local HAHA 3. now that multiple home agents can serve the same home prefix, we took the concept further and said the home agents can actually be geographically distributed 4. then we heard about the Boeing problem from Terry Davis. thats when we extended HAHA to address mobile networks on airplanes. I am not saying it addresses it completely. its possible that global HAHA sucks at solving the mobile networks on airplanes problem. Pascal would know more about this. so thats the history. it can be used for route optimization, but can't be standardized as _the_ route optimization solution since it is not realistic to assume that every operator would be able to deploy distributed home agents. Vijay From nemo-bounces@ietf.org Fri Apr 21 15:04:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FX0w7-0000DH-Rw; Fri, 21 Apr 2006 15:04:59 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FX0w7-0000D9-3d for nemo@ietf.org; Fri, 21 Apr 2006 15:04:59 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FX0w6-00013W-RD for nemo@ietf.org; Fri, 21 Apr 2006 15:04:59 -0400 Received: from [10.1.201.19] ([10.1.201.19]) by bart.corp.azairenet.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 21 Apr 2006 12:04:57 -0700 Message-ID: <44492CD9.4010500@azairenet.com> Date: Fri, 21 Apr 2006 12:04:57 -0700 From: Vijay Devarapalli User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Alexandru Petrescu Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC02151418@xmb-ams-337.emea.cisco.com> <4448CF68.60609@motorola.com> In-Reply-To: <4448CF68.60609@motorola.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 21 Apr 2006 19:04:57.0868 (UTC) FILETIME=[7B7884C0:01C66576] X-Spam-Score: 0.0 (/) X-Scan-Signature: bb8f917bb6b8da28fc948aeffb74aa17 Cc: marcelo bagnulo braun , nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Alexandru Petrescu wrote: > Pascal Thubert (pthubert) wrote: >> One of the motivations is to get rid of the issues associated with the >> Home Link (single site, single point of failure) > > Address "single site - single point of failure"???? > > I could understand addressing single HA - single point of failure (in > which case multiple HAs can co-exist on the home link), but "single site > - single point of failure" can't be addressed by any protocol, should > better be addressed by some administrative organizations. Or by shim6. > > Good food for thought but no protocol to be designed here. Imagine I > want to test that protocol; how could I inflict some failure to some > site to illustrate the benefits of my protocol? > >> and enable distributing Home and L3. > > I am afraid the network is made in such a way that IP addresses can't be > distributed. Have you looked at shim6? > >> RO comes when we introduce the satellites and route projection, which >> is another aspect of the draft. > > The draft seems to have too many aspects it tries to deal with. not really. thats why we split them into three different drafts. Vijay > > Alex > From nemo-bounces@ietf.org Sat Apr 22 18:13:52 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXQMQ-0006uQ-IY; Sat, 22 Apr 2006 18:13:50 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXQMP-0006uL-0d for nemo@ietf.org; Sat, 22 Apr 2006 18:13:49 -0400 Received: from laposte.rennes.enst-bretagne.fr ([192.44.77.17]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXQMN-00006q-KN for nemo@ietf.org; Sat, 22 Apr 2006 18:13:48 -0400 Received: from localhost (localhost.localdomain [127.0.0.1]) by laposte.rennes.enst-bretagne.fr (8.13.4/8.13.4/2004.10.03) with ESMTP id k3MMDkbX027339 for ; Sun, 23 Apr 2006 00:13:46 +0200 Received: from givry.rennes.enst-bretagne.fr (givry.rennes.enst-bretagne.fr [192.44.77.29]) by laposte.rennes.enst-bretagne.fr (8.13.4/8.13.4/2004.09.01) with ESMTP id k3MMDhXv027335 for ; Sun, 23 Apr 2006 00:13:43 +0200 Received: from givry.rennes.enst-bretagne.fr (localhost.rennes.enst-bretagne.fr [127.0.0.1]) by givry.rennes.enst-bretagne.fr (8.13.1/8.13.1) with ESMTP id k3MMDhil092347 for ; Sun, 23 Apr 2006 00:13:43 +0200 (CEST) (envelope-from dupont@givry.rennes.enst-bretagne.fr) Message-Id: <200604222213.k3MMDhil092347@givry.rennes.enst-bretagne.fr> From: Francis Dupont To: nemo@ietf.org Date: Sun, 23 Apr 2006 00:13:43 +0200 X-Virus-Scanned: amavisd-new at enst-bretagne.fr X-Spam-Score: 0.0 (/) X-Scan-Signature: 08e48e05374109708c00c6208b534009 Subject: [nemo] tunnel all mode X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org There is a discussion about the "tunnel all" mode in the mip6 WG mailing-list which is interesting for NEMO too. The initial message is the mip6 WG Last Call on draft-ietf-mip6-ikev2-ipsec-06.txt. Francis.Dupont@point6.net PS: A diff about the changes made from version 05 to version 06 are at: http://www.dvijay.com/ietf/internet-drafts/mip6/diff-05to06.html From nemo-bounces@ietf.org Sun Apr 23 03:52:43 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXZOZ-0007PN-4b; Sun, 23 Apr 2006 03:52:39 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXZOX-0007O2-Ra for nemo@ietf.org; Sun, 23 Apr 2006 03:52:37 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXZOW-0000KS-Ff for nemo@ietf.org; Sun, 23 Apr 2006 03:52:37 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 23 Apr 2006 09:52:36 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3N7qYUE014218; Sun, 23 Apr 2006 09:52:34 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Sun, 23 Apr 2006 09:52:34 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Sun, 23 Apr 2006 09:52:14 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC0215165F@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZlPtAHH5XnWZifR4uOOLOzPq3kdgBar0fw From: "Pascal Thubert \(pthubert\)" To: "Alexandru Petrescu" X-OriginalArrivalTime: 23 Apr 2006 07:52:34.0776 (UTC) FILETIME=[E1F0B580:01C666AA] X-Spam-Score: 0.0 (/) X-Scan-Signature: 39bd8f8cbb76cae18b7e23f7cf6b2b9f Cc: marcelo bagnulo braun , nemo@ietf.org, Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org and global HAHA (was Re: [nemo] RE: About route aggregation >in global HAHA > >Pascal Thubert (pthubert) wrote: >> One of the motivations is to get rid of the issues associated with >> the Home Link (single site, single point of failure) > >Address "single site - single point of failure"???? > >I could understand addressing single HA - single point of failure (in >which case multiple HAs can co-exist on the home link), but "single site >- single point of failure" can't be addressed by any protocol, should >better be addressed by some administrative organizations. Or by shim6. > >Good food for thought but no protocol to be designed here. Imagine I >want to test that protocol; how could I inflict some failure to some >site to illustrate the benefits of my protocol? > [Pascal] Alex, sometimes I wonder if we waste our time writing such docs as: http://www.ietf.org/internet-drafts/draft-ietf-nemo-ro-problem-statement -02.txt please have a look a chapter 2.2 If you disagree please comment on that draft. At the moment, this is considered as a problem and HAHA addresses it. MIP6 is based in ND on the home link. HAHA replaces that by a routing protocol.=20 This is true for both local and global HAHA. Pascal From nemo-bounces@ietf.org Sun Apr 23 04:24:06 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXZsx-0006sE-KA; Sun, 23 Apr 2006 04:24:03 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXZsw-0006s9-HX for nemo@ietf.org; Sun, 23 Apr 2006 04:24:02 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXZsw-0006zE-7J for nemo@ietf.org; Sun, 23 Apr 2006 04:24:02 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 23 Apr 2006 10:24:03 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3N8NwUE017583; Sun, 23 Apr 2006 10:24:00 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Sun, 23 Apr 2006 10:23:58 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Sun, 23 Apr 2006 10:23:42 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC02151661@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: globalHAHA can not guarantee shorter path benefits; can't predict placement of CN Thread-Index: AcZlLjmU+2ykbcY9RnOjERoSAkCm9ABfhaGA From: "Pascal Thubert \(pthubert\)" To: "Alexandru Petrescu" X-OriginalArrivalTime: 23 Apr 2006 08:23:58.0755 (UTC) FILETIME=[44E11730:01C666AF] X-Spam-Score: 0.0 (/) X-Scan-Signature: 7baded97d9887f7a0c7e8a33c2e3ea1b Cc: marcelo bagnulo braun , Hesham Soliman , Vijay.Devarapalli@AzaireNet.com, ml-nemo WG Subject: [nemo] RE: globalHAHA can not guarantee shorter path benefits; can't predict placement of CN X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >> I never understood that you thought it should be deployed that way, >> quite the contrary... And I agreed already to change the draft according >> to your comments... I expect only minor impacts on the text, though... > >Would it be possible to insert a section in the draft with an >illustration of the deployment case where globalHAHA forces MR to use >the HA closest to it and CN is placed such that MR had better used the >HA farthest from it? > [Pascal] Would you have some proposed text? Note that global HAHA resembles the GPRS core network, where SGSNs would be proxies sites and GGSNs could be Home Sites. In the GTP world, you find those pretty close to where you are. Coupling those features makes a lot of sense and cisco's GGSN already supports Home Agent. Compared to what you can do today with GTP, Global HAHA optimizes further, between proxies, while you can not shortcut between SGSNs today. Obviously, you could take an additional step in some situations, like within a conference room or within a plane. In that case, I'd suggest you consider a MANEMO based solution for local RO coupled with global HAHA for global RO. Global HAHA builds on accepted concepts (NEMO in one hand and GPRS core network in the other) to enable, amongst other things, 4G telephony. As TJ mentions, we have discussed HAHA in the WG for a very long time already, sorry for new comers. The WG keeps improving the text (eg. Marcelo's recent comments on the DFZ). So it could be expected that sometime soon we ask the WG to accept global HAHA as a WG document, which requires charter entry in the first place. Global HAHA is not The RO solution. From the ROPS, it addresses 2.1 and 2.2; and ROSA shows that there are many angles to solve the other RO problems, 2.3 to 2.7, and that there is not a single solution to solve them all. Thus for instance the MANEMO discussions. Pascal From kotoe_mai@ocn.ne.jp Sun Apr 23 08:14:39 2006 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXdU7-0001tR-LZ for NEMO-ARCHIVE@LISTS.IETF.ORG; Sun, 23 Apr 2006 08:14:39 -0400 Received: from [221.207.213.6] (helo=ocn.ne.jp) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXdU5-00027z-7I for NEMO-ARCHIVE@LISTS.IETF.ORG; Sun, 23 Apr 2006 08:14:39 -0400 Received: from bdc6 (unknown [71.82.157.90]) by smtp32 (Coremail) with SMTP id 2mjIzNmKs8UPOUGh.1 for ; Wed, 16 Apr 2003 06:34:06 +0800 (CST) X-Originating-IP: [71.82.157.90] Subject: =?iso-2022-jp?B?GyRCJEEkZyRDJEgjRyNXJEtGLyQrJEokJCEpGyhC?= From: =?shift-jis?B?a2l5b2tv?= To: X-Mailer: Microsoft Outlook Express 6.00.2800.1478 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0090_01C666E3.38973200" X-Priority: 3 X-MSMail-Priority: Normal X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2869 X-Spam-Score: 4.8 (++++) X-Scan-Signature: 7da5a831c477fb6ef97f379a05fb683c This is a multi-part message in MIME format. ------=_NextPart_000_0090_01C666E3.38973200 Content-Type: text/plain; charset="iso-2022-jp" Content-Transfer-Encoding: 7bit GW$B4k2hBh0lCF!*!*(B $BD6C;4|4V#G#W8BDj4k2h!*5U1g=u6b3[:GDc#1#0#0K|1_!A!!:#F|$+$i%9%?!<%H$G$9!*!*(B $B$*$$$7$9$.$^$9$3$N4k2h!*!*!*%,%C%]%j2T$$$8$c$C$F2<$5$$!#(B $B>\:Y$O$3$A$i"*!!F~$j8}(B $B5qH](B pp_info_0001@yahoo.ca ------=_NextPart_000_0090_01C666E3.38973200 Content-Type: text/html; charset="iso-2022-jp" Content-Transfer-Encoding: quoted-printable
GW=1B$B4k=1B(B=1B$B2h=1B(B=1B$BBh=1B(B=1B$B0l=1B(B=1B$BCF=1B(B=1B$B!*=1B(B=1B$B!*=1B(B
 
=1B$BD6C;4|4V#G#W8BDj4k2h!*5U1g=3Du6b3[:GDc#1#0#0K|1_!A!!:#F|$+$= i%9%?!<%H$G$9!*!*=1B(B
 
=1B$B$*$$$7$9$.$^$9$3$N4k2h!*!*!*%,%C%]%j2T$$$8$c$C$F2<$5$$!#=1B= (B
 
=1B$B>\:Y$O$3$A$i"*!!=1B(B=1B$BF~$j8}=1B(B
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
=1B$B5qH]=1B(B
pp_info_0001@yahoo.ca
 
------=_NextPart_000_0090_01C666E3.38973200-- From nemo-bounces@ietf.org Sun Apr 23 14:33:01 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXjOB-0003ji-PM; Sun, 23 Apr 2006 14:32:55 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXjOA-0003jY-PU for nemo@ietf.org; Sun, 23 Apr 2006 14:32:54 -0400 Received: from motgate7.mot.com ([129.188.136.7]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXjOA-00029D-HK for nemo@ietf.org; Sun, 23 Apr 2006 14:32:54 -0400 Received: from az33exr01.mot.com ([10.64.251.231]) by motgate7.mot.com (8.12.11/Motgate7) with ESMTP id k3NJ4M5q020744; Sun, 23 Apr 2006 12:04:22 -0700 (MST) Received: from [144.189.72.159] (mvp-144-189-72-159.corp.mot.com [144.189.72.159]) by az33exr01.mot.com (8.13.1/8.13.0) with ESMTP id k3NIss3P019435; Sun, 23 Apr 2006 13:54:55 -0500 (CDT) Message-ID: <444BC84A.6080104@motorola.com> Date: Sun, 23 Apr 2006 20:32:42 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Vijay Devarapalli Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <4446B3CA.3070302@azairenet.com> <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> <4447C434.6050703@azairenet.com> <4448B00C.2080200@motorola.com> <44492C35.2020509@azairenet.com> In-Reply-To: <44492C35.2020509@azairenet.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 0a7aa2e6e558383d84476dc338324fab Cc: marcelo bagnulo braun , nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Vijay Devarapalli wrote: > Alexandru Petrescu wrote: > >>> since global HAHA is not a generic route optimization >>> solution, >> >> What is global HAHA for if not for shortening paths? Isn't its main >> goal to "use a HA that is closer to MR"? > > global HAHA can do some sort of route optimization. > but we didn't start working on it to solve the > route optimization problem. the history goes some > thing like this > > 1. we needed a solution where multiple home agents > on the same home link can share the binding cache. > this was first mentioned four years ago when the 'S' > bit in the binding update was being discussed in the > Mobile IP working group. I agree with this need and DHAAD method. The HAs stay on the same link. > 2. then there was a need for load balancing, HA > reliability, etc... so we came up with local HAHA Any method to separate HAs to _different_ links, i.e. a router between two HAs serving same HoA, makes very little sense. Be it local or global. If one can do that (two different HAs separated by one or more routers, the HAs serving a unique Home Address) then one does not need Mobile IP. > 3. now that multiple home agents can serve the same > home prefix, we took the concept further and said > the home agents can actually be geographically > distributed Which is not feasible. Could be drawn on paper but not implemented. > 4. then we heard about the Boeing problem from Terry > Davis. thats when we extended HAHA to address mobile > networks on airplanes. I am not saying it addresses > it completely. its possible that global HAHA sucks > at solving the mobile networks on airplanes problem. > Pascal would know more about this. > > so thats the history. Thanks for explaining the conceptual evolution of it. > it can be used for route optimization, but can't be > standardized as _the_ route optimization solution > since it is not realistic to assume that every > operator would be able to deploy distributed home > agents. Ok, Alex From nemo-bounces@ietf.org Sun Apr 23 14:50:21 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXjf1-0007hl-VO; Sun, 23 Apr 2006 14:50:19 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXjf1-0007hf-Bl for nemo@ietf.org; Sun, 23 Apr 2006 14:50:19 -0400 Received: from motgate2.mot.com ([144.189.100.101]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXjf0-0002m2-W6 for nemo@ietf.org; Sun, 23 Apr 2006 14:50:19 -0400 Received: from az33exr01.mot.com (az33exr01.mot.com [10.64.251.231]) by motgate2.mot.com (8.12.11/Motgate2) with ESMTP id k3NJBZ57019338; Sun, 23 Apr 2006 12:11:35 -0700 (MST) Received: from [144.189.72.159] (mvp-144-189-72-159.corp.mot.com [144.189.72.159]) by az33exr01.mot.com (8.13.1/8.13.0) with ESMTP id k3NJCMJj022259; Sun, 23 Apr 2006 14:12:24 -0500 (CDT) Message-ID: <444BCC62.30905@motorola.com> Date: Sun, 23 Apr 2006 20:50:10 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "Pascal Thubert (pthubert)" Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0215165F@xmb-ams-337.emea.cisco.com> In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC0215165F@xmb-ams-337.emea.cisco.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: f4c2cf0bccc868e4cc88dace71fb3f44 Cc: marcelo bagnulo braun , nemo@ietf.org, Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Pascal Thubert (pthubert) wrote: > and global HAHA (was Re: [nemo] RE: About route aggregation >> in global HAHA >> >> Pascal Thubert (pthubert) wrote: >>> One of the motivations is to get rid of the issues associated >>> with the Home Link (single site, single point of failure) >> Address "single site - single point of failure"???? >> >> I could understand addressing single HA - single point of failure >> (in which case multiple HAs can co-exist on the home link), but >> "single > site >> - single point of failure" can't be addressed by any protocol, >> should better be addressed by some administrative organizations. Or >> by shim6. >> >> Good food for thought but no protocol to be designed here. Imagine >> I want to test that protocol; how could I inflict some failure to >> some site to illustrate the benefits of my protocol? >> > [Pascal] > > Alex, sometimes I wonder if we waste our time writing such docs as: > > http://www.ietf.org/internet-drafts/draft-ietf-nemo-ro-problem-statement > -02.txt > > please have a look a chapter 2.2 If you disagree please comment on > that draft. Section 2.2 "Bottleneck in home network" describes that there is a problem because all traffic goes through the home link and Home Agent. The home link may become a bottleneck and a single-point of failure. I do not disagree with that. But it's not because the problem is listed there that (1) globalHAHA is a feasible solution or that (2) anybody needs globalHAHA and (3) globalHAHA may solve in any way the "site failure" problems you were mentioning. I see no direct relationship between aforementioned section 2.2 and globalHAHA. > At the moment, this is considered as a problem and HAHA addresses it. I simply disagree. > MIP6 is based in ND on the home link. It is. > HAHA replaces that by a routing protocol. I somehow agree, and to go further I'd say there are at least a couple deployed routing protocols doing just that. Alex From nemo-bounces@ietf.org Sun Apr 23 14:57:18 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXjlm-0002FV-MZ; Sun, 23 Apr 2006 14:57:18 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXjll-0002FQ-Fc for nemo@ietf.org; Sun, 23 Apr 2006 14:57:17 -0400 Received: from motgate2.mot.com ([144.189.100.101]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXjll-00032V-4Z for nemo@ietf.org; Sun, 23 Apr 2006 14:57:17 -0400 Received: from az33exr03.mot.com (az33exr03.mot.com [10.64.251.233]) by motgate2.mot.com (8.12.11/Motgate2) with ESMTP id k3NJIPmM022244; Sun, 23 Apr 2006 12:18:25 -0700 (MST) Received: from [144.189.72.159] (mvp-144-189-72-159.corp.mot.com [144.189.72.159]) by az33exr03.mot.com (8.13.1/8.13.0) with ESMTP id k3NJDdEt021315; Sun, 23 Apr 2006 14:13:40 -0500 (CDT) Message-ID: <444BCDFD.7070705@motorola.com> Date: Sun, 23 Apr 2006 20:57:01 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: "Pascal Thubert (pthubert)" References: <7892795E1A87F04CADFCCF41FADD00FC02151661@xmb-ams-337.emea.cisco.com> In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC02151661@xmb-ams-337.emea.cisco.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: 50a516d93fd399dc60588708fd9a3002 Cc: marcelo bagnulo braun , Hesham Soliman , Vijay.Devarapalli@AzaireNet.com, ml-nemo WG Subject: [nemo] Re: globalHAHA can not guarantee shorter path benefits; can't predict placement of CN X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Pascal Thubert (pthubert) wrote: >>> I never understood that you thought it should be deployed that >>> way, quite the contrary... And I agreed already to change the >>> draft > according >>> to your comments... I expect only minor impacts on the text, > though... >> Would it be possible to insert a section in the draft with an >> illustration of the deployment case where globalHAHA forces MR to >> use the HA closest to it and CN is placed such that MR had better >> used the HA farthest from it? >> > [Pascal] Would you have some proposed text? I have posted several texts already on this, would you select the most appropriate and paste it first phrase in the abstract? > Note that global HAHA resembles the GPRS core network, where SGSNs > would be proxies sites and GGSNs could be Home Sites. In a GPRS network, or any similar cell network, there is absolutely no need of Mobile IP's encapsulating tunnels. The network is small, while the tunnels are so that they traverse an entire Internet, not such a small network where route updates are feasible. > In the GTP world, you find those pretty close to where you are. > Coupling those features makes a lot of sense and cisco's GGSN already > supports Home Agent. Cisco's routers support lots of protocols. > Compared to what you can do today with GTP, Global HAHA optimizes > further, between proxies, while you can not shortcut between SGSNs > today. I think netlmm concepts are better adapted for this. > Global HAHA builds on accepted concepts (NEMO in one hand and GPRS > core network in the other) to enable, amongst other things, 4G > telephony. As TJ mentions, we have discussed HAHA in the WG for a > very long time already, sorry for new comers. The WG keeps improving > the text (eg. Marcelo's recent comments on the DFZ). So it could be > expected that sometime soon we ask the WG to accept global HAHA as a > WG document, which requires charter entry in the first place. The WG members can still comment on its feasibility, no? Alex From nemo-bounces@ietf.org Sun Apr 23 18:22:01 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXmxm-0000wl-PF; Sun, 23 Apr 2006 18:21:54 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXmxl-0000wg-8H for nemo@ietf.org; Sun, 23 Apr 2006 18:21:53 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXmxk-0004ax-V9 for nemo@ietf.org; Sun, 23 Apr 2006 18:21:53 -0400 Received: from [10.1.210.8] ([10.1.210.8]) by bart.corp.azairenet.com with Microsoft SMTPSVC(6.0.3790.1830); Sun, 23 Apr 2006 15:21:51 -0700 Message-ID: <444BFDFE.6030209@azairenet.com> Date: Sun, 23 Apr 2006 15:21:50 -0700 From: Vijay Devarapalli User-Agent: Thunderbird 1.5.0.2 (Windows/20060308) MIME-Version: 1.0 To: Alexandru Petrescu Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA References: <7892795E1A87F04CADFCCF41FADD00FC0208D018@xmb-ams-337.emea.cisco.com> <476f03fef6d31140a0db4f5d085ddf44@it.uc3m.es> <4446B3CA.3070302@azairenet.com> <5f35a6b0ace73dc0b9ad072134d03141@it.uc3m.es> <4447C434.6050703@azairenet.com> <4448B00C.2080200@motorola.com> <44492C35.2020509@azairenet.com> <444BC84A.6080104@motorola.com> In-Reply-To: <444BC84A.6080104@motorola.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 23 Apr 2006 22:21:51.0446 (UTC) FILETIME=[51BCF760:01C66724] X-Spam-Score: 0.0 (/) X-Scan-Signature: 4b800b1eab964a31702fa68f1ff0e955 Cc: marcelo bagnulo braun , nemo@ietf.org, "Pascal Thubert \(pthubert\)" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Alexandru Petrescu wrote: > Vijay Devarapalli wrote: >> Alexandru Petrescu wrote: >> >>>> since global HAHA is not a generic route optimization >>>> solution, >>> >>> What is global HAHA for if not for shortening paths? Isn't its main >>> goal to "use a HA that is closer to MR"? >> >> global HAHA can do some sort of route optimization. >> but we didn't start working on it to solve the >> route optimization problem. the history goes some >> thing like this >> >> 1. we needed a solution where multiple home agents >> on the same home link can share the binding cache. >> this was first mentioned four years ago when the 'S' >> bit in the binding update was being discussed in the >> Mobile IP working group. > > I agree with this need and DHAAD method. The HAs stay on the same link. > >> 2. then there was a need for load balancing, HA >> reliability, etc... so we came up with local HAHA > > Any method to separate HAs to _different_ links, i.e. a router between > two HAs serving same HoA, makes very little sense. Be it local or > global. If one can do that (two different HAs separated by one or more > routers, the HAs serving a unique Home Address) then one does not need > Mobile IP. I would re-word to say "it can be done without using Mobile IP". sure, I agree with that. consider this as another approach. >> 3. now that multiple home agents can serve the same >> home prefix, we took the concept further and said >> the home agents can actually be geographically >> distributed > > Which is not feasible. Could be drawn on paper but not implemented. disagree. Vijay > >> 4. then we heard about the Boeing problem from Terry >> Davis. thats when we extended HAHA to address mobile >> networks on airplanes. I am not saying it addresses >> it completely. its possible that global HAHA sucks >> at solving the mobile networks on airplanes problem. >> Pascal would know more about this. >> >> so thats the history. > > Thanks for explaining the conceptual evolution of it. > >> it can be used for route optimization, but can't be >> standardized as _the_ route optimization solution >> since it is not realistic to assume that every >> operator would be able to deploy distributed home >> agents. > > Ok, > > Alex > From nemo-bounces@ietf.org Mon Apr 24 05:48:41 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXxgF-000882-Vy; Mon, 24 Apr 2006 05:48:31 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXxgE-00087x-Eb for nemo@ietf.org; Mon, 24 Apr 2006 05:48:30 -0400 Received: from motgate8.mot.com ([129.188.136.8]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXxgD-0001XO-7w for nemo@ietf.org; Mon, 24 Apr 2006 05:48:30 -0400 Received: from az33exr02.mot.com ([10.64.251.232]) by motgate8.mot.com (8.12.11/Motgate7) with ESMTP id k3OA59fd008176; Mon, 24 Apr 2006 03:05:09 -0700 (MST) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by az33exr02.mot.com (8.13.1/8.13.0) with ESMTP id k3OA1ikd003409; Mon, 24 Apr 2006 05:01:45 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 8DC99865980; Mon, 24 Apr 2006 11:48:22 +0200 (CEST) Message-ID: <444C9EE6.7070006@motorola.com> Date: Mon, 24 Apr 2006 11:48:22 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: Francis Dupont Subject: Re: [nemo] tunnel all mode References: <200604222213.k3MMDhil092347@givry.rennes.enst-bretagne.fr> In-Reply-To: <200604222213.k3MMDhil092347@givry.rennes.enst-bretagne.fr> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE X-Spam-Score: 0.0 (/) X-Scan-Signature: cf4fa59384e76e63313391b70cd0dd25 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Francis Dupont wrote: > There is a discussion about the "tunnel all" mode in the mip6 WG > mailing-list which is interesting for NEMO too. Why is the discussion interesting to NEMO? Is it more than just the fact that a NEMO implementation sends BU to HA? Is the discussion somehow different if MNP is reckoned? (Mobile Network Prefix) (as I understand it it revolves around having or not having MIP6 signalling between MN and HA encapsulated. The claimed advantage is easier security with only one SA instead of 4 if it were not encapsulated. The inconvenient is existing implementation don't encapsulate MIP6 signalling.) Alex From nemo-bounces@ietf.org Mon Apr 24 07:21:11 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXz7t-0003WM-Mj; Mon, 24 Apr 2006 07:21:09 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FXz7s-0003WG-Oh for nemo@ietf.org; Mon, 24 Apr 2006 07:21:08 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FXz7s-0005XG-8W for nemo@ietf.org; Mon, 24 Apr 2006 07:21:08 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 1346B212C59; Mon, 24 Apr 2006 14:21:07 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id A0357212C44; Mon, 24 Apr 2006 14:21:06 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 315FCBDC40; Mon, 24 Apr 2006 14:21:06 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id ED712BDC38; Mon, 24 Apr 2006 14:21:05 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021513FB@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021513FB@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <2d2a0ef017d82ead29759c088e1a7109@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Deployment models for global HAHA (was Re: RO approaches and the proposed charter (was Re: About theproposed charter and global HAHA (was Re: [nemo] RE: Aboutroute aggregation in global HAHA Date: Mon, 24 Apr 2006 14:21:10 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 5d7a7e767f20255fce80fa0b77fb2433 Cc: ml-nemo WG , Hesham Soliman , Vijay.Devarapalli@AzaireNet.com X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Pascal, thanks for the effort with the new text, but i was thinking about a=20 deployment consideratiosn section rather than including changes in the=20= solution text... I think that as some folks mentioned earlier (maybe even you did) there=20= are two separated issues: on one hand there is the actual protocol spec and otoh there is how to=20= deploy the solution. after this thread, i think we have identified two possible ways to=20 deploy global HAHA: - one option is to use the mobile ISP consortium model that you=20 described - the other option is to use PI addresses for global companies that=20 need to service and make arrangements with local ISPs in each location,=20= so that the PI prefix is injected. these two alternatives are enabled by the global HAHA protocol, agree? now, as we already discussed they have different effects in the global=20= routing table My suggestion with respect to the actual draft would be to include a=20 deployment section ackowledging the different deployment models and the=20= issues with the different models and eventually making a recomendation=20= that only the scalable model is deployed. I mean, i don't think it is a good approach to mix actual proto spec=20 with deployment issues, like the prefix length for instance. In any case, i guess it should be made clear that the global haha tool=20= enables also non so global routing table firendly deployment models and=20= that this is a liability if we decide to go through this path, since=20 peope may decide to use it in this other ways. Regards, marcelo El 21/04/2006, a las 14:21, Pascal Thubert (pthubert) escribi=F3: > >>> >>> Please note that we have sites and providers here and the prefix >>> announced belongs to the SITE and NOT to the provider. This i read = as >>> PI for end sites being injected in the DFZ. >>> > [Pascal] Hi Marcelo: > > Again, thanks for your comments on global HAHA. > > What do you think of the following changes: > > Before: > > In return, each site advertises the aggregation that encompasses = the > Distributed Home Network aggregation back into the service provider > network, to the outside world (the internet). > > Proposed text: > > In return, each site advertises a Home aggregation to the internet. > The Home aggregation has a very short prefix which should be > partitioned amongst a number of Service Providers and subnetted to > serve as Distributed Home Networks for their customers. One could > visualize this aggregation as a subway for Mobile Nodes. > > Before: > > In each area, border HAs are elected to mesh with peers in other > sites. Again, they discover each other, mesh using tunnels, and=20 > form > a backbone. It might be preferable to form a fully meshed = backbone, > in order to limit the cost of routing between sites. Also, making > the backbone a transit area enables the use of a specific HA for = the > HAHA protocol that acts as an OSPF Designated Router or a BGP Route > Reflector. > > > > Proposed text: > > In each site, border HAs are elected to mesh with peers in other > sites. Sites are interconnected over a mesh tunnels and private > links. Routing between sites obeys the traditional rules of the > Internet, using an Exterior Gateway Protocol (like BGP) between=20 > sites > of different service providers, and an IGP within a Distributed = Home > Network. > > Between sites of a given Distributed Home Network, it might be > preferable to form a fully meshed backbone, in order to limit the > cost of routing and optimize the paths. > > > > Also: replace Home://32 by Home::/16 which might make more sense in=20 > this > space. > > What do you think? > > Pascal > From nemo-bounces@ietf.org Mon Apr 24 08:41:24 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FY0NO-0005eY-1u; Mon, 24 Apr 2006 08:41:14 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FY0NM-0005de-UP for nemo@ietf.org; Mon, 24 Apr 2006 08:41:12 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FY0NL-0001Zf-Fk for nemo@ietf.org; Mon, 24 Apr 2006 08:41:12 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id B6E91212C59; Mon, 24 Apr 2006 15:41:09 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 50C04212C44; Mon, 24 Apr 2006 15:41:09 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 1757ABDC40; Mon, 24 Apr 2006 15:41:09 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id D473CBDC38; Mon, 24 Apr 2006 15:41:08 +0300 (EEST) In-Reply-To: <4448CF68.60609@motorola.com> References: <7892795E1A87F04CADFCCF41FADD00FC02151418@xmb-ams-337.emea.cisco.com> <4448CF68.60609@motorola.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <443d7a3cacc143462ceff171f803db32@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Mon, 24 Apr 2006 15:41:13 +0300 To: Alexandru Petrescu X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 4d87d2aa806f79fed918a62e834505ca Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Alex, El 21/04/2006, a las 15:26, Alexandru Petrescu escribi=F3: > Pascal Thubert (pthubert) wrote: >> One of the motivations is to get rid of the issues associated with=20 >> the Home Link (single site, single point of failure) > > Address "single site - single point of failure"???? > > I could understand addressing single HA - single point of failure (in > which case multiple HAs can co-exist on the home link), but "single=20 > site > - single point of failure" can't be addressed by any protocol, should > better be addressed by some administrative organizations. Or by = shim6. > agree that shim6 could be used to provide some support for this > >> and enable distributing Home and L3. > > I am afraid the network is made in such a way that IP addresses can't=20= > be > distributed. Have you looked at shim6? > same thing >> RO comes when we introduce the satellites and route projection, which >> is another aspect of the draft. > > The draft seems to have too many aspects it tries to deal with. > probably you could even do some shim6 based RO also... that is one of the reasons why i am asking about the requirements and=20 the use cases that we want to address... first we need to define the problem that we want to work on and then=20 move to solution space... regards, marcelo > Alex > From nemo-bounces@ietf.org Tue Apr 25 03:04:51 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYHbB-0001vs-IX; Tue, 25 Apr 2006 03:04:37 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYHb9-0001vV-NX for nemo@ietf.org; Tue, 25 Apr 2006 03:04:35 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYHb8-00057Z-BJ for nemo@ietf.org; Tue, 25 Apr 2006 03:04:35 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 25 Apr 2006 09:04:34 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3P74VUE025553; Tue, 25 Apr 2006 09:04:33 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Tue, 25 Apr 2006 09:04:30 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 25 Apr 2006 09:04:12 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021A9265@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZnnGztCjmdpcbLSN+/xzE9kexT3wAmajog From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" , "Alexandru Petrescu" X-OriginalArrivalTime: 25 Apr 2006 07:04:30.0932 (UTC) FILETIME=[7FDC8940:01C66836] X-Spam-Score: 0.0 (/) X-Scan-Signature: c0bedb65cce30976f0bf60a0a39edea4 Cc: nemo@ietf.org, Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Marcelo, Alexandru: I fail to understand this. Are you saying that SHIM6 can help resolve = the case where the MIP6/NEMO Home Link goes down?=20 Pascal >-----Original Message----- >From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >Sent: Monday, April 24, 2006 2:41 PM >To: Alexandru Petrescu >Cc: Pascal Thubert (pthubert); Vijay Devarapalli; nemo@ietf.org >Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] = RE: About route aggregation >in global HAHA > >Hi Alex, > >El 21/04/2006, a las 15:26, Alexandru Petrescu escribi=F3: > >> Pascal Thubert (pthubert) wrote: >>> One of the motivations is to get rid of the issues associated with >>> the Home Link (single site, single point of failure) >> >> Address "single site - single point of failure"???? >> >> I could understand addressing single HA - single point of failure (in >> which case multiple HAs can co-exist on the home link), but "single >> site >> - single point of failure" can't be addressed by any protocol, should >> better be addressed by some administrative organizations. Or by = shim6. >> > >agree that shim6 could be used to provide some support for this > >> >>> and enable distributing Home and L3. >> >> I am afraid the network is made in such a way that IP addresses can't >> be >> distributed. Have you looked at shim6? >> > >same thing > >>> RO comes when we introduce the satellites and route projection, = which >>> is another aspect of the draft. >> >> The draft seems to have too many aspects it tries to deal with. >> > >probably you could even do some shim6 based RO also... > > >that is one of the reasons why i am asking about the requirements and >the use cases that we want to address... > >first we need to define the problem that we want to work on and then >move to solution space... > >regards, marcelo > > >> Alex >> From nemo-bounces@ietf.org Tue Apr 25 04:20:18 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYImG-0000fM-Im; Tue, 25 Apr 2006 04:20:08 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYImE-0000fG-Ep for nemo@ietf.org; Tue, 25 Apr 2006 04:20:06 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYI2e-0006Py-Tq for nemo@ietf.org; Tue, 25 Apr 2006 03:33:00 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FYHmS-0005EE-E3 for nemo@ietf.org; Tue, 25 Apr 2006 03:16:17 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id EDEA7212C61; Tue, 25 Apr 2006 10:16:14 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 8624B212C59; Tue, 25 Apr 2006 10:16:14 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 4B414BDC40; Tue, 25 Apr 2006 10:16:14 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 1509ABDC38; Tue, 25 Apr 2006 10:16:14 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021A9265@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021A9265@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 25 Apr 2006 10:16:19 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: -2.6 (--) X-Scan-Signature: 6e922792024732fb1bb6f346e63517e4 Cc: nemo@ietf.org, Vijay Devarapalli , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Pascal, El 25/04/2006, a las 10:04, Pascal Thubert (pthubert) escribi=F3: > Marcelo, Alexandru: > > I fail to understand this. Are you saying that SHIM6 can help resolve=20= > the case where the MIP6/NEMO Home Link goes down? > when the HA are distributed among different sites, yes, and in a bgp=20 routing table friendly manner i mean, suppose that you have multiple HAs located in different sites=20 and that each HA has a different PA home prefix assigned. then each=20 nemo has multiple home prefixes and each MNN has multiple addresses one=20= per available home prefix then you can use the shim6 to move from one home prefix to another, in=20= case that there is a failure in one of the HAs that is exactly what the shim6 protocol has been designed to do Again, maybe the global moving networks have some additional=20 requirements that don't fit in this approach, that is why i have been=20 asking for a requirement specification, in order to find out which=20 solution is acceptable Regards, marcelo > Pascal > >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >> Sent: Monday, April 24, 2006 2:41 PM >> To: Alexandru Petrescu >> Cc: Pascal Thubert (pthubert); Vijay Devarapalli; nemo@ietf.org >> Subject: Re: About the proposed charter and global HAHA (was Re:=20 >> [nemo] RE: About route aggregation >> in global HAHA >> >> Hi Alex, >> >> El 21/04/2006, a las 15:26, Alexandru Petrescu escribi=F3: >> >>> Pascal Thubert (pthubert) wrote: >>>> One of the motivations is to get rid of the issues associated with >>>> the Home Link (single site, single point of failure) >>> >>> Address "single site - single point of failure"???? >>> >>> I could understand addressing single HA - single point of failure = (in >>> which case multiple HAs can co-exist on the home link), but "single >>> site >>> - single point of failure" can't be addressed by any protocol, = should >>> better be addressed by some administrative organizations. Or by=20 >>> shim6. >>> >> >> agree that shim6 could be used to provide some support for this >> >>> >>>> and enable distributing Home and L3. >>> >>> I am afraid the network is made in such a way that IP addresses = can't >>> be >>> distributed. Have you looked at shim6? >>> >> >> same thing >> >>>> RO comes when we introduce the satellites and route projection,=20 >>>> which >>>> is another aspect of the draft. >>> >>> The draft seems to have too many aspects it tries to deal with. >>> >> >> probably you could even do some shim6 based RO also... >> >> >> that is one of the reasons why i am asking about the requirements and >> the use cases that we want to address... >> >> first we need to define the problem that we want to work on and then >> move to solution space... >> >> regards, marcelo >> >> >>> Alex >>> > From nemo-bounces@ietf.org Tue Apr 25 05:44:09 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYK5J-0007cE-KC; Tue, 25 Apr 2006 05:43:53 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYK5H-0007c6-LU for nemo@ietf.org; Tue, 25 Apr 2006 05:43:51 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYK5G-0005kR-Vv for nemo@ietf.org; Tue, 25 Apr 2006 05:43:51 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id CC63E212C61; Tue, 25 Apr 2006 12:43:47 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 994F5212C59; Tue, 25 Apr 2006 12:43:47 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 32F52BDC40; Tue, 25 Apr 2006 12:43:47 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id F1FEFBDC38; Tue, 25 Apr 2006 12:43:46 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021A92CB@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021A92CB@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <8c9ad7af1b401f0eb4d6ff77569d37ca@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 25 Apr 2006 12:43:53 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: b8f3559805f7873076212d6f63ee803e Cc: nemo@ietf.org, Vijay Devarapalli , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 25/04/2006, a las 11:12, Pascal Thubert (pthubert) escribi=F3: > Thanks for the explanation Marcelo. Now I see what you have in mind.=20= > This approach is interesting; I guess it takes a mix of MONAMI6 and=20 > SHIM6 to get the result done. > yes > IMHO, though, It looks fairly complex and remote, compared to the use=20= > of existing routing protocols proposed in the global HAHA approach. > well, otoh, if any of the home networks is multihomed, then you need=20 shim6 in anycase to support multihoming features and so this comes "for=20= free" with respect to timeline, i am not sure but i wouldn't expect=20 commercial airlines to start providing IPv6 connectivity in the short=20 term, so i would argue that the timeline of this solution is compatible=20= with the times of the provision of v6 services in commercial flights...=20= (i guess involved people can correct me if i am wrong) > Note that: > > - the NEMO assumption is to have the MR handle the mobility on behalf=20= > of LFNs which are unaware of the mobility. We want a simple operation=20= > for simple Mobile Network Nodes. right, but this wouldn't result in exposing mobility to the nodes,=20 since only the set of stable home address is exposed to the MNN. So,=20 actual this preserves the mobility transparent to the MNN. No problem=20 here AFAICT > > - You are not solving the MIP6 Home Link issue, you are bypassing it.=20= > Global HAHA gets rid of the ND operations on the Home Link, therefore=20= > it removes the associated issues, replacing that mechanism by=20 > traditional routing protocol operations. > agree this is not a substitute for local HAHA, just for global HAHA > I'd say that before we assemble complex things that we do not have, we=20= > could start thinking of using what we mostly have and have been=20 > discussing for some time already. i would say that the relevant parameter is when is the expect time to=20 deploy this... when are we expecting to have v6 services in commercial=20= flights? As i said, perhaps the shim6 approach is not really appropriate for=20 this scenario, but the way we can determine that is if we understand=20 the requirements of this use case BEsides, note that shim6 approach could also be used as a RO mechanisms=20= (if a topologically meaningful CoA is provided to the nodes) and it=20 also is needed for providing multihoming support in multihomed nemos regards, marcelo > > Pascal > >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >> Sent: Tuesday, April 25, 2006 9:16 AM >> To: Pascal Thubert (pthubert) >> Cc: Alexandru Petrescu; nemo@ietf.org; Vijay Devarapalli >> Subject: Re: About the proposed charter and global HAHA (was Re:=20 >> [nemo] RE: About route aggregation >> in global HAHA >> >> Hi Pascal, >> >> El 25/04/2006, a las 10:04, Pascal Thubert (pthubert) escribi=F3: >> >>> Marcelo, Alexandru: >>> >>> I fail to understand this. Are you saying that SHIM6 can help = resolve >>> the case where the MIP6/NEMO Home Link goes down? >>> >> >> when the HA are distributed among different sites, yes, and in a bgp >> routing table friendly manner >> >> i mean, suppose that you have multiple HAs located in different sites >> and that each HA has a different PA home prefix assigned. then each >> nemo has multiple home prefixes and each MNN has multiple addresses=20= >> one >> per available home prefix >> >> then you can use the shim6 to move from one home prefix to another, = in >> case that there is a failure in one of the HAs >> >> that is exactly what the shim6 protocol has been designed to do >> >> Again, maybe the global moving networks have some additional >> requirements that don't fit in this approach, that is why i have been >> asking for a requirement specification, in order to find out which >> solution is acceptable >> >> Regards, marcelo >> >> >>> Pascal >>> >>>> -----Original Message----- >>>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>>> Sent: Monday, April 24, 2006 2:41 PM >>>> To: Alexandru Petrescu >>>> Cc: Pascal Thubert (pthubert); Vijay Devarapalli; nemo@ietf.org >>>> Subject: Re: About the proposed charter and global HAHA (was Re: >>>> [nemo] RE: About route aggregation >>>> in global HAHA >>>> >>>> Hi Alex, >>>> >>>> El 21/04/2006, a las 15:26, Alexandru Petrescu escribi=F3: >>>> >>>>> Pascal Thubert (pthubert) wrote: >>>>>> One of the motivations is to get rid of the issues associated = with >>>>>> the Home Link (single site, single point of failure) >>>>> >>>>> Address "single site - single point of failure"???? >>>>> >>>>> I could understand addressing single HA - single point of failure=20= >>>>> (in >>>>> which case multiple HAs can co-exist on the home link), but = "single >>>>> site >>>>> - single point of failure" can't be addressed by any protocol,=20 >>>>> should >>>>> better be addressed by some administrative organizations. Or by >>>>> shim6. >>>>> >>>> >>>> agree that shim6 could be used to provide some support for this >>>> >>>>> >>>>>> and enable distributing Home and L3. >>>>> >>>>> I am afraid the network is made in such a way that IP addresses=20 >>>>> can't >>>>> be >>>>> distributed. Have you looked at shim6? >>>>> >>>> >>>> same thing >>>> >>>>>> RO comes when we introduce the satellites and route projection, >>>>>> which >>>>>> is another aspect of the draft. >>>>> >>>>> The draft seems to have too many aspects it tries to deal with. >>>>> >>>> >>>> probably you could even do some shim6 based RO also... >>>> >>>> >>>> that is one of the reasons why i am asking about the requirements=20= >>>> and >>>> the use cases that we want to address... >>>> >>>> first we need to define the problem that we want to work on and = then >>>> move to solution space... >>>> >>>> regards, marcelo >>>> >>>> >>>>> Alex >>>>> >>> > From nemo-bounces@ietf.org Tue Apr 25 08:07:48 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYMKU-0007kQ-CO; Tue, 25 Apr 2006 08:07:42 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYMKT-0007kJ-3w for nemo@ietf.org; Tue, 25 Apr 2006 08:07:41 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYJQk-0002IE-I6 for nemo@ietf.org; Tue, 25 Apr 2006 05:01:58 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FYIfa-0005uX-FY for nemo@ietf.org; Tue, 25 Apr 2006 04:13:17 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 25 Apr 2006 10:13:14 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3P8DCUE014542; Tue, 25 Apr 2006 10:13:13 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Tue, 25 Apr 2006 10:13:12 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Tue, 25 Apr 2006 10:12:53 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021A92CB@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Thread-Index: AcZoOFY0mS++4c77SCezhUpkOrUg4wABmHNw From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 25 Apr 2006 08:13:12.0730 (UTC) FILETIME=[18A4FFA0:01C66840] X-Spam-Score: -1.4 (-) X-Scan-Signature: 8fbbaa16f9fd29df280814cb95ae2290 Cc: nemo@ietf.org, Vijay Devarapalli , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Thanks for the explanation Marcelo. Now I see what you have in mind. = This approach is interesting; I guess it takes a mix of MONAMI6 and = SHIM6 to get the result done. IMHO, though, It looks fairly complex and remote, compared to the use of = existing routing protocols proposed in the global HAHA approach. Note that: - the NEMO assumption is to have the MR handle the mobility on behalf of = LFNs which are unaware of the mobility. We want a simple operation for = simple Mobile Network Nodes. - You are not solving the MIP6 Home Link issue, you are bypassing it. = Global HAHA gets rid of the ND operations on the Home Link, therefore it = removes the associated issues, replacing that mechanism by traditional = routing protocol operations. I'd say that before we assemble complex things that we do not have, we = could start thinking of using what we mostly have and have been = discussing for some time already. Pascal >-----Original Message----- >From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >Sent: Tuesday, April 25, 2006 9:16 AM >To: Pascal Thubert (pthubert) >Cc: Alexandru Petrescu; nemo@ietf.org; Vijay Devarapalli >Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] = RE: About route aggregation >in global HAHA > >Hi Pascal, > >El 25/04/2006, a las 10:04, Pascal Thubert (pthubert) escribi=F3: > >> Marcelo, Alexandru: >> >> I fail to understand this. Are you saying that SHIM6 can help resolve >> the case where the MIP6/NEMO Home Link goes down? >> > >when the HA are distributed among different sites, yes, and in a bgp >routing table friendly manner > >i mean, suppose that you have multiple HAs located in different sites >and that each HA has a different PA home prefix assigned. then each >nemo has multiple home prefixes and each MNN has multiple addresses one >per available home prefix > >then you can use the shim6 to move from one home prefix to another, in >case that there is a failure in one of the HAs > >that is exactly what the shim6 protocol has been designed to do > >Again, maybe the global moving networks have some additional >requirements that don't fit in this approach, that is why i have been >asking for a requirement specification, in order to find out which >solution is acceptable > >Regards, marcelo > > >> Pascal >> >>> -----Original Message----- >>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>> Sent: Monday, April 24, 2006 2:41 PM >>> To: Alexandru Petrescu >>> Cc: Pascal Thubert (pthubert); Vijay Devarapalli; nemo@ietf.org >>> Subject: Re: About the proposed charter and global HAHA (was Re: >>> [nemo] RE: About route aggregation >>> in global HAHA >>> >>> Hi Alex, >>> >>> El 21/04/2006, a las 15:26, Alexandru Petrescu escribi=F3: >>> >>>> Pascal Thubert (pthubert) wrote: >>>>> One of the motivations is to get rid of the issues associated with >>>>> the Home Link (single site, single point of failure) >>>> >>>> Address "single site - single point of failure"???? >>>> >>>> I could understand addressing single HA - single point of failure = (in >>>> which case multiple HAs can co-exist on the home link), but "single >>>> site >>>> - single point of failure" can't be addressed by any protocol, = should >>>> better be addressed by some administrative organizations. Or by >>>> shim6. >>>> >>> >>> agree that shim6 could be used to provide some support for this >>> >>>> >>>>> and enable distributing Home and L3. >>>> >>>> I am afraid the network is made in such a way that IP addresses = can't >>>> be >>>> distributed. Have you looked at shim6? >>>> >>> >>> same thing >>> >>>>> RO comes when we introduce the satellites and route projection, >>>>> which >>>>> is another aspect of the draft. >>>> >>>> The draft seems to have too many aspects it tries to deal with. >>>> >>> >>> probably you could even do some shim6 based RO also... >>> >>> >>> that is one of the reasons why i am asking about the requirements = and >>> the use cases that we want to address... >>> >>> first we need to define the problem that we want to work on and then >>> move to solution space... >>> >>> regards, marcelo >>> >>> >>>> Alex >>>> >> From nemo-bounces@ietf.org Tue Apr 25 17:20:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYUxK-0002yi-5A; Tue, 25 Apr 2006 17:20:22 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYUxI-0002yc-MX for nemo@ietf.org; Tue, 25 Apr 2006 17:20:20 -0400 Received: from laposte.rennes.enst-bretagne.fr ([192.44.77.17]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYUxH-0000Zy-9K for nemo@ietf.org; Tue, 25 Apr 2006 17:20:20 -0400 Received: from localhost (localhost.localdomain [127.0.0.1]) by laposte.rennes.enst-bretagne.fr (8.13.4/8.13.4/2004.10.03) with ESMTP id k3PLK87K029687; Tue, 25 Apr 2006 23:20:08 +0200 Received: from givry.rennes.enst-bretagne.fr (givry.rennes.enst-bretagne.fr [192.44.77.29]) by laposte.rennes.enst-bretagne.fr (8.13.4/8.13.4/2004.09.01) with ESMTP id k3PLGUXT029227; Tue, 25 Apr 2006 23:16:30 +0200 Received: from givry.rennes.enst-bretagne.fr (localhost.rennes.enst-bretagne.fr [127.0.0.1]) by givry.rennes.enst-bretagne.fr (8.13.1/8.13.1) with ESMTP id k3PLGU4C002179; Tue, 25 Apr 2006 23:16:30 +0200 (CEST) (envelope-from dupont@givry.rennes.enst-bretagne.fr) Message-Id: <200604252116.k3PLGU4C002179@givry.rennes.enst-bretagne.fr> From: Francis Dupont To: Alexandru Petrescu Subject: Re: [nemo] tunnel all mode In-reply-to: Your message of Mon, 24 Apr 2006 11:48:22 +0200. <444C9EE6.7070006@motorola.com> Date: Tue, 25 Apr 2006 23:16:30 +0200 X-Virus-Scanned: amavisd-new at enst-bretagne.fr X-Spam-Score: 0.0 (/) X-Scan-Signature: 8abaac9e10c826e8252866cbe6766464 Cc: nemo@ietf.org X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org In your previous mail you wrote: Francis Dupont wrote: > There is a discussion about the "tunnel all" mode in the mip6 WG > mailing-list which is interesting for NEMO too. Why is the discussion interesting to NEMO? => because NEMO uses most of the MIPv6 IPsec stuff. Is it more than just the fact that a NEMO implementation sends BU to HA? => no, there is the possibility to protect the whole MR-HA tunnel with IPsec too (perhaps a better option for NEMO than for MIPv6). Is the discussion somehow different if MNP is reckoned? (Mobile Network Prefix) => this can be a part of the discussion too. (as I understand it it revolves around having or not having MIP6 signalling between MN and HA encapsulated. The claimed advantage is easier security with only one SA instead of 4 if it were not encapsulated. The inconvenient is existing implementation don't encapsulate MIP6 signalling.) => yes, I consider the tunnel all mode as a super sharing of the tunnel SAs. The other idea is better privacy. Regards Francis.Dupont@point6.net From nemo-bounces@ietf.org Wed Apr 26 03:07:44 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYe7g-0000Np-OU; Wed, 26 Apr 2006 03:07:40 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYe7f-0000NO-3X for nemo@ietf.org; Wed, 26 Apr 2006 03:07:39 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYe6v-0008T9-07 for nemo@ietf.org; Wed, 26 Apr 2006 03:06:54 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 26 Apr 2006 09:06:53 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3Q76nUG020417; Wed, 26 Apr 2006 09:06:51 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Wed, 26 Apr 2006 09:06:49 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Wed, 26 Apr 2006 09:06:44 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021A9798@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: is global HAHA so complex? Thread-Index: AcZoaX0/biXDvXc4TX24Izxx4/7o7gAlRpEA From: "Pascal Thubert \(pthubert\)" To: "Henrik Levkowetz" X-OriginalArrivalTime: 26 Apr 2006 07:06:49.0553 (UTC) FILETIME=[FCE62C10:01C668FF] X-Spam-Score: 0.0 (/) X-Scan-Signature: 69a74e02bbee44ab4f8eafdbcedd94a1 Cc: marcelo bagnulo braun , nemo@ietf.org, Vijay Devarapalli Subject: [nemo] is global HAHA so complex? X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org > >On 2006-04-25 10:12 Pascal Thubert (pthubert) said the following: >> Thanks for the explanation Marcelo. Now I see what you have in mind. >> This approach is interesting; I guess it takes a mix of MONAMI6 and >> SHIM6 to get the result done. >> >> IMHO, though, It looks fairly complex and remote, compared to the use of >> existing routing protocols proposed in the global HAHA approach. > >Funny you should say that - to me the SHIM6 + MIP6 solution ( I'm not sure >MONAMI6 need to enter into this case) has been quite obvious, while the >HAHA solution has not. I've recently read up a bit more on HAHA, and can >see where you're going with it, but I don't think the major difference is >in complexity, but rather in potential deployment time-frame... [Pascal] I guess that the complexity you perceive is there because global HAHA goes into deployment which bears its own difficulties. I suppose that to make an honest comparison, one should have the equivalent level of information on the SHIM 6 alternate. Doing that effort, you'd find things like:=20 - to deploy a MR, you need to get a prefix from various regions into each Mobile Router. This multiplies the complexity of a deployment. - You need to deploy SHIM6 in all LFN. I agree with you that there is no visible time-frame for getting this. - And that gives you a very limited RO since you still do not have the proxies, which are the real reason why global HAHA improves the path. In terms of protocol, Global HAHA introduces one new component, the proxy HA. This much is not very complex. Then it changes the ND based exchanges in the Home Link by a routing protocol. Again, that complexity is manageable.=20 Pascal From nemo-bounces@ietf.org Wed Apr 26 03:17:37 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYeHH-0000ty-GQ; Wed, 26 Apr 2006 03:17:35 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYeHG-0000ts-Gp for nemo@ietf.org; Wed, 26 Apr 2006 03:17:34 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYeHF-0000ew-Qd for nemo@ietf.org; Wed, 26 Apr 2006 03:17:34 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id C6323212C61; Wed, 26 Apr 2006 10:17:32 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 8DEDD212C59; Wed, 26 Apr 2006 10:17:32 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 53D87BDC40; Wed, 26 Apr 2006 10:17:32 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 1D6B5BDC38; Wed, 26 Apr 2006 10:17:32 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021A9798@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021A9798@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Date: Wed, 26 Apr 2006 10:17:39 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: b280b4db656c3ca28dd62e5e0b03daa8 Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli Subject: [nemo] Re: is global HAHA so complex? X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 26/04/2006, a las 10:06, Pascal Thubert (pthubert) escribi=F3: >> >> On 2006-04-25 10:12 Pascal Thubert (pthubert) said the following: >>> Thanks for the explanation Marcelo. Now I see what you have in mind. >>> This approach is interesting; I guess it takes a mix of MONAMI6 and >>> SHIM6 to get the result done. >>> >>> IMHO, though, It looks fairly complex and remote, compared to the = use > of >>> existing routing protocols proposed in the global HAHA approach. >> >> Funny you should say that - to me the SHIM6 + MIP6 solution ( I'm not > sure >> MONAMI6 need to enter into this case) has been quite obvious, while=20= >> the >> HAHA solution has not. I've recently read up a bit more on HAHA, and > can >> see where you're going with it, but I don't think the major = difference > is >> in complexity, but rather in potential deployment time-frame... > > [Pascal] I guess that the complexity you perceive is there because > global HAHA goes into deployment which bears its own difficulties. I > suppose that to make an honest comparison, one should have the > equivalent level of information on the SHIM 6 alternate. > > Doing that effort, you'd find things like: > - to deploy a MR, you need to get a prefix from various regions into > each Mobile Router. This multiplies the complexity of a deployment. > well, managing that each local ISP injects the gloabla HAHA prefix or=20 deploying a global ISP consortium seems quite complex also > - You need to deploy SHIM6 in all LFN. I agree with you that there is=20= > no > visible time-frame for getting this. right, but you still need this to support other features like=20 multihoming. I mean, a multihomed nemo or a nemo which home network is multihomed=20 will need to have shim6 in the LFN on order to benefit from=20 multihoming, > - And that gives you a very limited RO since you still do not have the > proxies, which are the real reason why global HAHA improves the path. > not sure about this one... i mean you can use the closest HA which is=20 similar to what you achieve in global HAHA (you could also expose CoAs=20= to the LFN, and you would achieve optimal path, but this is other=20 discussion, i guess) > In terms of protocol, Global HAHA introduces one new component, the > proxy HA. This much is not very complex. Then it changes the ND based > exchanges in the Home Link by a routing protocol. Again, that=20 > complexity > is manageable. > In terms of protocol, shim6 does not introduces any new component,=20 since the shim6 protocol is needed to support multihoming Anyway, i still argue that we need to flesh out the requirements that=20 gloablly moving networks have in order to figure out which solution are=20= best suited for them Regards, marcelo > Pascal > From nemo-bounces@ietf.org Wed Apr 26 03:38:59 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYebu-0003Gl-1U; Wed, 26 Apr 2006 03:38:54 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYebs-0003Gg-Nm for nemo@ietf.org; Wed, 26 Apr 2006 03:38:52 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYebp-0001sn-Be for nemo@ietf.org; Wed, 26 Apr 2006 03:38:49 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 26 Apr 2006 09:38:49 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3Q7cmUE028057; Wed, 26 Apr 2006 09:38:48 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Wed, 26 Apr 2006 09:38:48 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Wed, 26 Apr 2006 09:38:47 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021A97C0@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: is global HAHA so complex? Thread-Index: AcZpAYEZinPii5t2RbWvZe7lwrRojAAAE/Gg From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 26 Apr 2006 07:38:48.0065 (UTC) FILETIME=[746BDF10:01C66904] X-Spam-Score: 0.0 (/) X-Scan-Signature: 6cca30437e2d04f45110f2ff8dc1b1d5 Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli Subject: [nemo] RE: is global HAHA so complex? X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >> [Pascal] I guess that the complexity you perceive is there because >> global HAHA goes into deployment which bears its own difficulties. I >> suppose that to make an honest comparison, one should have the >> equivalent level of information on the SHIM 6 alternate. >> >> Doing that effort, you'd find things like: >> - to deploy a MR, you need to get a prefix from various regions into >> each Mobile Router. This multiplies the complexity of a deployment. >> > >well, managing that each local ISP injects the gloabla HAHA prefix or >deploying a global ISP consortium seems quite complex also > >> - You need to deploy SHIM6 in all LFN. I agree with you that there is >> no >> visible time-frame for getting this. > >right, but you still need this to support other features like >multihoming. >I mean, a multihomed nemo or a nemo which home network is multihomed >will need to have shim6 in the LFN on order to benefit from >multihoming, [Pascal] I fail to see why a telephone needs to be multihomed?=20 And there are other things you can do than SHIM6 to benefit from multihoming. For instance you can use MIP/NEMO to get back to ISP-B via ISP-A. While this is hardly a generic multihoming solution, this is acceptable when you are already in the MIP/NEMO context. > >> - And that gives you a very limited RO since you still do not have the >> proxies, which are the real reason why global HAHA improves the path. >> > >not sure about this one... i mean you can use the closest HA which is >similar to what you achieve in global HAHA (you could also expose CoAs >to the LFN, and you would achieve optimal path, but this is other >discussion, i guess) [Pascal] How many prefixes do you have in your MR in the SHIM6 approach? If you do not get rid of the home link, then that is as many anchor points for MR to reattach. Not even one per continent. I do not call that RO. This is why global HAHA removes the geographical constraint of the home link. > >> In terms of protocol, Global HAHA introduces one new component, the >> proxy HA. This much is not very complex. Then it changes the ND based >> exchanges in the Home Link by a routing protocol. Again, that >> complexity >> is manageable. >> > >In terms of protocol, shim6 does not introduces any new component, >since the shim6 protocol is needed to support multihoming > [Pascal] The proxy is added to be very distributed - sort of collocated with SGSNs type of box. This is the component that gives RO to global HAHA. Now you see why HAHA is built that way. >Anyway, i still argue that we need to flesh out the requirements that >gloablly moving networks have in order to figure out which solution are >best suited for them > [Pascal] Well, you were not there all the time and I personally do welcome you now.=20 But there is a long long history in NEMO about requirements (ask Thierry). After that, we went into a long and difficult work on Problem Statement (RO-PS). And since people had worked on RO for years, we also compiled the RO SA, solution space analysis.=20 Global HAHA builds on all that and sorry, I do not wish to start over... Pascal From nemo-bounces@ietf.org Wed Apr 26 03:56:47 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYetB-0000Ye-BB; Wed, 26 Apr 2006 03:56:45 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYetA-0000YZ-EL for nemo@ietf.org; Wed, 26 Apr 2006 03:56:44 -0400 Received: from stsc1260-eth-s1-s1p1-vip.va.neustar.com ([156.154.16.129] helo=chiedprmail1.ietf.org) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYetA-00034h-Bo for nemo@ietf.org; Wed, 26 Apr 2006 03:56:44 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by chiedprmail1.ietf.org with esmtp (Exim 4.43) id 1FYet8-0002Tn-Vp for nemo@ietf.org; Wed, 26 Apr 2006 03:56:44 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 8FB95212C8E; Wed, 26 Apr 2006 10:56:40 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 5C3C6212C8D; Wed, 26 Apr 2006 10:56:40 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id E7A24BDC40; Wed, 26 Apr 2006 10:56:39 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id AF3DABDC38; Wed, 26 Apr 2006 10:56:39 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021A97C0@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021A97C0@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Date: Wed, 26 Apr 2006 10:56:46 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: -2.6 (--) X-Scan-Signature: 7da5a831c477fb6ef97f379a05fb683c Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli Subject: [nemo] Re: is global HAHA so complex? X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 26/04/2006, a las 10:38, Pascal Thubert (pthubert) escribi=F3: > >>> [Pascal] I guess that the complexity you perceive is there because >>> global HAHA goes into deployment which bears its own difficulties. I >>> suppose that to make an honest comparison, one should have the >>> equivalent level of information on the SHIM 6 alternate. >>> >>> Doing that effort, you'd find things like: >>> - to deploy a MR, you need to get a prefix from various regions into >>> each Mobile Router. This multiplies the complexity of a deployment. >>> >> >> well, managing that each local ISP injects the gloabla HAHA prefix or >> deploying a global ISP consortium seems quite complex also >> >>> - You need to deploy SHIM6 in all LFN. I agree with you that there = is >>> no >>> visible time-frame for getting this. >> >> right, but you still need this to support other features like >> multihoming. >> I mean, a multihomed nemo or a nemo which home network is multihomed >> will need to have shim6 in the LFN on order to benefit from >> multihoming, > > [Pascal] I fail to see why a telephone needs to be multihomed? > 4G telephones are multihomed by definition. besides, the end device=20 itself does need to be multihomed in order to need ths shim6 support,=20 just attaching it to a multihomed network will imply multiple HoAs and=20= you need something like SHIM6 to support this... > And there are other things you can do than SHIM6 to benefit from > multihoming. For instance you can use MIP/NEMO to get back to ISP-B = via > ISP-A. While this is hardly a generic multihoming solution, this is > acceptable when you are already in the MIP/NEMO context. > no, this doesn't work multihoming is about fault tolerance and mip relies in going through=20 the HA, (i guess you know this already) So mip and nemo is not enough... so need something else for this, shim6=20= or global haha for instance :-) >> >>> - And that gives you a very limited RO since you still do not have > the >>> proxies, which are the real reason why global HAHA improves the = path. >>> >> >> not sure about this one... i mean you can use the closest HA which is >> similar to what you achieve in global HAHA (you could also expose = CoAs >> to the LFN, and you would achieve optimal path, but this is other >> discussion, i guess) > > [Pascal] How many prefixes do you have in your MR in the SHIM6=20 > approach? shim6 relies on PA addressing, so you need one prefix per ISP serving=20 the home netowrk. This is so becasue shim6 is gloabl routing table friendly and it does=20 not requires additional bgp entries > If you do not get rid of the home link, then that is as many anchor > points for MR to reattach. Not even one per continent. I do not call > that RO. This is why global HAHA removes the geographical constraint = of > the home link. > not sure what your point is here... >> >>> In terms of protocol, Global HAHA introduces one new component, the >>> proxy HA. This much is not very complex. Then it changes the ND = based >>> exchanges in the Home Link by a routing protocol. Again, that >>> complexity >>> is manageable. >>> >> >> In terms of protocol, shim6 does not introduces any new component, >> since the shim6 protocol is needed to support multihoming >> > [Pascal] The proxy is added to be very distributed - sort of = collocated > with SGSNs type of box. This is the component that gives RO to global > HAHA. Now you see why HAHA is built that way. > >> Anyway, i still argue that we need to flesh out the requirements that >> gloablly moving networks have in order to figure out which solution=20= >> are >> best suited for them >> > [Pascal] Well, you were not there all the time and I personally do > welcome you now. > But there is a long long history in NEMO about requirements (ask > Thierry). After that, we went into a long and difficult work on = Problem > Statement (RO-PS). And since people had worked on RO for years, we = also > compiled the RO SA, solution space analysis. > > Global HAHA builds on all that and sorry, I do not wish to start=20 > over... > I am sorry but we are discussing rechartering at this very moment My point is that if we are going to work on a solution we need to agree=20= on the requiremetns for that solution and then define the solution. I=20 don't we are ready to accept a full fleshed solution like global haha=20 that implies at least a liability to the gloabl routing health without=20= proper assesment of the requirements that justify the need for that=20 specific solution and that other alternatives need to be ruled out. Regards, marcelo > Pascal > From nemo-bounces@ietf.org Wed Apr 26 06:32:42 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYhJZ-0000NB-M4; Wed, 26 Apr 2006 06:32:09 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYhJY-0000N6-I9 for nemo@ietf.org; Wed, 26 Apr 2006 06:32:08 -0400 Received: from [129.188.136.9] (helo=ftpbox.mot.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYhJX-000345-AN for nemo@ietf.org; Wed, 26 Apr 2006 06:32:08 -0400 Received: from az33exr02.mot.com (az33exr02.mot.com [10.64.251.232]) by ftpbox.mot.com (8.12.11/Motgate7) with ESMTP id k3QAbWb9020400; Wed, 26 Apr 2006 05:37:32 -0500 (CDT) Received: from zfr01srv02.crm.mot.com (zfr01srv02.crm.mot.com [10.161.201.8]) by az33exr02.mot.com (8.13.1/8.13.0) with ESMTP id k3QAjJO3004536; Wed, 26 Apr 2006 05:45:20 -0500 (CDT) Received: from [10.161.201.117] (zfr01-2117.crm.mot.com [10.161.201.117]) by zfr01srv02.crm.mot.com (Postfix) with ESMTP id 32F68865980; Wed, 26 Apr 2006 12:31:52 +0200 (CEST) Message-ID: <444F4C17.4030700@motorola.com> Date: Wed, 26 Apr 2006 12:31:51 +0200 From: Alexandru Petrescu User-Agent: Thunderbird 1.5 (Windows/20051201) MIME-Version: 1.0 To: marcelo bagnulo braun Subject: Re: phones with multiple interfaces (was: [nemo] Re: is global HAHA so complex?) References: <7892795E1A87F04CADFCCF41FADD00FC021A97C0@xmb-ams-337.emea.cisco.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed X-Brightmail-Tracker: AAAAAQAAAAQ= X-White-List-Member: TRUE Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by ftpbox.mot.com id k3QAbWb9020400 X-Spam-Score: 0.0 (/) X-Scan-Signature: 7baded97d9887f7a0c7e8a33c2e3ea1b Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" , Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org marcelo bagnulo braun wrote: >=20 > El 26/04/2006, a las 10:38, Pascal Thubert (pthubert) escribi=F3: >=20 >>=20 >>>> [Pascal] I guess that the complexity you perceive is there >>>> because global HAHA goes into deployment which bears its own >>>> difficulties. I suppose that to make an honest comparison, one >>>> should have the equivalent level of information on the SHIM 6 >>>> alternate. >>>>=20 >>>> Doing that effort, you'd find things like: - to deploy a MR, >>>> you need to get a prefix from various regions into each Mobile >>>> Router. This multiplies the complexity of a deployment. >>>>=20 >>>=20 >>> well, managing that each local ISP injects the gloabla HAHA >>> prefix or deploying a global ISP consortium seems quite complex >>> also >>>=20 >>>> - You need to deploy SHIM6 in all LFN. I agree with you that >>>> there is no visible time-frame for getting this. >>>=20 >>> right, but you still need this to support other features like=20 >>> multihoming. I mean, a multihomed nemo or a nemo which home >>> network is multihomed will need to have shim6 in the LFN on order >>> to benefit from multihoming, >>=20 >> [Pascal] I fail to see why a telephone needs to be multihomed? >>=20 >=20 > 4G telephones are multihomed by definition. Not only 4G, there exist at least two mass-marketed high-end 2.5G handsets with triple interfaces: GPRS, bluetooth, wifi. I.e. they have multiple interfaces. But finding what are their needs to run them simultaneously, or routing between them is still a matter of user requirements. Alex From nemo-bounces@ietf.org Wed Apr 26 11:05:30 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYla3-0005Mw-DY; Wed, 26 Apr 2006 11:05:27 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYla1-0005Mr-Jq for nemo@ietf.org; Wed, 26 Apr 2006 11:05:25 -0400 Received: from smtp03.uc3m.es ([163.117.136.123]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYla0-00016y-6T for nemo@ietf.org; Wed, 26 Apr 2006 11:05:25 -0400 Received: from smtp03.uc3m.es (localhost [127.0.0.1]) by localhost.uc3m.es (Postfix) with ESMTP id E376F90174; Wed, 26 Apr 2006 17:05:22 +0200 (CEST) Received: from [163.117.203.135] (unknown [163.117.203.135]) by smtp03.uc3m.es (Postfix) with ESMTP id 2EFED90013; Wed, 26 Apr 2006 17:05:22 +0200 (CEST) In-Reply-To: <444F851B.90407@ericsson.com> References: <7892795E1A87F04CADFCCF41FADD00FC021A9798@xmb-ams-337.emea.cisco.com> <444F851B.90407@ericsson.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <3ca32b78fed9382670bc817896445577@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Date: Wed, 26 Apr 2006 18:05:29 +0300 To: Henrik Levkowetz X-Mailer: Apple Mail (2.623) X-Spam-Score: 0.0 (/) X-Scan-Signature: 50a516d93fd399dc60588708fd9a3002 Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" , Vijay Devarapalli Subject: [nemo] Re: is global HAHA so complex? X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 26/04/2006, a las 17:35, Henrik Levkowetz escribi=F3: > > > On 2006-04-26 09:17 marcelo bagnulo braun said the following: > > > >> Anyway, i still argue that we need to flesh out the requirements that >> gloablly moving networks have in order to figure out which solution=20= >> are >> best suited for them > > Yes. That's it exactly. If we don't have a clear picture of the=20 > potential > users, after all this long thread, i think that the case of globally moving=20 networks on board of airplanes or other transportation mechanisms are a=20= use case for which there is interest and that we could try to work on a=20= solution for that. > use cases and requirements we can't evaluate the benefit of any > particular solution... > so imho we would need a charter item to flesh out the requirement for=20 this case in particular and potentially other charter items for other=20 use cases that we identify in this discussion (but no other use cases=20 have been brought up to the list so far during this charter discussion=20= AFAICT Then another charter item would be the design of a solution that suits=20= the identified requirements regards, marcelo > > Henrik > From nemo-bounces@ietf.org Wed Apr 26 13:21:43 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYnhm-00024W-8P; Wed, 26 Apr 2006 13:21:34 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FYnhk-00024R-Ri for nemo@ietf.org; Wed, 26 Apr 2006 13:21:32 -0400 Received: from mail1.azairenet.com ([66.92.223.4] helo=bart.corp.azairenet.com) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FYnhj-0000IN-Hk for nemo@ietf.org; Wed, 26 Apr 2006 13:21:32 -0400 Received: from [10.1.201.8] ([10.1.201.8]) by bart.corp.azairenet.com with Microsoft SMTPSVC(6.0.3790.1830); Wed, 26 Apr 2006 10:21:26 -0700 Message-ID: <444FAC15.1070103@azairenet.com> Date: Wed, 26 Apr 2006 10:21:25 -0700 From: Vijay Devarapalli User-Agent: Thunderbird 1.5.0.2 (Windows/20060308) MIME-Version: 1.0 To: Alexandru Petrescu References: <7892795E1A87F04CADFCCF41FADD00FC021A97C0@xmb-ams-337.emea.cisco.com> <444F4C17.4030700@motorola.com> In-Reply-To: <444F4C17.4030700@motorola.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 8bit X-OriginalArrivalTime: 26 Apr 2006 17:21:26.0106 (UTC) FILETIME=[D9099BA0:01C66955] X-Spam-Score: 0.0 (/) X-Scan-Signature: c1c65599517f9ac32519d043c37c5336 Cc: marcelo bagnulo braun , nemo@ietf.org, "Pascal Thubert \(pthubert\)" , Henrik Levkowetz Subject: [nemo] Re: phones with multiple interfaces X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org now I think there is confusion between site multihoming and host multihoming. Vijay Alexandru Petrescu wrote: > marcelo bagnulo braun wrote: >> >> El 26/04/2006, a las 10:38, Pascal Thubert (pthubert) escribió: >> >>> >>>>> [Pascal] I guess that the complexity you perceive is there >>>>> because global HAHA goes into deployment which bears its own >>>>> difficulties. I suppose that to make an honest comparison, one >>>>> should have the equivalent level of information on the SHIM 6 >>>>> alternate. >>>>> >>>>> Doing that effort, you'd find things like: - to deploy a MR, >>>>> you need to get a prefix from various regions into each Mobile >>>>> Router. This multiplies the complexity of a deployment. >>>>> >>>> >>>> well, managing that each local ISP injects the gloabla HAHA >>>> prefix or deploying a global ISP consortium seems quite complex >>>> also >>>> >>>>> - You need to deploy SHIM6 in all LFN. I agree with you that >>>>> there is no visible time-frame for getting this. >>>> >>>> right, but you still need this to support other features like >>>> multihoming. I mean, a multihomed nemo or a nemo which home >>>> network is multihomed will need to have shim6 in the LFN on order >>>> to benefit from multihoming, >>> >>> [Pascal] I fail to see why a telephone needs to be multihomed? >>> >> >> 4G telephones are multihomed by definition. > > Not only 4G, there exist at least two mass-marketed high-end 2.5G > handsets with triple interfaces: GPRS, bluetooth, wifi. I.e. they have > multiple interfaces. But finding what are their needs to run them > simultaneously, or routing between them is still a matter of user > requirements. > > Alex From nemo-bounces@ietf.org Thu Apr 27 03:43:30 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ18v-0007Oq-Hn; Thu, 27 Apr 2006 03:42:29 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ18u-0007Ol-EG for nemo@ietf.org; Thu, 27 Apr 2006 03:42:28 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZ18r-0001OF-Pc for nemo@ietf.org; Thu, 27 Apr 2006 03:42:28 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 27 Apr 2006 09:42:25 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3R7gMUE005856; Thu, 27 Apr 2006 09:42:24 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Thu, 27 Apr 2006 09:42:21 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Re: is global HAHA so complex? Date: Thu, 27 Apr 2006 09:42:07 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021A9C90@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Re: is global HAHA so complex? Thread-Index: AcZpBwM+Vgj9sXntQ/G8a2GeQX7z7QAw/CUw From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 27 Apr 2006 07:42:21.0910 (UTC) FILETIME=[1E4BC760:01C669CE] X-Spam-Score: 0.0 (/) X-Scan-Signature: 2086112c730e13d5955355df27e3074b Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >> [Pascal] I fail to see why a telephone needs to be multihomed? >> > >4G telephones are multihomed by definition. besides, the end device [Pascal]=20 [Pascal] I am afraid there is some confusion here. I think we were talking about multiple Home Networks not multiple interfaces on the phone. In that context, could you please elaborate why 4G phones are multi-homed by definition?=20 >itself does need to be multihomed in order to need ths shim6 support, >just attaching it to a multihomed network will imply multiple HoAs and >you need something like SHIM6 to support this... [Pascal] And if a Home Link fails what happens to the connections from these non SHIM6 end devices? > >> And there are other things you can do than SHIM6 to benefit from >> multihoming. For instance you can use MIP/NEMO to get back to ISP-B via >> ISP-A. While this is hardly a generic multihoming solution, this is >> acceptable when you are already in the MIP/NEMO context. >> > >no, this doesn't work >multihoming is about fault tolerance and mip relies in going through >the HA, (i guess you know this already) [Pascal]=20 [Pascal] Again we are off sync. I meant NEMO can be used similarly to a good old ISDN call when your direct link to your ISP dies. But, agreed, nothing can save you if the home link craches.=20 > >So mip and nemo is not enough... so need something else for this, shim6 >or global haha for instance :-) [Pascal] Back in sync.=20 Though I need to read your draft because I do not have a 100 percent clear view on your proposal.=20 I'm not convinced by the answers on my issues (which RO do you get, how it works with stupid LFN, complexity of having multiple Home prefix on each phone). > >>> >>>> - And that gives you a very limited RO since you still do not have >> the >>>> proxies, which are the real reason why global HAHA improves the path. >>>> >>> >>> not sure about this one... i mean you can use the closest HA which is >>> similar to what you achieve in global HAHA (you could also expose CoAs >>> to the LFN, and you would achieve optimal path, but this is other >>> discussion, i guess) >> >> [Pascal] How many prefixes do you have in your MR in the SHIM6 >> approach? > >shim6 relies on PA addressing, so you need one prefix per ISP serving >the home netowrk. [Pascal] That was not my question. The RO ypu get with this depends on how close from your location the closest HA is. The SHIM approach needs one prefix in the MR per such location. So you end up with 1-3 locations, and that does not give you Route Optimization. Distributing the same Home does better, and multiplying proxies does even better. That was my point. >This is so becasue shim6 is gloabl routing table friendly and it does >not requires additional bgp entries [Pascal] There a 170K entries in the V4 tables. If NEMO RO is for all Phones and cars and trucks and planes is obtained at the expense of 1-100 more entries, no one will complain.=20 > >> If you do not get rid of the home link, then that is as many anchor >> points for MR to reattach. Not even one per continent. I do not call >> that RO. This is why global HAHA removes the geographical constraint of >> the home link. >> > >not sure what your point is here... [Pascal] See just above. SHIM does not give you any RO. Proxies do. =20 > >> But there is a long long history in NEMO about requirements (ask >> Thierry). After that, we went into a long and difficult work on Problem >> Statement (RO-PS). And since people had worked on RO for years, we also >> compiled the RO SA, solution space analysis. >> >> Global HAHA builds on all that and sorry, I do not wish to start >> over... >> > >I am sorry but we are discussing rechartering at this very moment >My point is that if we are going to work on a solution we need to agree >on the requiremetns for that solution and then define the solution. I >don't we are ready to accept a full fleshed solution like global haha >that implies at least a liability to the gloabl routing health without >proper assesment of the requirements that justify the need for that >specific solution and that other alternatives need to be ruled out. [Pascal]=20 [Pascal] I understand that and would not oppose in general; that'd be the normal way for a new WG.=20 But we are not chartering and starting from scratch, we are re-chartering to cover an evolution. What I am saying is that the requirements are pretty much the same as when NEMO was created. We restrained ourselves to do basic as a protocol while doing only PS for RO. If you think we need to evolve the requirements as well then you need to post to Thierry. The feeling there was that the charter needed polishing but that the requirements held still. Pascal From nemo-bounces@ietf.org Thu Apr 27 03:51:54 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ1I1-0001I6-UL; Thu, 27 Apr 2006 03:51:53 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ1I0-0001I1-UD for nemo@ietf.org; Thu, 27 Apr 2006 03:51:52 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZ1Hz-0001i5-K0 for nemo@ietf.org; Thu, 27 Apr 2006 03:51:52 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 27 Apr 2006 09:51:52 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3R7poUE008422; Thu, 27 Apr 2006 09:51:50 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Thu, 27 Apr 2006 09:51:50 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Re: is global HAHA so complex? Date: Thu, 27 Apr 2006 09:51:44 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021A9C9E@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Re: is global HAHA so complex? Thread-Index: AcZpQua5JLPU1GepS5KVhTyE1VvzyAAi8xmA From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" , "Henrik Levkowetz" X-OriginalArrivalTime: 27 Apr 2006 07:51:50.0464 (UTC) FILETIME=[712E3C00:01C669CF] X-Spam-Score: 0.0 (/) X-Scan-Signature: cab78e1e39c4b328567edb48482b6a69 Cc: nemo@ietf.org, Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >> >> >>> Anyway, i still argue that we need to flesh out the requirements that >>> gloablly moving networks have in order to figure out which solution >>> are >>> best suited for them >> >> Yes. That's it exactly. If we don't have a clear picture of the >> potential >> users, > >after all this long thread, i think that the case of globally moving >networks on board of airplanes or other transportation mechanisms are a >use case for which there is interest and that we could try to work on a >solution for that. > [Pascal] Yes, this is already in the NEMO requirements :)=20 " # Networks of sensors and computers deployed in vehicles: vehicles are increasingly embedded with a number of processing units for safety and ease of driving reasons, as advocated by ITS (Intelligent Transportation Systems) applications ([9], CALM - Medium and Long Range, High Speed, Air Interfaces parameters and protocols for broadcast, point to point, vehicle to vehicle, and vehicle to point communication in the ITS sector - Networking Protocol - Complementary Element, February 2005., [8]Ernst, T. and K. Uehara, Connecting Automobiles to the Internet, November 2002.). # Access networks deployed in public transportation (buses, trains, taxis, aircrafts): they provide Internet access to IP devices carried by passengers: laptop, camera, mobile phone: host mobility within network mobility or PANs: network mobility within network mobility, i.e. nested mobility (see [4]Ernst, T. and H. Lach, Network Mobility Support Terminology, February 2005. for the definition of nested mobility). " But we can move forward and detail more if you wish. Note that for devices like cars, they expect to rely on cheap common devices such as 4G phones with some application capability built in (like JAVA or .net). As a result, you can not do most of the vehicules if you do not include 4G phones... Pascal From nemo-bounces@ietf.org Thu Apr 27 05:55:48 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ3Dr-0007bX-ML; Thu, 27 Apr 2006 05:55:43 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ3Dq-0007bN-QY for nemo@ietf.org; Thu, 27 Apr 2006 05:55:42 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZ3Dq-0001LS-3C for nemo@ietf.org; Thu, 27 Apr 2006 05:55:42 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id ED8FC212C8E; Thu, 27 Apr 2006 11:32:29 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 851A1212C8D; Thu, 27 Apr 2006 11:32:29 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 1A58FBDC40; Thu, 27 Apr 2006 11:32:29 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id D8C5DBDC38; Thu, 27 Apr 2006 11:32:28 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021A9C90@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021A9C90@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <66629944f58444fa3bdeaa9a7b777f50@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: [nemo] Re: is global HAHA so complex? Date: Thu, 27 Apr 2006 11:32:36 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: 16c9da4896bf5539ae3547c6c25f06a0 Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 27/04/2006, a las 10:42, Pascal Thubert (pthubert) escribi=F3: > > >>> [Pascal] I fail to see why a telephone needs to be multihomed? >>> >> >> 4G telephones are multihomed by definition. besides, the end device > [Pascal] > [Pascal] I am afraid there is some confusion here. I think we were > talking about multiple Home Networks not multiple interfaces on the > phone. In that context, could you please elaborate why 4G phones are > multi-homed by definition? > Ok, you mean why a telephone may need multiple Home Networks? Well, i guess that if you have a 4G phone and you have multiple=20 provider, you want to be able to be reachable through all the=20 providers. We do agree that is a desired feature to be able to have differetn=20 providers for different technologies, right? So that you don't have to=20= rely on a single provider giving UMTS, WIFI services for instance. In this scenario, you probably also want to have a stable address from=20= each of the providers, so that you can by pass any of the providers. I=20= mean if you are going to have a communication through the WiFI=20 interface, you don't want to have to exchange the initial packets and=20 the HoTI packets through the UMTS interface, right? So i really think that it is a very useful feature that 4G phones hame=20= multiple Home Addresses associated to different providers and different=20= technologies makes sense to you? >> itself does need to be multihomed in order to need ths shim6 support, >> just attaching it to a multihomed network will imply multiple HoAs = and >> you need something like SHIM6 to support this... > [Pascal] And if a Home Link fails what happens to the connections from > these non SHIM6 end devices? > shim6 requires support from both peers of the communication, that is=20 for sure. but shim6 is used to provide multihoming support for fixed scenarios,=20 so there other reasons that just RO to implement and support the shim6=20= (I do agree that the deployment time frames for shim6 and a global HAHA=20= solution are different and that shim6 requires modification of more=20 devices, but as i already mention, i still not convinced that the=20 timing required for this solution is so tight, i mean i haven't heard=20 that airlines will provide v6 support anytime soon) >> >>> And there are other things you can do than SHIM6 to benefit from >>> multihoming. For instance you can use MIP/NEMO to get back to ISP-B > via >>> ISP-A. While this is hardly a generic multihoming solution, this is >>> acceptable when you are already in the MIP/NEMO context. >>> >> >> no, this doesn't work >> multihoming is about fault tolerance and mip relies in going through >> the HA, (i guess you know this already) > [Pascal] > [Pascal] Again we are off sync. I meant NEMO can be used similarly to = a > good old ISDN call when your direct link to your ISP dies. But, = agreed, > nothing can save you if the home link craches. > ok, we agree then that mip/nemo approach cannot be used to provide home=20= network fault tolerance >> >> So mip and nemo is not enough... so need something else for this,=20 >> shim6 >> or global haha for instance :-) > > [Pascal] Back in sync. > Though I need to read your draft because I do not have a 100 percent > clear view on your proposal. > I'm not convinced by the answers on my issues (which RO do you get, = how > it works with stupid LFN, complexity of having multiple Home prefix on > each phone). > >> >>>> >>>>> - And that gives you a very limited RO since you still do not have >>> the >>>>> proxies, which are the real reason why global HAHA improves the > path. >>>>> >>>> >>>> not sure about this one... i mean you can use the closest HA which > is >>>> similar to what you achieve in global HAHA (you could also expose > CoAs >>>> to the LFN, and you would achieve optimal path, but this is other >>>> discussion, i guess) >>> >>> [Pascal] How many prefixes do you have in your MR in the SHIM6 >>> approach? >> >> shim6 relies on PA addressing, so you need one prefix per ISP serving >> the home netowrk. > [Pascal] That was not my question. The RO ypu get with this depends on > how close from your location the closest HA is. The SHIM approach = needs > one prefix in the MR per such location. So you end up with 1-3 > locations, and that does not give you Route Optimization. Distributing > the same Home does better, and multiplying proxies does even better. > That was my point. well with a shim6 approach you would get as many prefixes as=20 distributed locations and you would be able to use the prefix=20 associated with the closer location. So you end up suing the HA which=20 is closer to the MR. That should provide the same level of RO than=20 global HAHA... > > >> This is so becasue shim6 is gloabl routing table friendly and it does >> not requires additional bgp entries > [Pascal] There a 170K entries in the V4 tables. If NEMO RO is for all > Phones and cars and trucks and planes is obtained at the expense of > 1-100 more entries, no one will complain. > why do these guys get to pollute the global routing table and others=20 don't? being a small number of entries is not enough justification what need to be explained is why other solutions that do not pollute=20 the global routing table fail to meet the requirements that this=20 problem has. hence the need for explicitly flesh out the requirements >> I am sorry but we are discussing rechartering at this very moment >> My point is that if we are going to work on a solution we need to=20 >> agree >> on the requiremetns for that solution and then define the solution. I >> don't we are ready to accept a full fleshed solution like global haha >> that implies at least a liability to the gloabl routing health = without >> proper assesment of the requirements that justify the need for that >> specific solution and that other alternatives need to be ruled out. > [Pascal] > [Pascal] I understand that and would not oppose in general; that'd be > the normal way for a new WG. > > But we are not chartering and starting from scratch, we are > re-chartering to cover an evolution. What I am saying is that the > requirements are pretty much the same as when NEMO was created. so what is the document describing the requirements of gloablly moving=20= networks? I mean what is the document that define the requirements for=20= the problem that global HAHA is solving and that cannot be solved with=20= a solution that does not pollute the global routing table? I mean, global HAHA is not a general RO mechanism, as i been told, so=20 what is the requirements for the specific problem that is being solved=20= by global HAHA? > We > restrained ourselves to do basic as a protocol while doing only PS for > RO. > > If you think we need to evolve the requirements as well then you need=20= > to > post to Thierry. The feeling there was that the charter needed=20 > polishing > but that the requirements held still. > the requirements for what? for RO? for nemo? i guess nemo is addressed in the basic spec and global HAHA is not a=20 general RO mechanism, so i fail to see where are the requirements for=20 the problem that global HAHA is solving rgeards, marcelo > Pascal > From nemo-bounces@ietf.org Thu Apr 27 06:21:57 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ3dD-0000HL-74; Thu, 27 Apr 2006 06:21:55 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZ3dB-0000HF-Vc for nemo@ietf.org; Thu, 27 Apr 2006 06:21:53 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZ3dA-0002bW-G0 for nemo@ietf.org; Thu, 27 Apr 2006 06:21:53 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 8B1E2212C8E; Thu, 27 Apr 2006 13:21:51 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id 593ED212C8D; Thu, 27 Apr 2006 13:21:51 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 24790BDC40; Thu, 27 Apr 2006 13:21:51 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id E2FF2BDC38; Thu, 27 Apr 2006 13:21:50 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021A9C9E@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021A9C9E@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <2bba36c41512bb8b5862650c81d35693@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: [nemo] Re: is global HAHA so complex? Date: Thu, 27 Apr 2006 13:21:58 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: fb6060cb60c0cea16e3f7219e40a0a81 Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 27/04/2006, a las 10:51, Pascal Thubert (pthubert) escribi=F3: >>> >>> >>>> Anyway, i still argue that we need to flesh out the requirements > that >>>> gloablly moving networks have in order to figure out which solution >>>> are >>>> best suited for them >>> >>> Yes. That's it exactly. If we don't have a clear picture of the >>> potential >>> users, >> >> after all this long thread, i think that the case of globally moving >> networks on board of airplanes or other transportation mechanisms are=20= >> a >> use case for which there is interest and that we could try to work on=20= >> a >> solution for that. >> > [Pascal] Yes, this is already in the NEMO requirements :) > > " # Networks of sensors and computers deployed in vehicles: vehicles > are increasingly embedded with a number of processing units for safety > and ease of driving reasons, as advocated by ITS (Intelligent > Transportation Systems) applications ([9], CALM - Medium and Long=20 > Range, > High Speed, Air Interfaces parameters and protocols for broadcast,=20 > point > to point, vehicle to vehicle, and vehicle to point communication in = the > ITS sector - Networking Protocol - Complementary Element, February > 2005., [8]Ernst, T. and K. Uehara, Connecting Automobiles to the > Internet, November 2002.). > > # Access networks deployed in public transportation (buses, trains, > taxis, aircrafts): they provide Internet access to IP devices carried=20= > by > passengers: laptop, camera, mobile phone: host mobility within network > mobility or PANs: network mobility within network mobility, i.e. = nested > mobility (see [4]Ernst, T. and H. Lach, Network Mobility Support > Terminology, February 2005. for the definition of nested mobility). > " > right we knew since the begining that the networks in planes and cars=20 are a clear target for nemo protocol. The question is what are the specific requiremetns of globally moving=20 networks located in planes that they need a specific RO mechanism like=20= gloabl HAHA that cannot be addressed by a general RO mechanism for=20 nemo? Regards, marcelo > But we can move forward and detail more if you wish. > > Note that for devices like cars, they expect to rely on cheap common > devices such as 4G phones with some application capability built in > (like JAVA or .net). As a result, you can not do most of the vehicules > if you do not include 4G phones... > > Pascal > From nemo-bounces@ietf.org Fri Apr 28 02:44:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZMiC-0007fV-8Z; Fri, 28 Apr 2006 02:44:20 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZMiB-0007fP-8f for nemo@ietf.org; Fri, 28 Apr 2006 02:44:19 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZMi7-0003By-Ta for nemo@ietf.org; Fri, 28 Apr 2006 02:44:19 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 28 Apr 2006 08:44:15 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3S6iEUE019312; Fri, 28 Apr 2006 08:44:14 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Fri, 28 Apr 2006 08:44:14 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Date: Fri, 28 Apr 2006 08:43:58 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021F15BB@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: What RO do you get Thread-Index: AcZp1VShVFz9FtxWSgCB8b7xZGKKzgAuMQoA From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 28 Apr 2006 06:44:14.0452 (UTC) FILETIME=[2A05AB40:01C66A8F] X-Spam-Score: 0.0 (/) X-Scan-Signature: 7baded97d9887f7a0c7e8a33c2e3ea1b Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli Subject: [nemo] What RO do you get X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >>>> [Pascal] How many prefixes do you have in your MR in the SHIM6 >>>> approach? >>> >>> shim6 relies on PA addressing, so you need one prefix per ISP serving >>> the home netowrk. >> [Pascal] That was not my question. The RO ypu get with this depends on >> how close from your location the closest HA is. The SHIM approach needs >> one prefix in the MR per such location. So you end up with 1-3 >> locations, and that does not give you Route Optimization. Distributing >> the same Home does better, and multiplying proxies does even better. >> That was my point. > >well with a shim6 approach you would get as many prefixes as >distributed locations and you would be able to use the prefix >associated with the closer location. So you end up suing the HA which >is closer to the MR. That should provide the same level of RO than >global HAHA... > [Pascal] No, by orders of magnitude.=20 Global HAHA allows you to place proxies wherever you place SGSNs todays, and HAs wherever you place GGSNs. You'll find them in your city. So the wasted part of the path (the non ROed) is a few blocks. If I have to have one prefix by HA and you have no proxy, then the number of location is limited to the number of prefix. How many prefixes would you manage in a telephone? As a result, the non ROed part (between you and the closest HA) is half a continent, though you might be talking to someone next door (Alex's point in this discussion). I see where the SHIM6 approach provides connectivity redundancy (though not transparent to LFN as should be), but no RO at all, sorry.=20 Pascal From nemo-bounces@ietf.org Fri Apr 28 03:01:38 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZMyr-0000y1-II; Fri, 28 Apr 2006 03:01:33 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZMyp-0000xD-FQ for nemo@ietf.org; Fri, 28 Apr 2006 03:01:31 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZMsJ-0003hH-IQ for nemo@ietf.org; Fri, 28 Apr 2006 02:54:50 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 28 Apr 2006 08:54:44 +0200 Received: from xbh-ams-331.emea.cisco.com (xbh-ams-331.cisco.com [144.254.231.71]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3S6sgUE021678; Fri, 28 Apr 2006 08:54:43 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-331.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Fri, 28 Apr 2006 08:54:42 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Re: is global HAHA so complex? Date: Fri, 28 Apr 2006 08:54:37 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021F15CB@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Re: is global HAHA so complex? Thread-Index: AcZp4NNGYhikuqntT/y2wYJqAvk0XgAro5hA From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 28 Apr 2006 06:54:42.0759 (UTC) FILETIME=[A085B570:01C66A90] X-Spam-Score: 0.0 (/) X-Scan-Signature: 9ed51c9d1356100bce94f1ae4ec616a9 Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >>> >>> 4G telephones are multihomed by definition. besides, the end device >> [Pascal] >> [Pascal] I am afraid there is some confusion here. I think we were >> talking about multiple Home Networks not multiple interfaces on the >> phone. In that context, could you please elaborate why 4G phones are >> multi-homed by definition? >> > >Ok, you mean why a telephone may need multiple Home Networks? > >Well, i guess that if you have a 4G phone and you have multiple >provider, you want to be able to be reachable through all the >providers. > >We do agree that is a desired feature to be able to have differetn >providers for different technologies, right? So that you don't have to >rely on a single provider giving UMTS, WIFI services for instance. > >In this scenario, you probably also want to have a stable address from >each of the providers, so that you can by pass any of the providers. I >mean if you are going to have a communication through the WiFI >interface, you don't want to have to exchange the initial packets and >the HoTI packets through the UMTS interface, right? > >So i really think that it is a very useful feature that 4G phones hame >multiple Home Addresses associated to different providers and different >technologies > >makes sense to you? > [Pascal] May happen... for some geeks who can tweak their phones.=20 That would not be the normal case for my grand ma. And I would not base any RO on this. Again redundancy but no RO (and a lot of hidden complexities). When my phone number is an IP address, I want to be always reachable at that IP address. So providing a real redundancy to Home makes sense,=20 before you enter complex solutions based on presence or SHIM6 to=20 duplicate everything on the clients. Pascal From nemo-bounces@ietf.org Fri Apr 28 04:15:21 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZO8C-0001H9-OG; Fri, 28 Apr 2006 04:15:16 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZO8A-0001Gy-UE for nemo@ietf.org; Fri, 28 Apr 2006 04:15:14 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZO89-0000zA-EH for nemo@ietf.org; Fri, 28 Apr 2006 04:15:14 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id D7C01212C61; Fri, 28 Apr 2006 11:15:11 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id A1DF3212C44; Fri, 28 Apr 2006 11:15:11 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 5F15DBDC40; Fri, 28 Apr 2006 11:15:11 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 2A607BDC38; Fri, 28 Apr 2006 11:15:11 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021F15BB@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021F15BB@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Date: Fri, 28 Apr 2006 11:15:19 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: f60d0f7806b0c40781eee6b9cd0b2135 Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli Subject: [nemo] Re: What RO do you get X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 28/04/2006, a las 9:43, Pascal Thubert (pthubert) escribi=F3: > >>>>> [Pascal] How many prefixes do you have in your MR in the SHIM6 >>>>> approach? >>>> >>>> shim6 relies on PA addressing, so you need one prefix per ISP > serving >>>> the home netowrk. >>> [Pascal] That was not my question. The RO ypu get with this depends > on >>> how close from your location the closest HA is. The SHIM approach > needs >>> one prefix in the MR per such location. So you end up with 1-3 >>> locations, and that does not give you Route Optimization. > Distributing >>> the same Home does better, and multiplying proxies does even better. >>> That was my point. >> >> well with a shim6 approach you would get as many prefixes as >> distributed locations and you would be able to use the prefix >> associated with the closer location. So you end up suing the HA which >> is closer to the MR. That should provide the same level of RO than >> global HAHA... >> > [Pascal] No, by orders of magnitude. > > Global HAHA allows you to place proxies wherever you place SGSNs=20 > todays, > and HAs wherever you place GGSNs. You'll find them in your city. So = the > wasted part of the path (the non ROed) is a few blocks. > but the important point is how many attachment point do you have to the=20= Internet, right? I mean having one HA per ISP would provide the same level of RO, since=20= no matter how many proxies you have you always end up accessing the=20 internet through the same number of ISPs... I think i am not understanding your point here... Are you saying that if you have more HA/proxies than the number of=20 attachment ISPs you get a shorter path? How can that be if you still=20 need to go through these limited number of ISPs? regards, marcelo > If I have to have one prefix by HA and you have no proxy, then the > number of location is limited to the number of prefix. How many=20 > prefixes > would you manage in a telephone? As a result, the non ROed part=20 > (between > you and the closest HA) is half a continent, though you might be=20 > talking > to someone next door (Alex's point in this discussion). > > I see where the SHIM6 approach provides connectivity redundancy = (though > not transparent to LFN as should be), but no RO at all, sorry. > > Pascal > From nemo-bounces@ietf.org Fri Apr 28 04:21:33 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZOEH-0003v1-0d; Fri, 28 Apr 2006 04:21:33 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZOEF-0003ut-QR for nemo@ietf.org; Fri, 28 Apr 2006 04:21:31 -0400 Received: from n2.nomadiclab.com ([193.234.219.2]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZOEF-0001Pb-BD for nemo@ietf.org; Fri, 28 Apr 2006 04:21:31 -0400 Received: from n2.nomadiclab.com (localhost [127.0.0.1]) by n2.nomadiclab.com (Postfix) with ESMTP id 275BC212C61; Fri, 28 Apr 2006 11:21:30 +0300 (EEST) Received: from outside.nomadiclab.com (d146.nomadiclab.com [193.234.218.146]) by n2.nomadiclab.com (Postfix) with ESMTP id E3F0B212C44; Fri, 28 Apr 2006 11:21:29 +0300 (EEST) Received: from outside.nomadiclab.com (localhost [127.0.0.1]) by outside.nomadiclab.com (Postfix) with ESMTP id 9F3CBBDC40; Fri, 28 Apr 2006 11:21:29 +0300 (EEST) Received: from [193.234.219.179] (w179.nomadiclab.com [193.234.219.179]) by outside.nomadiclab.com (Postfix) with ESMTP id 65FAFBDC38; Fri, 28 Apr 2006 11:21:29 +0300 (EEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021F15CB@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021F15CB@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: [nemo] Re: is global HAHA so complex? Date: Fri, 28 Apr 2006 11:21:38 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Virus-Scanned: ClamAV using ClamSMTP X-Virus-Scanned: ClamAV using ClamSMTP X-Spam-Score: 0.0 (/) X-Scan-Signature: d185fa790257f526fedfd5d01ed9c976 Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 28/04/2006, a las 9:54, Pascal Thubert (pthubert) escribi=F3: > >>>> >>>> 4G telephones are multihomed by definition. besides, the end device >>> [Pascal] >>> [Pascal] I am afraid there is some confusion here. I think we were >>> talking about multiple Home Networks not multiple interfaces on the >>> phone. In that context, could you please elaborate why 4G phones are >>> multi-homed by definition? >>> >> >> Ok, you mean why a telephone may need multiple Home Networks? >> >> Well, i guess that if you have a 4G phone and you have multiple >> provider, you want to be able to be reachable through all the >> providers. >> >> We do agree that is a desired feature to be able to have differetn >> providers for different technologies, right? So that you don't have = to >> rely on a single provider giving UMTS, WIFI services for instance. >> >> In this scenario, you probably also want to have a stable address = from >> each of the providers, so that you can by pass any of the providers. = I >> mean if you are going to have a communication through the WiFI >> interface, you don't want to have to exchange the initial packets and >> the HoTI packets through the UMTS interface, right? >> >> So i really think that it is a very useful feature that 4G phones = hame >> multiple Home Addresses associated to different providers and=20 >> different >> technologies >> >> makes sense to you? >> > [Pascal] May happen... for some geeks who can tweak their phones. I don't know.. i would expect that if i pay my mobile phone, this is=20 supported wihtout tweaks (of course this would be differetn if the=20 operator is subsidizing the phone) > That would not be the normal case for my grand ma. And I would not = base > any RO on this. Again redundancy but no RO (and a lot of hidden > complexities). > > When my phone number is an IP address, not really i would expect that for IP phones my phone number is a URI,=20= FQDN or something like this that is easier to remember and it can=20 easily be mapped to multiple IP addresses > I want to be always reachable at > that IP address. So providing a real redundancy to Home makes sense, > before you enter complex solutions based on presence or SHIM6 to > duplicate everything on the clients. > well, that is idea in putting intelegence in the edges and not in the=20 core, which seems to be the basis for scalability. I guess bottom line=20= disucssion here is about design principles. Of course that having the=20 clients to perform the functions imply additional complexities in the=20 clients, but making the network providing the fucntions, adds=20 compelxity to the netowrk, makes it less scalable and makes it more=20 rigid. regards, marcelo > Pascal > From nemo-bounces@ietf.org Fri Apr 28 21:30:03 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZeHR-0002oo-Vw; Fri, 28 Apr 2006 21:29:53 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZeHR-0002oj-Gh for nemo@ietf.org; Fri, 28 Apr 2006 21:29:53 -0400 Received: from yui.nc.u-tokyo.ac.jp ([130.69.251.116]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZeHQ-00029Q-S0 for nemo@ietf.org; Fri, 28 Apr 2006 21:29:53 -0400 Received: from [172.20.0.27] (p2139-ipad33hodogaya.kanagawa.ocn.ne.jp [221.184.101.139]) (authenticated bits=0) by yui.nc.u-tokyo.ac.jp (8.12.10/8.12.3/Debian-6.4) with ESMTP id k3T1SrUp029119; Sat, 29 Apr 2006 10:29:16 +0900 In-Reply-To: <8c9ad7af1b401f0eb4d6ff77569d37ca@it.uc3m.es> References: <7892795E1A87F04CADFCCF41FADD00FC021A92CB@xmb-ams-337.emea.cisco.com> <8c9ad7af1b401f0eb4d6ff77569d37ca@it.uc3m.es> Mime-Version: 1.0 (Apple Message framework v749.3) Content-Type: text/plain; charset=UTF-8; delsp=yes; format=flowed Message-Id: <39BF352B-F0CD-4235-BE31-4C1177C71C67@sfc.wide.ad.jp> Content-Transfer-Encoding: quoted-printable From: Ryuji Wakikawa Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Sat, 29 Apr 2006 04:20:00 +0900 To: marcelo bagnulo braun X-Mailer: Apple Mail (2.749.3) X-Spam-Score: 0.2 (/) X-Scan-Signature: a0534e6179a1e260079328e8b03c7901 Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" , Vijay Devarapalli , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org On 2006/04/25, at 18:43, marcelo bagnulo braun wrote: > > El 25/04/2006, a las 11:12, Pascal Thubert (pthubert) escribi=C3=B3: > >> Thanks for the explanation Marcelo. Now I see what you have in =20 >> mind. This approach is interesting; I guess it takes a mix of =20 >> MONAMI6 and SHIM6 to get the result done. >> > > yes it may be interesting, but you should first write a draft. >> IMHO, though, It looks fairly complex and remote, compared to the =20 >> use of existing routing protocols proposed in the global HAHA =20 >> approach. >> > > well, otoh, if any of the home networks is multihomed, then you =20 > need shim6 in anycase to support multihoming features and so this =20 > comes "for free" > > with respect to timeline, i am not sure but i wouldn't expect =20 > commercial airlines to start providing IPv6 connectivity in the =20 > short term, so i would argue that the timeline of this solution is =20 > compatible with the times of the provision of v6 services in =20 > commercial flights... (i guess involved people can correct me if i =20 > am wrong) HAHA is transparent to end nodes. I believe this is good feature of =20 HAHA,since modifying end nodes are burden. For getting RO, all the node must support SHIM6? This is somehow =20 difficult requirement. >> Note that: >> >> - the NEMO assumption is to have the MR handle the mobility on =20 >> behalf of LFNs which are unaware of the mobility. We want a simple =20= >> operation for simple Mobile Network Nodes. > > right, but this wouldn't result in exposing mobility to the nodes, =20 > since only the set of stable home address is exposed to the MNN. =20 > So, actual this preserves the mobility transparent to the MNN. No =20 > problem here AFAICT I don't think all multihomed node will support SHIM6. There is =20 another mechanism discussing in Monami6. ryuji >> >> - You are not solving the MIP6 Home Link issue, you are bypassing =20 >> it. Global HAHA gets rid of the ND operations on the Home Link, =20 >> therefore it removes the associated issues, replacing that =20 >> mechanism by traditional routing protocol operations. >> > > agree this is not a substitute for local HAHA, just for global HAHA > >> I'd say that before we assemble complex things that we do not =20 >> have, we could start thinking of using what we mostly have and =20 >> have been discussing for some time already. > > i would say that the relevant parameter is when is the expect time =20 > to deploy this... when are we expecting to have v6 services in =20 > commercial flights? > > As i said, perhaps the shim6 approach is not really appropriate for =20= > this scenario, but the way we can determine that is if we =20 > understand the requirements of this use case > > BEsides, note that shim6 approach could also be used as a RO =20 > mechanisms (if a topologically meaningful CoA is provided to the =20 > nodes) and it also is needed for providing multihoming support in =20 > multihomed nemos > > > regards, marcelo > > >> >> Pascal >> >>> -----Original Message----- >>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>> Sent: Tuesday, April 25, 2006 9:16 AM >>> To: Pascal Thubert (pthubert) >>> Cc: Alexandru Petrescu; nemo@ietf.org; Vijay Devarapalli >>> Subject: Re: About the proposed charter and global HAHA (was Re: =20 >>> [nemo] RE: About route aggregation >>> in global HAHA >>> >>> Hi Pascal, >>> >>> El 25/04/2006, a las 10:04, Pascal Thubert (pthubert) escribi=C3=B3: >>> >>>> Marcelo, Alexandru: >>>> >>>> I fail to understand this. Are you saying that SHIM6 can help =20 >>>> resolve >>>> the case where the MIP6/NEMO Home Link goes down? >>>> >>> >>> when the HA are distributed among different sites, yes, and in a bgp >>> routing table friendly manner >>> >>> i mean, suppose that you have multiple HAs located in different =20 >>> sites >>> and that each HA has a different PA home prefix assigned. then each >>> nemo has multiple home prefixes and each MNN has multiple =20 >>> addresses one >>> per available home prefix >>> >>> then you can use the shim6 to move from one home prefix to =20 >>> another, in >>> case that there is a failure in one of the HAs >>> >>> that is exactly what the shim6 protocol has been designed to do >>> >>> Again, maybe the global moving networks have some additional >>> requirements that don't fit in this approach, that is why i have =20 >>> been >>> asking for a requirement specification, in order to find out which >>> solution is acceptable >>> >>> Regards, marcelo >>> >>> >>>> Pascal >>>> >>>>> -----Original Message----- >>>>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>>>> Sent: Monday, April 24, 2006 2:41 PM >>>>> To: Alexandru Petrescu >>>>> Cc: Pascal Thubert (pthubert); Vijay Devarapalli; nemo@ietf.org >>>>> Subject: Re: About the proposed charter and global HAHA (was Re: >>>>> [nemo] RE: About route aggregation >>>>> in global HAHA >>>>> >>>>> Hi Alex, >>>>> >>>>> El 21/04/2006, a las 15:26, Alexandru Petrescu escribi=C3=B3: >>>>> >>>>>> Pascal Thubert (pthubert) wrote: >>>>>>> One of the motivations is to get rid of the issues associated =20= >>>>>>> with >>>>>>> the Home Link (single site, single point of failure) >>>>>> >>>>>> Address "single site - single point of failure"???? >>>>>> >>>>>> I could understand addressing single HA - single point of =20 >>>>>> failure (in >>>>>> which case multiple HAs can co-exist on the home link), but =20 >>>>>> "single >>>>>> site >>>>>> - single point of failure" can't be addressed by any protocol, =20= >>>>>> should >>>>>> better be addressed by some administrative organizations. Or by >>>>>> shim6. >>>>>> >>>>> >>>>> agree that shim6 could be used to provide some support for this >>>>> >>>>>> >>>>>>> and enable distributing Home and L3. >>>>>> >>>>>> I am afraid the network is made in such a way that IP =20 >>>>>> addresses can't >>>>>> be >>>>>> distributed. Have you looked at shim6? >>>>>> >>>>> >>>>> same thing >>>>> >>>>>>> RO comes when we introduce the satellites and route projection, >>>>>>> which >>>>>>> is another aspect of the draft. >>>>>> >>>>>> The draft seems to have too many aspects it tries to deal with. >>>>>> >>>>> >>>>> probably you could even do some shim6 based RO also... >>>>> >>>>> >>>>> that is one of the reasons why i am asking about the =20 >>>>> requirements and >>>>> the use cases that we want to address... >>>>> >>>>> first we need to define the problem that we want to work on and =20= >>>>> then >>>>> move to solution space... >>>>> >>>>> regards, marcelo >>>>> >>>>> >>>>>> Alex >>>>>> >>>> >> > From nemo-bounces@ietf.org Sat Apr 29 00:45:58 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZhLA-0007VV-2K; Sat, 29 Apr 2006 00:45:56 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZhL7-0007VQ-Ry for nemo@ietf.org; Sat, 29 Apr 2006 00:45:53 -0400 Received: from blv-smtpout-01.boeing.com ([130.76.32.69]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZhL5-0005jB-Aj for nemo@ietf.org; Sat, 29 Apr 2006 00:45:53 -0400 Received: from blv-av-01.boeing.com ([192.42.227.216]) by blv-smtpout-01.boeing.com (8.9.2.MG.10092003/8.8.5-M2) with ESMTP id VAA21606; Fri, 28 Apr 2006 21:45:49 -0700 (PDT) Received: from XCH-NWBH-11.nw.nos.boeing.com (localhost [127.0.0.1]) by blv-av-01.boeing.com (8.11.3/8.11.3/MBS-AV-LDAP-01) with ESMTP id k3T4jnr13342; Fri, 28 Apr 2006 21:45:49 -0700 (PDT) Received: from XCH-NW-8V1.nw.nos.boeing.com ([130.247.55.69]) by XCH-NWBH-11.nw.nos.boeing.com with Microsoft SMTPSVC(6.0.3790.1830); Fri, 28 Apr 2006 21:45:49 -0700 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Rewording of RO work in the charter Date: Fri, 28 Apr 2006 21:45:49 -0700 Message-ID: <0D090F1E0F5536449C7E6527AFFA280A21BFCB@XCH-NW-8V1.nw.nos.boeing.com> In-Reply-To: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Rewording of RO work in the charter Thread-Index: AcZlLCwIEWurnQZZSneCM5ITT1Sp3AGFdXqQ From: "Davis, Terry L" To: "T.J. Kniveton" , "ml-nemo WG" X-OriginalArrivalTime: 29 Apr 2006 04:45:49.0992 (UTC) FILETIME=[C9D8DA80:01C66B47] X-Spam-Score: 0.0 (/) X-Scan-Signature: 2beba50d0fcdeee5f091c59f204d4365 Cc: X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org T.J. You'll probably wish that you hadn't asked for this... These would be our ideal mobility solution. I realize that meeting all of these is probably an extreme stretch. Take care Terry =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D Generally the aviation mobility requirements are: 1- Ability to "dock" with numerous Internet service providers utilizing different media; satellite (KU & L), EVDO, 802.11/802.16 airport; hardwired Ethernet "gatelink". 2- Ability to be multihomed to different providers over different links simultaneously. The desired aircraft state will be to have two links active always. 3- Ability to assign traffic for specific providers links only (safety or operational certified providers) 4- Ability to re-profile traffic to match available link capacity and constraints when handing off between links. 5- Ability to seamlessly hand off existing network connections without dropping the session (we do this today). 6- Ability to maintain security through link hand off. 7- Ability to continue to receive multicast traffic after hand off. 8- Route optimization such that on hand off to a new link, connection/session routing is optimized to the new ground site. 9- Ability to maintain QoS on hand off with the capacity of the new link (You may hand off to a link with a smaller capacity or higher error rate or from an EVDO low latency link to a satellite high latency link) 10- Ability to support both IP-v6 and IP-v4 mobile platforms from an IP-v6 network. 11- Ability to receive priority multicast/broadcast messages. 12- Ability to establish connections to multiple onboard networks over a single link and hand off them simultaneously. (We expect the aircraft to have at three independent onboard isolated networks; air traffic control, airline operations, and passenger services. 13- Ability to initiate encrypted tunnels on all links with IPSec before passing traffic. > -----Original Message----- > From: T.J. Kniveton [mailto:tj@kniveton.com] > Sent: Friday, April 21, 2006 3:13 AM > To: ml-nemo WG > Subject: [nemo] Rewording of RO work in the charter >=20 > Folks, >=20 > I have been watching the conversation started by Marcelo a couple of > weeks ago, regarding the rewording of the (n,n,n) item in the > charter. (Sorry I haven't been commenting lately). >=20 > We need to figure out how to massage the concepts being batted around > so that we have something tractable that we can work on and complete. > From the discussion so far, I tend to think: >=20 > 1) We need somewhat more concrete idea of the requirements for the > type of situation we want to solve. I agree with Andrew that we don't > want to get bogged down in this, and I do believe Terry gave us a > list for the Conexion scenario quite a while ago. Perhaps he or > Andrew could give an updated list. >=20 > 2) We need a bit more analysis still on the RO problem space, based > on some idea of requirements for the solution to be worked out > (according to charter items). Specifically, the effects of having a > "Mobile ISP" vs. "ISP-independent mobile home networks", the > implications of that for the global BGP tables, how likely it is to > get custom agreements between a mobility provider and ISPs it (or its > customers) uses, etc. >=20 > 3) Route optimization - "Performance" as Marcelo called it - avoiding > long tunneling (geographically) - or avoiding expensive links. This > is what we have discussed so far, and we need a stronger boundary so > that we can define what is to be solved in these specific > circumstances. Otherwise this walks down the road toward many other > long-standing issues. My view is that having a good list of > requirements for the RO situation of interest would be helpful here. >=20 > 4) Aviation - question of "if" to use Mobile IP... NEMO base does use > it because it made sense. It's possible we have another deployment > case where it does not make sense, but we still want a solution for > Network Mobility. However, we have to be careful here. > There are some general network mobility problems.. such as > multiple ISPs, having high-cost links, and how/whether to inject > routes into the BGP tables. I believe Terry already gave us a list of > requirements for this type of situation - can we review that? >=20 > 5) Regarding HAHA - As Pascal mentioned, this wasn't supposed to be > "the only" or even "the" solution for RO. It is simply something that > addressed a shown need, and so I considered it to be something that > could be added for future work. But according to the list items > above, the group is not including or excluding any certain solution > at this point. We need more definition of what and how to solve, and > the implications of various approaches first. I would like to hear > the RO-analysis authors chime in on this one. >=20 > Besides Aviation, I have asked numerous times for anyone who is > actually deploying NEMO or NEMO-like technology to give us industry > input for requirements. Aside from aviation, I have heard a little > bit regarding Japanese auto makers, but not much else. I also used to > know things about some large cell phone makers that might be doing > these things, but never enough to make a good list of requirements. >=20 > Remember, our goal here is to reshape the charter to solve problems > so that NEMO can be deployed. If there are deployments that are > facing these issues, they can be brought forward to be studied at > this phase. >=20 > In conclusion, I would like to have a bit more discussion, and then > maybe people can offer suggestions on how specifically to reword > charter language and/or change milestones so that these things can be > accomplished. From nemo-bounces@ietf.org Sat Apr 29 03:03:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU8-00083J-3a; Sat, 29 Apr 2006 03:03:20 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU7-000838-FQ for nemo@ietf.org; Sat, 29 Apr 2006 03:03:19 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZjU5-00046G-31 for nemo@ietf.org; Sat, 29 Apr 2006 03:03:19 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 29 Apr 2006 09:03:17 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3T73GUE003107; Sat, 29 Apr 2006 09:03:16 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Sat, 29 Apr 2006 09:03:16 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Re: is global HAHA so complex? Date: Sat, 29 Apr 2006 09:02:52 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021F1A96@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Re: is global HAHA so complex? Thread-Index: AcZqnMy2XY+5cenMRyq5lPGor+eH9wAvO9gA From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 29 Apr 2006 07:03:16.0127 (UTC) FILETIME=[FCED32F0:01C66B5A] X-Spam-Score: 0.0 (/) X-Scan-Signature: 944ecb6e61f753561f559a497458fb4f Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Marcelo; There's a lot of good sense in your words; and I'd agree in general = terms.=20 Now please consider this angle: the design point we want to fix in MIP6 = is the Home Link, which is a single point of failure, a bottleneck and = anchors Home at a given location. We want to replace the ND based = operation by a protocol over IP and enable a real (site or HA) = redundancy using routing protocols. In other words, we want to fix MIP problems within MIP. Not to force ISP = to deploy multiple prefixes, and either SHIM6 or PRESENCE to fix a = limitation of another protocol. I do not think that HAHA breaks the end = to end principle. I think it obeys another principle, Give back to Cesar = what is Cesar's, fix a problem where it occurs. Pascal >-----Original Message----- >From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >Sent: Friday, April 28, 2006 10:22 AM >To: Pascal Thubert (pthubert) >Cc: nemo@ietf.org; Henrik Levkowetz; Vijay Devarapalli >Subject: Re: [nemo] Re: is global HAHA so complex? > > >El 28/04/2006, a las 9:54, Pascal Thubert (pthubert) escribi=F3: > >> >>>>> >>>>> 4G telephones are multihomed by definition. besides, the end = device >>>> [Pascal] >>>> [Pascal] I am afraid there is some confusion here. I think we were >>>> talking about multiple Home Networks not multiple interfaces on the >>>> phone. In that context, could you please elaborate why 4G phones = are >>>> multi-homed by definition? >>>> >>> >>> Ok, you mean why a telephone may need multiple Home Networks? >>> >>> Well, i guess that if you have a 4G phone and you have multiple >>> provider, you want to be able to be reachable through all the >>> providers. >>> >>> We do agree thFrom nemo-bounces@ietf.org Sat Apr 29 03:03:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU8-00083J-3a; Sat, 29 Apr 2006 03:03:20 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU7-000838-FQ for nemo@ietf.org; Sat, 29 Apr 2006 03:03:19 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZjU5-00046G-31 for nemo@ietf.org; Sat, 29 Apr 2006 03:03:19 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 29 Apr 2006 09:03:17 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3T73GUE003107; Sat, 29 Apr 2006 09:03:16 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Sat, 29 Apr 2006 09:03:16 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Re: is global HAHA so complex? Date: Sat, 29 Apr 2006 09:02:52 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021F1A96@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Re: is global HAHA so complex? Thread-Index: AcZqnMy2XY+5cenMRyq5lPGor+eH9wAvO9gA From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 29 Apr 2006 07:03:16.0127 (UTC) FILETIME=[FCED32F0:01C66B5A] X-Spam-Score: 0.0 (/) X-Scan-Signature: 944ecb6e61f753561f559a497458fb4f Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Marcelo; There's a lot of good sense in your words; and I'd agree in general = terms.=20 Now please consider this angle: the design point we want to fix in MIP6 = is the Home Link, which is a single point of failure, a bottleneck and = anchors Home at a given location. We want to replace the ND based = operation by a protocol over IP and enable a real (site or HA) = redundancy using routing protocols. In other words, we want to fix MIP problems within MIP. Not to force ISP = to deploy multiple prefixes, and either SHIM6 or PRESENCE to fix a = limitation of another protocol. I do not think that HAHA breaks the end = to end principle. I think it obeys another principle, Give back to Cesar = what is Cesar's, fix a problem where it occurs. Pascal >-----Original Message----- >From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >Sent: Friday, April 28, 2006 10:22 AM >To: Pascal Thubert (pthubert) >Cc: nemo@ietf.org; Henrik Levkowetz; Vijay Devarapalli >Subject: Re: [nemo] Re: is global HAHA so complex? > > >El 28/04/2006, a las 9:54, Pascal Thubert (pthubert) escribi=F3: > >> >>>>> >>>>> 4G telephones are multihomed by definition. besides, the end = device >>>> [Pascal] >>>> [Pascal] I am afraid there is some confusion here. I think we were >>>> talking about multiple Home Networks not multiple interfaces on the >>>> phone. In that context, could you please elaborate why 4G phones = are >>>> multi-homed by definition? >>>> >>> >>> Ok, you mean why a telephone may need multiple Home Networks? >>> >>> Well, i guess that if you have a 4G phone and you have multiple >>> provider, you want to be able to be reachable through all the >>> providers. >>> >>> We do agree that is a desired feature to be able to have differetn >>> providers for different technologies, right? So that you don't have = to >>> rely on a single provider giving UMTS, WIFI services for instance. >>> >>> In this scenario, you probably also want to have a stable address = from >>> each of the providers, so that you can by pass any of the providers. = I >>> mean if you are going to have a communication through the WiFI >>> interface, you don't want to have to exchange the initial packets = and >>> the HoTI packets through the UMTS interface, right? >>> >>> So i really think that it is a very useful feature that 4G phones = hame >>> multiple Home Addresses associated to different providers and >>> different >>> technologies >>> >>> makes sense to you? >>> >> [Pascal] May happen... for some geeks who can tweak their phones. > >I don't know.. i would expect that if i pay my mobile phone, this is >supported wihtout tweaks (of course this would be differetn if the >operator is subsidizing the phone) > >> That would not be the normal case for my grand ma. And I would not = base >> any RO on this. Again redundancy but no RO (and a lot of hidden >> complexities). > > > >> >> When my phone number is an IP address, > >not really i would expect that for IP phones my phone number is a URI, >FQDN or something like this that is easier to remember and it can >easily be mapped to multiple IP addresses > >> I want to be always reachable at >> that IP address. So providing a real redundancy to Home makes sense, >> before you enter complex solutions based on presence or SHIM6 to >> duplicate everything on the clients. >> > >well, that is idea in putting intelegence in the edges and not in the >core, which seems to be the basis for scalability. I guess bottom line >disucssion here is about design principles. Of course that having the >clients to perform the functions imply additional complexities in the >clients, but making the network providing the fucntions, adds >compelxity to the netowrk, makes it less scalable and makes it more >rigid. > >regards, marcelo > >> Pascal >> From nemo-bounces@ietf.org Sat Apr 29 03:03:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU7-00083D-Rh; Sat, 29 Apr 2006 03:03:19 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU6-000833-W2 for nemo@ietf.org; Sat, 29 Apr 2006 03:03:18 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZjU3-00046E-Gx for nemo@ietf.org; Sat, 29 Apr 2006 03:03:18 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 29 Apr 2006 09:03:16 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3T73DUE003103; Sat, 29 Apr 2006 09:03:14 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Sat, 29 Apr 2006 09:03:13 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Re: What RO do you get Date: Sat, 29 Apr 2006 09:03:13 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021F1A95@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Re: What RO do you get Thread-Index: AcZqm/6Pk7Y2ZHYKTteX+0aNI9PnlQAuSwWg From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 29 Apr 2006 07:03:13.0721 (UTC) FILETIME=[FB7E1290:01C66B5A] X-Spam-Score: 0.0 (/) X-Scan-Signature: c3a18ef96977fc9bcc21a621cbf1174b Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO at is a desired feature to be able to have differetn >>> providers for different technologies, right? So that you don't have = to >>> rely on a single provider giving UMTS, WIFI services for instance. >>> >>> In this scenario, you probably also want to have a stable address = from >>> each of the providers, so that you can by pass any of the providers. = I >>> mean if you are going to have a communication through the WiFI >>> interface, you don't want to have to exchange the initial packets = and >>> the HoTI packets through the UMTS interface, right? >>> >>> So i really think that it is a very useful feature that 4G phones = hame >>> multiple Home Addresses associated to different providers and >>> different >>> technologies >>> >>> makes sense to you? >>> >> [Pascal] May happen... for some geeks who can tweak their phones. > >I don't know.. i would expect that if i pay my mobile phone, this is >supported wihtout tweaks (of course this would be differetn if the >operator is subsidizing the phone) > >> That would not be the normal case for my grand ma. And I would not = base >> any RO on this. Again redundancy but no RO (and a lot of hidden >> complexities). > > > >> >> When my phone number is an IP address, > >not really i would expect that for IP phones my phone number is a URI, >FQDN or something like this that is easier to remember and it can >easily be mapped to multiple IP addresses > >> I want to be always reachable at >> that IP address. So providing a real redundancy to Home makes sense, >> before you enter complex solutions based on presence or SHIM6 to >> duplicate everything on the clients. >> > >well, that is idea in putting intelegence in the edges and not in the >core, which seems to be the basis for scalability. I guess bottom line >disucssion here is about design principles. Of course that having the >clients to perform the functions imply additional complexities in the >clients, but making the network providing the fucntions, adds >compelxity to the netowrk, makes it less scalable and makes it more >rigid. > >regards, marcelo > >> Pascal >> From nemo-bounces@ietf.org Sat Apr 29 03:03:29 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU7-00083D-Rh; Sat, 29 Apr 2006 03:03:19 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZjU6-000833-W2 for nemo@ietf.org; Sat, 29 Apr 2006 03:03:18 -0400 Received: from ams-iport-1.cisco.com ([144.254.224.140]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZjU3-00046E-Gx for nemo@ietf.org; Sat, 29 Apr 2006 03:03:18 -0400 Received: from ams-core-1.cisco.com ([144.254.224.150]) by ams-iport-1.cisco.com with ESMTP; 29 Apr 2006 09:03:16 +0200 Received: from xbh-ams-332.emea.cisco.com (xbh-ams-332.cisco.com [144.254.231.87]) by ams-core-1.cisco.com (8.12.10/8.12.6) with ESMTP id k3T73DUE003103; Sat, 29 Apr 2006 09:03:14 +0200 (MEST) Received: from xmb-ams-337.cisco.com ([144.254.231.82]) by xbh-ams-332.emea.cisco.com with Microsoft SMTPSVC(6.0.3790.0); Sat, 29 Apr 2006 09:03:13 +0200 X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Subject: RE: [nemo] Re: What RO do you get Date: Sat, 29 Apr 2006 09:03:13 +0200 Message-ID: <7892795E1A87F04CADFCCF41FADD00FC021F1A95@xmb-ams-337.emea.cisco.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: [nemo] Re: What RO do you get Thread-Index: AcZqm/6Pk7Y2ZHYKTteX+0aNI9PnlQAuSwWg From: "Pascal Thubert \(pthubert\)" To: "marcelo bagnulo braun" X-OriginalArrivalTime: 29 Apr 2006 07:03:13.0721 (UTC) FILETIME=[FB7E1290:01C66B5A] X-Spam-Score: 0.0 (/) X-Scan-Signature: c3a18ef96977fc9bcc21a621cbf1174b Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >>> well with a shim6 approach you would get as many prefixes as >>> distributed locations and you would be able to use the prefix >>> associated with the closer location. So you end up suing the HA which >>> is closer to the MR. That should provide the same level of RO than >>> global HAHA... >>> >> [Pascal] No, by orders of magnitude. >> >> Global HAHA allows you to place proxies wherever you place SGSNs >> todays, >> and HAs wherever you place GGSNs. You'll find them in your city. So the >> wasted part of the path (the non ROed) is a few blocks. >> > >but the important point is how many attachment point do you have to the >Internet, right? > [Pascal] How many anchor points, places that you can bind to. It's a bit similar to HMIP, if you wish. But a MIP proxy HA works like a classical proxy, not like a MAP; though both provide regional mobility. >I mean having one HA per ISP would provide the same level of RO, since >no matter how many proxies you have you always end up accessing the >internet through the same number of ISPs... > [Pascal] You access the Internet wherever you like and then bind to the closest proxy. Note that for 4G mobile phones or ISP 802.11 access, the proxy might be collocated with S/GGSNs. If can you join the APF/SSID at layer 2, it means that there is a proxy for you right behind at layer 3. If you access via a free connection, IP will take you to the nearest site that can proxy for you. >I think i am not understanding your point here... > >Are you saying that if you have more HA/proxies than the number of >attachment ISPs you get a shorter path? How can that be if you still >need to go through these limited number of ISPs? [Pascal]=20 Maybe thinking phone will help. In you mind image, think of current GPRS.=20 We have used it with Doors for years now. Wherever we roam, our packets go back to the HA (say in France). If I was using SHIM6, there could be another HA for me in the US so if I am there it could help a little bit. But not much if I am in Japan. In the global HAHA world, there is a Home site (not yours if you are a visitor in town but then it can proxy for you) at the GGSN and a maybe even a proxy HA at the SGSN. The RO you get initially depends on how well the ISPs are meshed inside the consortium. But from there, the proxies can discover each other and establish a direct tunnel over the Internet, bypassing the consortium network for the data path. The resulting path is no different then if you were doing a direct connection. But there is an encapsulation that hides the mobility to the end points (the MR =3D=3D 4G phones). Say that there are HAHA roaming agreements between ISP A and ISP B. Say that ISP A' belongs to the same ISP consortium as ISP A so they share the same "global HAHA subway", the same Home Aggregation that ends up in the DFZ. Then the RO path between phone A and phone B will be: Phone A <=3D=3D> proxy A' <=3D=3D> proxy B' <=3D=3D> phone B Looks like the normal path but for the fact that it's a proxies instead of a HAs. The difference is that proxy A' is very close to A and proxy B' is very close to B. So if there are proxies that you can use everywhere, and even though you have a single home site far, far away, you get RO and mostly fix ROPS problems 1 and 2. This builds on the way your phone roams today. You can attach to a base station that belongs to a partner if you have roaming agreements. In terms of HAHA, this is expressed by the fact that the partner (A') exposes the Home aggregation that encompasses your Home Network. I hope this helps? I might have missed your question. Pascal Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org >>> well with a shim6 approach you would get as many prefixes as >>> distributed locations and you would be able to use the prefix >>> associated with the closer location. So you end up suing the HA which >>> is closer to the MR. That should provide the same level of RO than >>> global HAHA... >>> >> [Pascal] No, by orders of magnitude. >> >> Global HAHA allows you to place proxies wherever you place SGSNs >> todays, >> and HAs wherever you place GGSNs. You'll find them in your city. So the >> wasted part of the path (the non ROed) is a few blocks. >> > >but the important point is how many attachment point do you have to the >Internet, right? > [Pascal] How many anchor points, places that you can bind to. It's a bit similar to HMIP, if you wish. But a MIP proxy HA works like a classical proxy, not like a MAP; though both provide regional mobility. >I mean having one HA per ISP would provide the same level of RO, since >no matter how many proxies you have you always end up accessing the >internet through the same number of ISPs... > [Pascal] You access the Internet wherever you like and then bind to the closest proxy. Note that for 4G mobile phones or ISP 802.11 access, the proxy might be collocated with S/GGSNs. If can you join the APF/SSID at layer 2, it means that there is a proxy for you right behind at layer 3. If you access via a free connection, IP will take you to the nearest site that can proxy for you. >I think i am not understanding your point here... > >Are you saying that if you have more HA/proxies than the number of >attachment ISPs you get a shorter path? How can that be if you still >need to go through these limited number of ISPs? [Pascal]=20 Maybe thinking phone will help. In you mind image, think of current GPRS.=20 We have used it with Doors for years now. Wherever we roam, our packets go back to the HA (say in France). If I was using SHIM6, there could be another HA for me in the US so if I am there it could help a little bit. But not much if I am in Japan. In the global HAHA world, there is a Home site (not yours if you are a visitor in town but then it can proxy for you) at the GGSN and a maybe even a proxy HA at the SGSN. The RO you get initially depends on how well the ISPs are meshed inside the consortium. But from there, the proxies can discover each other and establish a direct tunnel over the Internet, bypassing the consortium network for the data path. The resulting path is no different then if you were doing a direct connection. But there is an encapsulation that hides the mobility to the end points (the MR =3D=3D 4G phones). Say that there are HAHA roaming agreements between ISP A and ISP B. Say that ISP A' belongs to the same ISP consortium as ISP A so they share the same "global HAHA subway", the same Home Aggregation that ends up in the DFZ. Then the RO path between phone A and phone B will be: Phone A <=3D=3D> proxy A' <=3D=3D> proxy B' <=3D=3D> phone B Looks like the normal path but for the fact that it's a proxies instead of a HAs. The difference is that proxy A' is very close to A and proxy B' is very close to B. So if there are proxies that you can use everywhere, and even though you have a single home site far, far away, you get RO and mostly fix ROPS problems 1 and 2. This builds on the way your phone roams today. You can attach to a base station that belongs to a partner if you have roaming agreements. In terms of HAHA, this is expressed by the fact that the partner (A') exposes the Home aggregation that encompasses your Home Network. I hope this helps? I might have missed your question. Pascal From nemo-bounces@ietf.org Sat Apr 29 13:45:36 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZtVb-0000aS-PU; Sat, 29 Apr 2006 13:45:31 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZtVa-0000aN-Ie for nemo@ietf.org; Sat, 29 Apr 2006 13:45:30 -0400 Received: from smtp01.uc3m.es ([163.117.136.121]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZtVZ-0002To-PA for nemo@ietf.org; Sat, 29 Apr 2006 13:45:30 -0400 Received: from smtp01.uc3m.es (localhost [127.0.0.1]) by localhost.uc3m.es (Postfix) with ESMTP id 4F1CFD4B69; Sat, 29 Apr 2006 19:45:28 +0200 (CEST) Received: from [163.117.203.197] (unknown [163.117.203.197]) by smtp01.uc3m.es (Postfix) with ESMTP id 29664D4B7E; Sat, 29 Apr 2006 19:45:26 +0200 (CEST) In-Reply-To: <39BF352B-F0CD-4235-BE31-4C1177C71C67@sfc.wide.ad.jp> References: <7892795E1A87F04CADFCCF41FADD00FC021A92CB@xmb-ams-337.emea.cisco.com> <8c9ad7af1b401f0eb4d6ff77569d37ca@it.uc3m.es> <39BF352B-F0CD-4235-BE31-4C1177C71C67@sfc.wide.ad.jp> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: <85a0a98754e43faf045e164e83829481@it.uc3m.es> Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: About the proposed charter and global HAHA (was Re: [nemo] RE: About route aggregation in global HAHA Date: Sat, 29 Apr 2006 20:45:25 +0300 To: Ryuji Wakikawa X-Mailer: Apple Mail (2.623) X-Spam-Score: 0.0 (/) X-Scan-Signature: 515708a075ffdf0a79d1c83b601e2afd Cc: nemo@ietf.org, "Pascal Thubert \(pthubert\)" , Vijay Devarapalli , Alexandru Petrescu X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Ryuji, El 28/04/2006, a las 22:20, Ryuji Wakikawa escribi=F3: > > On 2006/04/25, at 18:43, marcelo bagnulo braun wrote: > >> >> El 25/04/2006, a las 11:12, Pascal Thubert (pthubert) escribi=F3: >> >>> Thanks for the explanation Marcelo. Now I see what you have in mind.=20= >>> This approach is interesting; I guess it takes a mix of MONAMI6 and=20= >>> SHIM6 to get the result done. >>> >> >> yes > > it may be interesting, but you should first write a draft. > sure, but we are still in the draft discussion phase, no specific RO=20 solutions were being considered by the wg so far AFAICT >>> IMHO, though, It looks fairly complex and remote, compared to the=20 >>> use of existing routing protocols proposed in the global HAHA=20 >>> approach. >>> >> >> well, otoh, if any of the home networks is multihomed, then you need=20= >> shim6 in anycase to support multihoming features and so this comes=20 >> "for free" >> >> with respect to timeline, i am not sure but i wouldn't expect=20 >> commercial airlines to start providing IPv6 connectivity in the short=20= >> term, so i would argue that the timeline of this solution is=20 >> compatible with the times of the provision of v6 services in=20 >> commercial flights... (i guess involved people can correct me if i am=20= >> wrong) > > HAHA is transparent to end nodes. I believe this is good feature of=20 > HAHA,since modifying end nodes are burden. > For getting RO, all the node must support SHIM6? This is somehow=20 > difficult requirement. > This is indeed a difficulty for deployment. But again there are several=20= issues to consider, in particular, the requirements of the solution=20 (without them we cannot properly evaluate what solution fits better the=20= problem), the deployment time frames that we are habdling and other=20 tradeoff between benefits and costs (including contribution to the=20 global routing tables and deployment issues that you mention and so on) Again, we first need to determine what are the requierements for a=20 solution, withotu them we cannot possibly tell which solution is better >>> Note that: >>> >>> - the NEMO assumption is to have the MR handle the mobility on=20 >>> behalf of LFNs which are unaware of the mobility. We want a simple=20= >>> operation for simple Mobile Network Nodes. >> >> right, but this wouldn't result in exposing mobility to the nodes,=20 >> since only the set of stable home address is exposed to the MNN. So,=20= >> actual this preserves the mobility transparent to the MNN. No problem=20= >> here AFAICT > > I don't think all multihomed node will support SHIM6. There is another=20= > mechanism discussing in Monami6. > not for multiple HoAs MONAMI is only scoped to deal with multiple CoAs. That does not provide=20= support for multiple home networks with different prefixes. regards, marcelo > ryuji > >>> >>> - You are not solving the MIP6 Home Link issue, you are bypassing=20 >>> it. Global HAHA gets rid of the ND operations on the Home Link,=20 >>> therefore it removes the associated issues, replacing that mechanism=20= >>> by traditional routing protocol operations. >>> >> >> agree this is not a substitute for local HAHA, just for global HAHA >> >>> I'd say that before we assemble complex things that we do not have,=20= >>> we could start thinking of using what we mostly have and have been=20= >>> discussing for some time already. >> >> i would say that the relevant parameter is when is the expect time to=20= >> deploy this... when are we expecting to have v6 services in=20 >> commercial flights? >> >> As i said, perhaps the shim6 approach is not really appropriate for=20= >> this scenario, but the way we can determine that is if we understand=20= >> the requirements of this use case >> >> BEsides, note that shim6 approach could also be used as a RO=20 >> mechanisms (if a topologically meaningful CoA is provided to the=20 >> nodes) and it also is needed for providing multihoming support in=20 >> multihomed nemos >> >> >> regards, marcelo >> >> >>> >>> Pascal >>> >>>> -----Original Message----- >>>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>>> Sent: Tuesday, April 25, 2006 9:16 AM >>>> To: Pascal Thubert (pthubert) >>>> Cc: Alexandru Petrescu; nemo@ietf.org; Vijay Devarapalli >>>> Subject: Re: About the proposed charter and global HAHA (was Re:=20 >>>> [nemo] RE: About route aggregation >>>> in global HAHA >>>> >>>> Hi Pascal, >>>> >>>> El 25/04/2006, a las 10:04, Pascal Thubert (pthubert) escribi=F3: >>>> >>>>> Marcelo, Alexandru: >>>>> >>>>> I fail to understand this. Are you saying that SHIM6 can help=20 >>>>> resolve >>>>> the case where the MIP6/NEMO Home Link goes down? >>>>> >>>> >>>> when the HA are distributed among different sites, yes, and in a = bgp >>>> routing table friendly manner >>>> >>>> i mean, suppose that you have multiple HAs located in different=20 >>>> sites >>>> and that each HA has a different PA home prefix assigned. then each >>>> nemo has multiple home prefixes and each MNN has multiple addresses=20= >>>> one >>>> per available home prefix >>>> >>>> then you can use the shim6 to move from one home prefix to another,=20= >>>> in >>>> case that there is a failure in one of the HAs >>>> >>>> that is exactly what the shim6 protocol has been designed to do >>>> >>>> Again, maybe the global moving networks have some additional >>>> requirements that don't fit in this approach, that is why i have=20 >>>> been >>>> asking for a requirement specification, in order to find out which >>>> solution is acceptable >>>> >>>> Regards, marcelo >>>> >>>> >>>>> Pascal >>>>> >>>>>> -----Original Message----- >>>>>> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >>>>>> Sent: Monday, April 24, 2006 2:41 PM >>>>>> To: Alexandru Petrescu >>>>>> Cc: Pascal Thubert (pthubert); Vijay Devarapalli; nemo@ietf.org >>>>>> Subject: Re: About the proposed charter and global HAHA (was Re: >>>>>> [nemo] RE: About route aggregation >>>>>> in global HAHA >>>>>> >>>>>> Hi Alex, >>>>>> >>>>>> El 21/04/2006, a las 15:26, Alexandru Petrescu escribi=F3: >>>>>> >>>>>>> Pascal Thubert (pthubert) wrote: >>>>>>>> One of the motivations is to get rid of the issues associated=20= >>>>>>>> with >>>>>>>> the Home Link (single site, single point of failure) >>>>>>> >>>>>>> Address "single site - single point of failure"???? >>>>>>> >>>>>>> I could understand addressing single HA - single point of=20 >>>>>>> failure (in >>>>>>> which case multiple HAs can co-exist on the home link), but=20 >>>>>>> "single >>>>>>> site >>>>>>> - single point of failure" can't be addressed by any protocol,=20= >>>>>>> should >>>>>>> better be addressed by some administrative organizations. Or by >>>>>>> shim6. >>>>>>> >>>>>> >>>>>> agree that shim6 could be used to provide some support for this >>>>>> >>>>>>> >>>>>>>> and enable distributing Home and L3. >>>>>>> >>>>>>> I am afraid the network is made in such a way that IP addresses=20= >>>>>>> can't >>>>>>> be >>>>>>> distributed. Have you looked at shim6? >>>>>>> >>>>>> >>>>>> same thing >>>>>> >>>>>>>> RO comes when we introduce the satellites and route projection, >>>>>>>> which >>>>>>>> is another aspect of the draft. >>>>>>> >>>>>>> The draft seems to have too many aspects it tries to deal with. >>>>>>> >>>>>> >>>>>> probably you could even do some shim6 based RO also... >>>>>> >>>>>> >>>>>> that is one of the reasons why i am asking about the requirements=20= >>>>>> and >>>>>> the use cases that we want to address... >>>>>> >>>>>> first we need to define the problem that we want to work on and=20= >>>>>> then >>>>>> move to solution space... >>>>>> >>>>>> regards, marcelo >>>>>> >>>>>> >>>>>>> Alex >>>>>>> >>>>> >>> >> > From nemo-bounces@ietf.org Sat Apr 29 14:24:09 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZu6w-0005vr-WE; Sat, 29 Apr 2006 14:24:07 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZu6v-0005vm-Q4 for nemo@ietf.org; Sat, 29 Apr 2006 14:24:05 -0400 Received: from smtp01.uc3m.es ([163.117.136.121]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZu6v-0003r2-3M for nemo@ietf.org; Sat, 29 Apr 2006 14:24:05 -0400 Received: from smtp01.uc3m.es (localhost [127.0.0.1]) by localhost.uc3m.es (Postfix) with ESMTP id 38CCCD4A53; Sat, 29 Apr 2006 20:24:04 +0200 (CEST) Received: from [163.117.203.197] (unknown [163.117.203.197]) by smtp01.uc3m.es (Postfix) with ESMTP id 76AF4D2956; Sat, 29 Apr 2006 20:24:02 +0200 (CEST) In-Reply-To: <0D090F1E0F5536449C7E6527AFFA280A21BFCB@XCH-NW-8V1.nw.nos.boeing.com> References: <0D090F1E0F5536449C7E6527AFFA280A21BFCB@XCH-NW-8V1.nw.nos.boeing.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: [nemo] Rewording of RO work in the charter Date: Sat, 29 Apr 2006 21:24:00 +0300 To: "Davis, Terry L" X-Mailer: Apple Mail (2.623) X-Spam-Score: 0.0 (/) X-Scan-Signature: 36fb765c89ed47dab364ab702a78e8fd Cc: ml-nemo WG , "T.J. Kniveton" X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org Hi Terry, This is exactly what we need imho, thanks! Are you considering to put this in a ID?, i think this would be very=20 useful Some comments below... El 29/04/2006, a las 7:45, Davis, Terry L escribi=F3: > T.J. > > You'll probably wish that you hadn't asked for this... > > These would be our ideal mobility solution. I realize that meeting = all > of these is probably an extreme stretch. > maybe... perhaps it would be good to clasify these requirements according to=20 their importance. I mean i guess that there are some of them that are=20 critical and that a solution that doesn't fulfill those is a=20 non-starter, while maybe there are others that are more like=20 nice-to-have and you can maybe can live without (don't know, just=20 asking) > Take care > Terry > > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D > Generally the aviation mobility requirements are: > 1- Ability to "dock" with numerous Internet service providers = utilizing > different media; satellite (KU & L), EVDO, 802.11/802.16 airport; > hardwired Ethernet "gatelink". > > 2- Ability to be multihomed to different providers over different = links > simultaneously. The desired aircraft state will be to have two links > active always. > I am kind of confused w.r.t. the role of the providers in these two=20 points above... I mean in the first point, i understand that the goal is ubiquity,=20 meaning that the mobile network is able to obtain internet connectivity=20= using different technologies available. So when you talk about multiple=20= provider you mean multiple "first hop" providers. However, the second one, is perhaps more general, an it seems that you=20= are considering other issues like general fault tolerance, where you=20 want two independent paths to the Internet, wihtout a single point of=20 failure in the Internet connectivity provision, and you also seems to=20 be able to distribute the load among the different ISPs In summary, the first one is about having multiple access links to the=20= fixed internet while the second one seems to refer to the capability of=20= having multiple ISPs that provide indepndent paths to the internet , am=20= i reading this correctly? > 3- Ability to assign traffic for specific providers links only (safety > or operational certified providers) again, are talking about the access link only or the actual ISP to the=20= internet? > > 4- Ability to re-profile traffic to match available link capacity and > constraints when handing off between links. i guess we have two different cases here: when the new bandwidth is=20 enough for carrying the ongoing communications and when it is not. In=20 the first case, i guess that the goal is to distribute the load along=20 the different access links so that all can be fitted. On the second=20 case, i guess that what you want os to reduce the generated traffic for=20= instance by using the congestion avoidance mechanisms, like TCP slow=20 start... > > 5- Ability to seamlessly hand off existing network connections without > dropping the session (we do this today). > > 6- Ability to maintain security through link hand off. > i guess that you want this in general no only during handoff :-) what level of security are we cosnidering here? is it acceptable the=20 "do not introduce new vulnerabilities that are not present in the fixed=20= internet" criteria good enough or more is needed? > 7- Ability to continue to receive multicast traffic after hand off. > > 8- Route optimization such that on hand off to a new link, > connection/session routing is optimized to the new ground site. > > 9- Ability to maintain QoS on hand off with the capacity of the new=20 > link > (You may hand off to a link with a smaller capacity or higher error=20 > rate > or from an EVDO low latency link to a satellite high latency link) > sorry not following this one... could you expand? > 10- Ability to support both IP-v6 and IP-v4 mobile platforms from an > IP-v6 network. > this means that the access network is IPv6 and that you have nodes=20 within the mobile network that are v4 and v6? > 11- Ability to receive priority multicast/broadcast messages. > > 12- Ability to establish connections to multiple onboard networks over=20= > a > single link and hand off them simultaneously. (We expect the aircraft > to have at three independent onboard isolated networks; air traffic > control, airline operations, and passenger services. > this seems quite critical also... but how this differs from having three mobile networks colocated in the=20= same plane? > 13- Ability to initiate encrypted tunnels on all links with IPSec=20 > before > passing traffic. > thanks, marcelo > >> -----Original Message----- >> From: T.J. Kniveton [mailto:tj@kniveton.com] >> Sent: Friday, April 21, 2006 3:13 AM >> To: ml-nemo WG >> Subject: [nemo] Rewording of RO work in the charter >> >> Folks, >> >> I have been watching the conversation started by Marcelo a couple of >> weeks ago, regarding the rewording of the (n,n,n) item in the >> charter. (Sorry I haven't been commenting lately). >> >> We need to figure out how to massage the concepts being batted around >> so that we have something tractable that we can work on and complete. >> =46rom the discussion so far, I tend to think: >> >> 1) We need somewhat more concrete idea of the requirements for the >> type of situation we want to solve. I agree with Andrew that we don't >> want to get bogged down in this, and I do believe Terry gave us a >> list for the Conexion scenario quite a while ago. Perhaps he or >> Andrew could give an updated list. >> >> 2) We need a bit more analysis still on the RO problem space, based >> on some idea of requirements for the solution to be worked out >> (according to charter items). Specifically, the effects of having a >> "Mobile ISP" vs. "ISP-independent mobile home networks", the >> implications of that for the global BGP tables, how likely it is to >> get custom agreements between a mobility provider and ISPs it (or its >> customers) uses, etc. >> >> 3) Route optimization - "Performance" as Marcelo called it - avoiding >> long tunneling (geographically) - or avoiding expensive links. This >> is what we have discussed so far, and we need a stronger boundary so >> that we can define what is to be solved in these specific >> circumstances. Otherwise this walks down the road toward many other >> long-standing issues. My view is that having a good list of >> requirements for the RO situation of interest would be helpful here. >> >> 4) Aviation - question of "if" to use Mobile IP... NEMO base does use >> it because it made sense. It's possible we have another deployment >> case where it does not make sense, but we still want a solution for >> Network Mobility. However, we have to be careful here. >> There are some general network mobility problems.. such as >> multiple ISPs, having high-cost links, and how/whether to inject >> routes into the BGP tables. I believe Terry already gave us a list of >> requirements for this type of situation - can we review that? >> >> 5) Regarding HAHA - As Pascal mentioned, this wasn't supposed to be >> "the only" or even "the" solution for RO. It is simply something that >> addressed a shown need, and so I considered it to be something that >> could be added for future work. But according to the list items >> above, the group is not including or excluding any certain solution >> at this point. We need more definition of what and how to solve, and >> the implications of various approaches first. I would like to hear >> the RO-analysis authors chime in on this one. >> >> Besides Aviation, I have asked numerous times for anyone who is >> actually deploying NEMO or NEMO-like technology to give us industry >> input for requirements. Aside from aviation, I have heard a little >> bit regarding Japanese auto makers, but not much else. I also used to >> know things about some large cell phone makers that might be doing >> these things, but never enough to make a good list of requirements. >> >> Remember, our goal here is to reshape the charter to solve problems >> so that NEMO can be deployed. If there are deployments that are >> facing these issues, they can be brought forward to be studied at >> this phase. >> >> In conclusion, I would like to have a bit more discussion, and then >> maybe people can offer suggestions on how specifically to reword >> charter language and/or change milestones so that these things can be >> accomplished. > > From nemo-bounces@ietf.org Sat Apr 29 14:35:35 2006 Received: from [127.0.0.1] (helo=stiedprmman1.va.neustar.com) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZuI2-0001xP-Gq; Sat, 29 Apr 2006 14:35:34 -0400 Received: from [10.91.34.44] (helo=ietf-mx.ietf.org) by megatron.ietf.org with esmtp (Exim 4.43) id 1FZuI2-0001xK-6o for nemo@ietf.org; Sat, 29 Apr 2006 14:35:34 -0400 Received: from smtp01.uc3m.es ([163.117.136.121]) by ietf-mx.ietf.org with esmtp (Exim 4.43) id 1FZuI0-0004Xj-5O for nemo@ietf.org; Sat, 29 Apr 2006 14:35:32 -0400 Received: from smtp01.uc3m.es (localhost [127.0.0.1]) by localhost.uc3m.es (Postfix) with ESMTP id 40083B8DC3; Sat, 29 Apr 2006 20:35:29 +0200 (CEST) Received: from [163.117.203.197] (unknown [163.117.203.197]) by smtp01.uc3m.es (Postfix) with ESMTP id 208B2C8FFB; Sat, 29 Apr 2006 20:35:27 +0200 (CEST) In-Reply-To: <7892795E1A87F04CADFCCF41FADD00FC021F1A96@xmb-ams-337.emea.cisco.com> References: <7892795E1A87F04CADFCCF41FADD00FC021F1A96@xmb-ams-337.emea.cisco.com> Mime-Version: 1.0 (Apple Message framework v623) Content-Type: text/plain; charset=ISO-8859-1; format=flowed Message-Id: Content-Transfer-Encoding: quoted-printable From: marcelo bagnulo braun Subject: Re: [nemo] Re: is global HAHA so complex? Date: Sat, 29 Apr 2006 21:35:24 +0300 To: "Pascal Thubert (pthubert)" X-Mailer: Apple Mail (2.623) X-Spam-Score: 0.0 (/) X-Scan-Signature: 0770535483960d190d4a0d020e7060bd Cc: nemo@ietf.org, Henrik Levkowetz , Vijay Devarapalli X-BeenThere: nemo@ietf.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: NEMO Working Group List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Errors-To: nemo-bounces@ietf.org El 29/04/2006, a las 10:02, Pascal Thubert (pthubert) escribi=F3: > Hi Marcelo; > > There's a lot of good sense in your words; and I'd agree in general=20 > terms. > > Now please consider this angle: the design point we want to fix in=20 > MIP6 is the Home Link, which is a single point of failure, a=20 > bottleneck and anchors Home at a given location. We want to replace=20 > the ND based operation by a protocol over IP and enable a real (site=20= > or HA) redundancy using routing protocols. > > In other words, we want to fix MIP problems within MIP. Not to force=20= > ISP to deploy multiple prefixes, and either SHIM6 or PRESENCE to fix a=20= > limitation of another protocol. I do not think that HAHA breaks the=20 > end to end principle. I think it obeys another principle, Give back to=20= > Cesar what is Cesar's, fix a problem where it occurs. > but the goal is to try to provision certain capability as i understand=20= it. There are different tools that can be used to provide (or can be=20 tweaked to provide) the required capabilities. the tool that provides a better trade-offs will is the one that needs=20 to be selected. Please note that my point is that we need to have a scale to evaluate=20 the different proposals. This scale are the requirements imposed by the=20= user (in this case the aviation community) Once we have those we can see which solution fits their needs better.=20 If the case is that their needs require a global haha like and it make=20= sense to add routing table entries for this, i am all for it. But this=20= analysis needs still to be done and this needs to be part of the=20 charter imho. First define requirements for aviation Second define a solution (i.e. select the solution that fits the best=20 the requirements defined) (as opposed to pick one solution based on other criteria) regards, marcelo > Pascal > >> -----Original Message----- >> From: marcelo bagnulo braun [mailto:marcelo@it.uc3m.es] >> Sent: Friday, April 28, 2006 10:22 AM >> To: Pascal Thubert (pthubert) >> Cc: nemo@ietf.org; Henrik Levkowetz; Vijay Devarapalli >> Subject: Re: [nemo] Re: is global HAHA so complex? >> >> >> El 28/04/2006, a las 9:54, Pascal Thubert (pthubert) escribi=F3: >> >>> >>>>>> >>>>>> 4G telephones are multihomed by definition. besides, the end=20 >>>>>> device >>>>> [Pascal] >>>>> [Pascal] I am afraid there is some confusion here. I think we were >>>>> talking about multiple Home Networks not multiple interfaces on = the >>>>> phone. In that context, could you please elaborate why 4G phones=20= >>>>> are >>>>> multi-homed by definition? >>>>> >>>> >>>> Ok, you mean why a telephone may need multiple Home Networks? >>>> >>>> Well, i guess that if you have a 4G phone and you have multiple >>>> provider, you want to be able to be reachable through all the >>>> providers. >>>> >>>> We do agree that is a desired feature to be able to have differetn >>>> providers for different technologies, right? So that you don't have=20= >>>> to >>>> rely on a single provider giving UMTS, WIFI services for instance. >>>> >>>> In this scenario, you probably also want to have a stable address=20= >>>> from >>>> each of the providers, so that you can by pass any of the=20 >>>> providers. I >>>> mean if you are going to have a communication through the WiFI >>>> interface, you don't want to have to exchange the initial packets=20= >>>> and >>>> the HoTI packets through the UMTS interface, right? >>>> >>>> So i really think that it is a very useful feature that 4G phones=20= >>>> hame >>>> multiple Home Addresses associated to different providers and >>>> different >>>> technologies >>>> >>>> makes sense to you? >>>> >>> [Pascal] May happen... for some geeks who can tweak their phones. >> >> I don't know.. i would expect that if i pay my mobile phone, this is >> supported wihtout tweaks (of course this would be differetn if the >> operator is subsidizing the phone) >> >>> That would not be the normal case for my grand ma. And I would not=20= >>> base >>> any RO on this. Again redundancy but no RO (and a lot of hidden >>> complexities). >> >> >> >>> >>> When my phone number is an IP address, >> >> not really i would expect that for IP phones my phone number is a = URI, >> FQDN or something like this that is easier to remember and it can >> easily be mapped to multiple IP addresses >> >>> I want to be always reachable at >>> that IP address. So providing a real redundancy to Home makes sense, >>> before you enter complex solutions based on presence or SHIM6 to >>> duplicate everything on the clients. >>> >> >> well, that is idea in putting intelegence in the edges and not in the >> core, which seems to be the basis for scalability. I guess bottom = line >> disucssion here is about design principles. Of course that having the >> clients to perform the functions imply additional complexities in the >> clients, but making the network providing the fucntions, adds >> compelxity to the netowrk, makes it less scalable and makes it more >> rigid. >> >> regards, marcelo >> >>> Pascal >>> >