From prvs=337a41da8=acee@redback.com Fri Apr 3 07:30:30 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id E24A63A68A3 for ; Fri, 3 Apr 2009 07:30:30 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.446 X-Spam-Level: X-Spam-Status: No, score=-2.446 tagged_above=-999 required=5 tests=[AWL=0.153, BAYES_00=-2.599] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id y4qhZRvPnfpC for ; Fri, 3 Apr 2009 07:30:29 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id DF0A83A679C for ; Fri, 3 Apr 2009 07:30:29 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.39,319,1235980800"; d="scan'208";a="691084" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 03 Apr 2009 07:31:32 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id C1751211B51; Fri, 3 Apr 2009 07:31:32 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 23187-08; Fri, 3 Apr 2009 07:31:32 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id AFEC5211B4E; Fri, 3 Apr 2009 07:31:31 -0700 (PDT) In-Reply-To: <030361BA-3A10-4B76-B31C-4518507EFDCA@redback.com> References: <1BEEAE11-F1A5-40DD-B4A4-54C6743B18D4@redback.com> <030361BA-3A10-4B76-B31C-4518507EFDCA@redback.com> Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Acee Lindem Date: Fri, 3 Apr 2009 10:31:30 -0400 To: Acee Lindem X-Mailer: Apple Mail (2.753.1) Cc: OSPF List Subject: Re: [OSPF] WG Last Call - Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-ospf-dynamic-hostname-02 (Corrected) X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Apr 2009 14:31:17 -0000 Ok - this time the WG last call actually has ended. A new revision addressing Dave's comment on multi-instances will be passed on to the ADs for review. Thanks, Acee On Mar 30, 2009, at 11:23 AM, Acee Lindem wrote: > The WG last call has ended. The authors are adding a statement with > respect to multiple OSPF instances and it will go to the routing ADs. > > > Thanks, > Acee > On Mar 12, 2009, at 5:56 AM, Acee Lindem wrote: > >> This starts the Working Group Last Call for the subject document. >> The target status is proposed standard. >> The WG Last Call will start today and end April 3rd at 12:00 AM >> EDT. I've added an additional week due to the upcoming IETF. If >> anyone needs more time to review the document, let Abhay or myself >> know. It is a very simple document and the last remaining issue >> (specification of hostname encoding) has been resolved to my >> satisfaction. >> >> Thanks, >> Acee >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf From prvs=337a41da8=acee@redback.com Fri Apr 3 07:41:24 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id B7BC33A6E5B; Fri, 3 Apr 2009 07:41:24 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.437 X-Spam-Level: X-Spam-Status: No, score=-2.437 tagged_above=-999 required=5 tests=[AWL=0.161, BAYES_00=-2.599, HTML_MESSAGE=0.001] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id MKOIu08GcVQl; Fri, 3 Apr 2009 07:41:24 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id 0C42D3A6D27; Fri, 3 Apr 2009 07:41:24 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.39,319,1235980800"; d="scan'208,217";a="691289" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 03 Apr 2009 07:42:26 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id DFFF86C30D; Fri, 3 Apr 2009 07:42:26 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 28720-03; Fri, 3 Apr 2009 07:42:26 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id 45CC06C30F; Fri, 3 Apr 2009 07:42:26 -0700 (PDT) Mime-Version: 1.0 (Apple Message framework v753.1) To: Ross Callon , Adrian Farrel Message-Id: <8953272F-32B4-4380-90D5-EAA1E69196EC@redback.com> Content-Type: multipart/alternative; boundary=Apple-Mail-4-617024557 From: Acee Lindem Date: Fri, 3 Apr 2009 10:42:25 -0400 X-Mailer: Apple Mail (2.753.1) Cc: OSPF List , iesg-secretary@ietf.org Subject: [OSPF] Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-ospf-dynamic-hostname-03.txt X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Apr 2009 14:41:24 -0000 --Apple-Mail-4-617024557 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Ross, Adrian, We have completed the WG last call for Dynamic Hostname Exchange Mechanism for OSPF and version draft-ietf-ospf-dynamic- hostname-03.txt addresses a couple minor comments. It is the desire of the WG that the document be published as a standards track document. Here is an html link for your convenience. http://www.ietf.org/internet-drafts/draft-ietf-ospf-dynamic- hostname-03.txt. Thanks, Acee and Abhay --Apple-Mail-4-617024557 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1

Ross, Adrian,
We have completed the WG = last call for=A0Dynamic Hostname Exchange Mechanism for OSPF and = version=A0draft-ietf-ospf-dynamic-hostname-03.txt addresses a couple = minor comments. It is the desire of the WG that the document be = published as a standards track document. Here is an html link for your = convenience.

X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: OSPF and AS External LSA Thread-Index: Acm0arCsM9MF+UVYR2qUAcXjiOlGFA== From: "Perez, Stephane" To: Subject: [OSPF] OSPF and AS External LSA X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Apr 2009 14:59:22 -0000 Hello guys, I would like to raise a generic question about the OSPF AS External LSA handling to which I have only found a workaround. I had experienced a specific corner case with the OSPF process which can cause some problems in some network environment. The scenario is a firewall cluster composed of 2 firewalls using VRRP, running. Each firewall announces a specific set of AS External routes to the OSPF with identical parameters including the same next-hop (VRRP-VIP) for L2 redundancy. In this configuration both firewalls are redistributing into the OSPF process the same external routes with the same next-hop the VRRP-VIP and the same metric. The only difference between these LSAs is the router-ID. As per OSPF standard 2328 (page 142) this will be considered as two functionally identical LSA for each external route. To avoid effort duplication, only the one with the highest router-id of the two LSA will be kept into the LSDB. Router A ---+-- Firewall A (rid ra.rb.rc.1) --+- Network ABC | vip w.x.y.z | |-- Firewall B (rid ra.rb.rc.2) --| =20 The main problem comes when the router with the router with the highest router-id dies (Firewall B). At that point the following happen: - Firewall B is marked as unreachable after the Dead interval timer - routes are removed from the Router A FIB - All External LSAs from FW-B are kept in the LSDB until the LSA age reaches the MaxAge time of 3600s - when the External LSA is being withdrawn from the LSDB the FW-A flood a new LSA and Djikstra is calculated again. In such a case this create an outage in a network for about 1 hour. There is an easy workaround which is to redistribute the Network ABC with a different metric from FW-A and FW-B, however this just comes back to a duplication of effort where we will have 2 LSAs functionally identical in the LSDB. Premature aging seems not of any help here since the router can only age an LSA which contains its own router-id. I would like to have your thought about this particular case where it can result in a 1 hour network outage. Thanks in advance Stephane Perez From rcallon@juniper.net Fri Apr 3 08:03:22 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 50EF63A6A58 for ; Fri, 3 Apr 2009 08:03:22 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.466 X-Spam-Level: X-Spam-Status: No, score=-6.466 tagged_above=-999 required=5 tests=[AWL=0.132, BAYES_00=-2.599, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0sUm7TgkUX5h for ; Fri, 3 Apr 2009 08:03:21 -0700 (PDT) Received: from exprod7og124.obsmtp.com (exprod7og124.obsmtp.com [64.18.2.26]) by core3.amsl.com (Postfix) with ESMTP id 7A2353A6BE6 for ; Fri, 3 Apr 2009 08:03:18 -0700 (PDT) Received: from source ([66.129.224.36]) (using TLSv1) by exprod7ob124.postini.com ([64.18.6.12]) with SMTP ID DSNKSdYlcF6NRk+SFWxkcPGHDpO369o5Oz09@postini.com; Fri, 03 Apr 2009 08:04:21 PDT Received: from p-emfe01-sac.jnpr.net (66.129.254.71) by P-EMHUB01-HQ.jnpr.net (172.24.192.35) with Microsoft SMTP Server id 8.1.340.0; Fri, 3 Apr 2009 08:02:47 -0700 Received: from p-emlb02-sac.jnpr.net ([66.129.254.47]) by p-emfe01-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Fri, 3 Apr 2009 08:02:47 -0700 Received: from emailwf1.jnpr.net ([10.10.2.33]) by p-emlb02-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Fri, 3 Apr 2009 08:02:47 -0700 x-mimeole: Produced By Microsoft Exchange V6.5 Content-Class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C9B46D.3ED9D379" Date: Fri, 3 Apr 2009 11:02:44 -0400 Message-ID: <3525C9833C09ED418C6FD6CD9514668C060CB487@emailwf1.jnpr.net> In-Reply-To: <8953272F-32B4-4380-90D5-EAA1E69196EC@redback.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-ospf-dynamic-hostname-03.txt Thread-Index: Acm0amn0zC7PyBEXRBmHL6A8sClMYQAAd6BQ References: <8953272F-32B4-4380-90D5-EAA1E69196EC@redback.com> From: Ross Callon To: "Acee Lindem" X-OriginalArrivalTime: 03 Apr 2009 15:02:47.0608 (UTC) FILETIME=[40114B80:01C9B46D] Cc: OSPF List , Adrian Farrel Subject: Re: [OSPF] Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-ospf-dynamic-hostname-03.txt X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Apr 2009 15:03:22 -0000 ------_=_NextPart_001_01C9B46D.3ED9D379 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable Can you send me a PROTO writeup? =20 Thanks, Ross =20 ________________________________ From: Acee Lindem [mailto:acee@redback.com]=20 Sent: 03 April 2009 10:42 To: Ross Callon; Adrian Farrel Cc: OSPF List; iesg-secretary@ietf.org Subject: Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-ospf-dynamic-hostname-03.txt =20 =20 Ross, Adrian, We have completed the WG last call for Dynamic Hostname Exchange Mechanism for OSPF and version draft-ietf-ospf-dynamic-hostname-03.txt addresses a couple minor comments. It is the desire of the WG that the document be published as a standards track document. Here is an html link for your convenience. =20 http://www.ietf.org/internet-drafts/draft-ietf-ospf-dynamic-hostname-03. txt. =20 Thanks, Acee and Abhay =20 ------_=_NextPart_001_01C9B46D.3ED9D379 Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable

Can you send me a PROTO = writeup?

 

Thanks, = Ross

 

From: Acee = Lindem [mailto:acee@redback.com]
Sent: 03 April 2009 = 10:42
To: Ross Callon; Adrian = Farrel
Cc: OSPF List; iesg-secretary@ietf.org
Subject: Dynamic Hostname = Exchange Mechanism for OSPF - = draft-ietf-ospf-dynamic-hostname-03.txt

 

 

Ross, Adrian,

We have completed the WG last call for Dynamic Hostname = Exchange Mechanism for OSPF and = version draft-ietf-ospf-dynamic-hostname-03.txt addresses a couple minor comments. It is the desire of the WG that the = document be published as a standards track document. Here is an html link for = your convenience.

 

Thanks,

Acee and Abhay

 

------_=_NextPart_001_01C9B46D.3ED9D379-- From prvs=337a41da8=acee@redback.com Fri Apr 3 08:21:12 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id AD97B28C212 for ; Fri, 3 Apr 2009 08:21:12 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.744 X-Spam-Level: X-Spam-Status: No, score=-1.744 tagged_above=-999 required=5 tests=[AWL=-0.542, BAYES_00=-2.599, HTML_MESSAGE=0.001, MIME_QP_LONG_LINE=1.396] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 5cgQ7XmSr5nb for ; Fri, 3 Apr 2009 08:21:11 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id 705D928C15E for ; Fri, 3 Apr 2009 08:21:11 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.39,320,1235980800"; d="scan'208,217";a="692220" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 03 Apr 2009 08:22:14 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id 5A8816A4691; Fri, 3 Apr 2009 08:22:14 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 07188-08; Fri, 3 Apr 2009 08:22:14 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id 5B71B6A4692; Fri, 3 Apr 2009 08:22:13 -0700 (PDT) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: multipart/alternative; boundary=Apple-Mail-6-619411596 Message-Id: <24DF32B1-A39F-4F16-87A3-5726B7BF309D@redback.com> From: Acee Lindem Date: Fri, 3 Apr 2009 11:22:12 -0400 To: "Perez, Stephane" X-Mailer: Apple Mail (2.753.1) Cc: ospf@ietf.org Subject: Re: [OSPF] OSPF and AS External LSA X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Apr 2009 15:21:12 -0000 --Apple-Mail-6-619411596 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Stephane, If one of the firewalls dies, the other should originate the suppressed As external advertisement immediately. Note the qualification "both reachable from one another" in the excerpted RFC 2328 text below. If one firewal is dead, it is clearly not reachable from the other. In figure 16, suppose instead that both RTA and RTB exchange BGP information with RTX. In this case, RTA and RTB would originate the same set of AS- external-LSAs. These LSAs, if they specify the same metric, would be functionally equivalent since they would specify the same destination and forwarding address (RTX). This leads to a clear duplication of effort. If only one of RTA or RTB originated the set of AS-external-LSAs, the routing would remain the same, and the size of the link state database would decrease. However, it must be unambiguously defined as to which router originates the LSAs (otherwise neither may, or the identity of the originator may oscillate). The following rule is thereby established: if two routers, both reachable from one another, originate functionally equivalent AS-external-LSAs (i.e., same destination, cost and non-zero forwarding address), then the LSA originated by the router having the highest OSPF Router ID is used. The router having the lower OSPF outer ID can then flush its LSA. Flushing an LSA is discussed in Section 14.1. If you don't need to run OSPF on the VRRP LAN, a workaround may be to advertise it as an OSPF external network to avoid it usage as a forwarding address. Hope this helps, Acee + On Apr 3, 2009, at 10:44 AM, Perez, Stephane wrote: > Hello guys, > > I would like to raise a generic question about the OSPF AS External > LSA > handling to which I have only found a workaround. > > I had experienced a specific corner case with the OSPF process > which can > cause some problems in some network environment. > The scenario is a firewall cluster composed of 2 firewalls using VRRP, > running. Each firewall announces a specific set of AS External > routes to > the OSPF with identical parameters including the same next-hop > (VRRP-VIP) for L2 redundancy. > > In this configuration both firewalls are redistributing into the OSPF > process the same external routes with the same next-hop the VRRP- > VIP and > the same metric. The only difference between these LSAs is the > router-ID. > > As per OSPF standard 2328 (page 142) this will be considered as two > functionally identical LSA for each external route. To avoid effort > duplication, only the one with the highest router-id of the two LSA > will > be kept into the LSDB. > > > Router A ---+-- Firewall A (rid ra.rb.rc.1) --+- Network ABC > | vip w.x.y.z | > |-- Firewall B (rid ra.rb.rc.2) --| > > > The main problem comes when the router with the router with the > highest > router-id dies (Firewall B). > > At that point the following happen: > - Firewall B is marked as unreachable after the Dead interval timer > - routes are removed from the Router A FIB > - All External LSAs from FW-B are kept in the LSDB until the LSA age > reaches the MaxAge time of 3600s > - when the External LSA is being withdrawn from the LSDB the FW-A > flood > a new LSA and Djikstra is calculated again. > > In such a case this create an outage in a network for about 1 hour. > There is an easy workaround which is to redistribute the Network ABC > with a different metric from FW-A and FW-B, however this just comes > back > to a duplication of effort where we will have 2 LSAs functionally > identical in the LSDB. > > Premature aging seems not of any help here since the router can > only age > an LSA which contains its own router-id. > > I would like to have your thought about this particular case where it > can result in a 1 hour network outage. > > Thanks in advance > Stephane Perez > > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf --Apple-Mail-6-619411596 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1 Stephane,

If one of the firewalls dies, the other = should originate the suppressed=A0
As external advertisement = immediately. Note the qualification=A0
"both reachable from = one=A0another" in the excerpted RFC 2328 text below.
If one = firewal is dead, it is clearly not=A0reachable from the = other.=A0

=A0=A0=A0In figure 16, = suppose instead that both RTA and RTB
=A0=A0 exchange = BGP information with RTX. =A0In this case,
=A0=A0 RTA = and RTB would originate the same set of AS-
=A0=A0 = external-LSAs. =A0These LSAs, if they specify the = same
=A0=A0 metric, would be functionally equivalent = since they
=A0=A0 would specify the same destination = and forwarding
=A0=A0 address (RTX). =A0This leads to a = clear duplication of
=A0=A0 effort. =A0If only one of = RTA or RTB originated the
=A0=A0 set of = AS-external-LSAs, the routing would remain
=A0=A0 the = same, and the size of the link state database
=A0=A0 = would decrease. =A0However, it must be unambiguously
=A0=A0= defined as to which router originates the LSAs
=A0=A0 = (otherwise neither may, or the identity of the
=A0=A0 = originator may oscillate). =A0The following rule is
=A0=A0= thereby established: if two routers, both = reachable
=A0=A0 from one = another, originate functionally = equivalent
=A0=A0 AS-external-LSAs (i.e., same = destination, cost and
=A0=A0 non-zero forwarding = address), then the LSA
=A0=A0 originated by the router = having the highest OSPF
=A0=A0 Router ID is used. =A0The = router having the lower OSPF
=A0=A0 outer ID can then = flush its LSA. =A0Flushing an LSA
=A0=A0 is discussed = in Section 14.1.


If = you don't need to run OSPF on the VRRP LAN, a workaround may be = to
advertise it as an OSPF external network to avoid it usage = as a forwarding
address.=A0

Hope this = helps,
Acee=A0
=A0=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 = =A0 =A0 =A0 =A0 =A0 =A0 = =A0+

=A0=A0=A0
On Apr 3, = 2009, at 10:44 AM, Perez, Stephane wrote:

Hello guys,

I would like to raise a generic = question about the OSPF AS External LSA
handling = to which I have only found a workaround.

I had = experienced a specific corner case with the OSPF process which = can
cause some problems in some = network environment.
The scenario is a firewall = cluster composed of 2 firewalls using VRRP,
the OSPF with identical parameters including = the same next-hop
(VRRP-VIP) for L2 = redundancy.

In this configuration both firewalls are = redistributing into the OSPF
process the = same external routes with the same next-hop the VRRP-VIP and
the same metric. The only difference between these = LSAs is the
router-ID.

=A0As per OSPF standard 2328 = (page 142) this will be considered as two
duplication, only the one with = the highest router-id of the two LSA will
be kept = into the LSDB.


Router A = ---+-- Firewall A=A0 (rid = ra.rb.rc.1) --+- Network ABC
=A0 =A0 =A0 =A0 =A0 =A0 |=A0 =A0 vip w.x.y.z =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 = |
=A0 =A0 =A0 =A0 =A0 =A0 |-- = Firewall B=A0 (rid = ra.rb.rc.2) --| =A0


The main = problem comes when the router with the router with the highest
router-id dies (Firewall B).

=A0At that point the following = happen:
- Firewall B is marked as = unreachable after the Dead interval timer
- routes = are removed from the Router A FIB
- All = External LSAs from FW-B are kept in the LSDB until the LSA age
reaches the MaxAge time of 3600s
- when the External LSA is being withdrawn from the = LSDB the FW-A flood
a new LSA and Djikstra is = calculated again.

=A0In = such a case this create an outage in a network for about 1 = hour.
There is an easy workaround = which is to redistribute the Network ABC
with a = different metric from FW-A and FW-B, however this just comes = back
to a duplication of effort where = we will have 2 LSAs functionally
identical in = the LSDB.

Premature aging seems not of any help here since the = router can only age
an LSA which contains its = own router-id.

I would like to have your thought about this = particular case where it
can result in = a 1 hour network outage.

Thanks in advance
Stephane Perez


OSPF mailing list
=

= --Apple-Mail-6-619411596-- From prvs=337a41da8=acee@redback.com Fri Apr 3 12:47:33 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 4CA973A691B for ; Fri, 3 Apr 2009 12:47:33 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.871 X-Spam-Level: X-Spam-Status: No, score=-1.871 tagged_above=-999 required=5 tests=[AWL=-0.383, BAYES_00=-2.599, HTML_MESSAGE=0.001, MPART_ALT_DIFF_COUNT=1.11] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pMorN-RexcEO for ; Fri, 3 Apr 2009 12:47:31 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id 74CCB3A6843 for ; Fri, 3 Apr 2009 12:47:31 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.39,321,1235980800"; d="txt'?scan'208,217";a="698555" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 03 Apr 2009 12:48:34 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id 7EC8717BA9A; Fri, 3 Apr 2009 12:48:34 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 32475-10; Fri, 3 Apr 2009 12:48:34 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id 226D917BA94; Fri, 3 Apr 2009 12:48:33 -0700 (PDT) In-Reply-To: <3525C9833C09ED418C6FD6CD9514668C060CB487@emailwf1.jnpr.net> References: <8953272F-32B4-4380-90D5-EAA1E69196EC@redback.com> <3525C9833C09ED418C6FD6CD9514668C060CB487@emailwf1.jnpr.net> Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: multipart/alternative; boundary=Apple-Mail-10-635391243 Message-Id: From: Acee Lindem Date: Fri, 3 Apr 2009 15:48:32 -0400 To: Ross Callon X-Mailer: Apple Mail (2.753.1) Cc: OSPF List , Adrian Farrel Subject: Re: [OSPF] Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-ospf-dynamic-hostname-03.txt X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Apr 2009 19:47:33 -0000 --Apple-Mail-10-635391243 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=UTF-8; delsp=yes; format=flowed Right - see attached. Thanks, Acee =EF=BF=BC On Apr 3, 2009, at 11:02 AM, Ross Callon wrote: > Can you send me a PROTO writeup? > > Thanks, Ross > > From: Acee Lindem [mailto:acee@redback.com] > Sent: 03 April 2009 10:42 > To: Ross Callon; Adrian Farrel > Cc: OSPF List; iesg-secretary@ietf.org > Subject: Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-=20 > ospf-dynamic-hostname-03.txt > > > Ross, Adrian, > We have completed the WG last call for Dynamic Hostname Exchange =20 > Mechanism for OSPF and version draft-ietf-ospf-dynamic-=20 > hostname-03.txt addresses a couple minor comments. It is the desire =20= > of the WG that the document be published as a standards track =20 > document. Here is an html link for your convenience. > > http://www.ietf.org/internet-drafts/draft-ietf-ospf-dynamic-=20 > hostname-03.txt. > > Thanks, > Acee and Abhay > --Apple-Mail-10-635391243 Content-Type: multipart/mixed; boundary=Apple-Mail-11-635391244 --Apple-Mail-11-635391244 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1 Right - see = attached.
Thanks,
Acee
=A0
= --Apple-Mail-11-635391244 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; x-unix-mode=0644; name=draft-ietf-ospf-dynamic-hostname-03.txt Content-Disposition: attachment; filename=draft-ietf-ospf-dynamic-hostname-03.txt =0D =0D =0D OSPF WG S. Venkata=0D= Internet-Draft Google Inc.=0D= Intended status: Standards Track S. Harwani=0D= Expires: October 5, 2009 C. Pignataro=0D= Cisco Systems=0D= D. McPherson=0D= Arbor Networks, Inc.=0D= April 3, 2009=0D= =0D =0D Dynamic Hostname Exchange Mechanism for OSPF=0D draft-ietf-ospf-dynamic-hostname-03=0D =0D Status of this Memo=0D =0D This Internet-Draft is submitted to IETF in full conformance with the=0D= provisions of BCP 78 and BCP 79. This document may contain material=0D= from IETF Documents or IETF Contributions published or made publicly=0D= available before November 10, 2008. The person(s) controlling the=0D copyright in some of this material may not have granted the IETF=0D Trust the right to allow modifications of such material outside the=0D= IETF Standards Process. Without obtaining an adequate license from=0D= the person(s) controlling the copyright in such materials, this=0D document may not be modified outside the IETF Standards Process, and=0D= derivative works of it may not be created outside the IETF Standards=0D= Process, except to format it for publication as an RFC or to=0D translate it into languages other than English.=0D =0D Internet-Drafts are working documents of the Internet Engineering=0D Task Force (IETF), its areas, and its working groups. Note that=0D other groups may also distribute working documents as Internet-=0D Drafts.=0D =0D Internet-Drafts are draft documents valid for a maximum of six months=0D= and may be updated, replaced, or obsoleted by other documents at any=0D= time. It is inappropriate to use Internet-Drafts as reference=0D material or to cite them other than as "work in progress."=0D =0D The list of current Internet-Drafts can be accessed at=0D http://www.ietf.org/ietf/1id-abstracts.txt.=0D =0D The list of Internet-Draft Shadow Directories can be accessed at=0D http://www.ietf.org/shadow.html.=0D =0D This Internet-Draft will expire on October 5, 2009.=0D =0D Copyright Notice=0D =0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 1]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D Copyright (c) 2009 IETF Trust and the persons identified as the=0D document authors. All rights reserved.=0D =0D This document is subject to BCP 78 and the IETF Trust's Legal=0D Provisions Relating to IETF Documents in effect on the date of=0D publication of this document (http://trustee.ietf.org/license-info).=0D= Please review these documents carefully, as they describe your rights=0D= and restrictions with respect to this document.=0D =0D Abstract=0D =0D Currently, there does not exist a simple and dynamic mechanism for=0D routers running OSPF to learn about symbolic hostnames just like for=0D= routers running IS-IS. This document defines a new OSPF Router=0D Information (RI) TLV which allows the OSPF routers to flood their=0D hostname-to-Router ID mapping information across the OSPF network.=0D This mechanism is applicable to both OSPFv2 and OSPFv3.=0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 2]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D Table of Contents=0D =0D 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4=0D= 1.1. Specification of Requirements . . . . . . . . . . . . . . . 4=0D= 2. Possible solutions . . . . . . . . . . . . . . . . . . . . . . 4=0D= 3. Implementation . . . . . . . . . . . . . . . . . . . . . . . . 5=0D= 3.1. Dynamic Hostname TLV . . . . . . . . . . . . . . . . . . . 5=0D= 3.1.1. Flooding Scope . . . . . . . . . . . . . . . . . . . . 6=0D= 3.1.2. Multiple OSPF Instances . . . . . . . . . . . . . . . . 7=0D= 4. IPv6 Considerations . . . . . . . . . . . . . . . . . . . . . . 7=0D= 5. Security Considerations . . . . . . . . . . . . . . . . . . . . 7=0D= 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . . 8=0D= 7. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . 8=0D= 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 8=0D= 8.1. Normative References . . . . . . . . . . . . . . . . . . . 8=0D= 8.2. Informative References . . . . . . . . . . . . . . . . . . 9=0D= Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 9=0D= =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 3]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D 1. Introduction=0D =0D OSPF uses a 32-bit Router ID to uniquely represent and identify a=0D node in the network. For management and operational reasons, network=0D= operators need to check the status of OSPF adjacencies, entries in=0D the routing table and the content of the OSPF link state database.=0D It is obvious that, when looking at diagnostic information, numerical=0D= representations of Router IDs (e.g., dotted-decimal or hexadecimal=0D representations) are less clear to humans than symbolic names.=0D =0D One way to overcome this problem is to define a hostname-to-Router ID=0D= mapping table on a router. This mapping can be used bidirectionally=0D= (e.g., to find symbolic names for Router IDs, and to find Router IDs=0D= for symbolic names) or unidirectionally (e.g., to find symbolic=0D hostnames for Router IDs). Thus every router has to maintain a table=0D= with mappings between router names and Router IDs.=0D =0D These tables need to contain all names and Router IDs of all routers=0D= in the network. If these mapping tables are built by static=0D definitions, it can become a manual and tedious process in=0D operational networks currently; modifying these static mapping=0D entries when additions, deletions or changes occur becomes a non-=0D scalable process very prone to error.=0D =0D This document analyzes possible solutions to this problem (see=0D Section 2) and provides a way to populate tables by defining a new=0D OSPF Router Information TLV for OSPF, the Dynamic Hostname TLV (see=0D= Section 3). This mechanism is applicable to both OSPFv2 and OSPFv3.=0D= =0D 1.1. Specification of Requirements=0D =0D The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",=0D= "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this=0D= document are to be interpreted as described in [RFC2119].=0D =0D =0D 2. Possible solutions=0D =0D There are various approaches to providing a name-to-Router ID mapping=0D= service.=0D =0D One way to build this table of mappings is by static definitions.=0D The problem with static definitions is that the network administrator=0D= needs to keep updating the mapping entries manually as the network=0D changes; this approach does not scale as the network grows, since=0D there needs to be an entry in the mapping table for each and every=0D router in the network, on every router in the network. Thus, this=0D approach greatly suffers from maintainability and scalability=0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 4]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D considerations.=0D =0D Another approach is having a centralized location where the name-to-=0D= Router ID mapping can be kept. DNS can be used for the same. A=0D disadvantage with this centralized solution is that its a single=0D point of failure; and although enhanced availability of the central=0D= mapping service can be designed, it may not be able to resolve the=0D hostname in the event of reachability or network problems. Also, the=0D= response time can be an issue with the centralized solution, which=0D can be particularly problematic in times of problem resolution. If=0D= DNS is used as the centralized mapping table, a network operator may=0D= desire a different name mapping than the existing in the DNS, or new=0D= routers may not yet be in DNS.=0D =0D Additionally for OSPFv3, in native IPv6 deployments, the 32-bit=0D Router ID value will not map to IPv4-addressed entities in the=0D network, nor will it be DNS resolvable (see Section 4).=0D =0D The third solution that we have defined in this document is to make=0D= use of the protocol itself to carry the name-to-Router ID mapping in=0D= a TLV. Routers that understand this TLV can use it to create the=0D symbolic name-to-Router ID mapping and Routers who don't understand=0D= can simply ignore it. This specification provides these semantics=0D and mapping mechanisms for OSPFv2 and OSPFv3, leveraging the OSPF=0D Router Information (RI) LSA ([RFC4970]).=0D =0D =0D 3. Implementation=0D =0D This extension makes use of the Router Information (RI) Opaque LSA=0D defined in [RFC4970] for both OSPFv2 and OSPFv3, by defining a new=0D OSPF Router Information (RI) TLV: The Dynamic Hostname TLV.=0D =0D The Dynamic Hostname TLV (see Section 3.1) is OPTIONAL. Upon receipt=0D= of the TLV a router may decide to ignore this TLV, or to install the=0D= symbolic name and Router ID in its hostname mapping table.=0D =0D 3.1. Dynamic Hostname TLV=0D =0D The format of Dynamic Hostname TLV is as follows:=0D =0D 0 1 2 3=0D 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1=0D +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+=0D | Type | Length |=0D +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+=0D | Hostname ... |=0D +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+=0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 5]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D Type Dynamic Hostname TLV Type (TBD, see Section 6)=0D =0D Length Total length of the hostname (value field) in octets, not=0D including the optional padding.=0D =0D Value Hostname, a string of 1 to 255 octets, padded to 4-octet=0D alignment, encoded in the US-ASCII charset.=0D =0D Routers that do not recognize the Dynamic Hostname TLV Type, ignore=0D= the TLV (see [RFC4970]).=0D =0D The value field identifies the symbolic hostname of the router=0D originating the LSA. The string is not null-terminated. The Router=0D= ID of this router is derived from the LSA header, in the Advertising=0D= Router field of the Router Information (RI) Opaque LSA.=0D =0D The Value field is encoded in 7-bit ASCII. If a user-interface for=0D= configuring or displaying this field permits Unicode characters, that=0D= user-interface is responsible for applying the ToASCII and/or=0D ToUnicode algorithm as described in [RFC3490] to achieve the correct=0D= format for transmission or display.=0D =0D The Dynamic Hostname TLV is applicable to both OSPFv2 and OSPFv3.=0D =0D 3.1.1. Flooding Scope=0D =0D The Dynamic Hostname TLV MAY be advertised within an area-local or AS=0D= scope Router Information (RI) LSA. But the Dynamic Hostname TLV=0D SHOULD NOT be advertised into an area in more than one RI LSA=0D irrespective of the scope of the LSA.=0D =0D In other words, if a router originates a Dynamic Hostname TLV with an=0D= IGP domain (AS) flooding scope, it SHOULD NOT send area-scoped=0D Dynamic Hostname TLV except into any attached Not-So-Stubby Area=0D (NSSA) area(s). Similarly, if a router originates area-scoped=0D Dynamic Hostname TLV (other than NSSA area scoped), it SHOULD NOT=0D send AS-scoped Dynamic Hostname TLV. When the Dynamic Hostname TLV=0D= is advertised in more than one LSA (e.g., multiple area-scoped LSAs,=0D= or AS-scoped LSAs plus NSSA area-scope LSA(s)), the hostname SHOULD=0D= be the same.=0D =0D If a router is advertising any AS scope LSA (other than Dynamic TLV=0D= RI LSA) such router SHOULD advertise Dynamic TLV RI LSA in AS scope.=0D= Otherwise, it SHOULD advertise Dynamic TLV RI LSA in area scope. For=0D= example, an AS boundary router (ASBR) SHOULD send an AS scope Dynamic=0D= Hostname TLV, whereas area boundary router (ABRs) and internal=0D routers SHOULD send an area scope Dynamic Hostname TLV.=0D =0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 6]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D The flooding scope is controlled by the Opaque LSA type in OSPFv2 and=0D= by the S1 and S2 bits in OSPFv3. For area scope, the Dynamic=0D Hostname TLV MUST be carried within an OSPFv2 Type 10 RI LSA or an=0D OSPFv3 RI LSA with the S1 bit set and S2 bit clear. If the flooding=0D= scope is the entire routing domain (AS scope), the Dynamic Hostname=0D= TLV MUST be carried within an OSPFv2 Type 11 RI LSA or OSPFv3 RI LSA=0D= with the S1 bit clear and the S2 bit set.=0D =0D 3.1.2. Multiple OSPF Instances=0D =0D When an OSPF Router Information (RI) LSA, including the Dynamic=0D Hostname TLV, is advertised in multiple OSPF instances, the hostname=0D= SHOULD either be preserved, or include a common base element. It may=0D= be useful for debugging or other purposes to assign separate=0D instances different hostnames with a consistent set of suffixes or=0D prefixes that can be associated with a specific instance. In=0D particular, when an instance is used for a discrete address family or=0D= non-routing information.=0D =0D =0D 4. IPv6 Considerations=0D =0D Both OSPFv2 and OSPFv3 employ Router IDs with a common size of 32-=0D bits. In IPv4 the Router ID values were typically derived=0D automatically from an IPv4 address configured on a loopback or=0D physical interface defined on the local system, or explicitly defined=0D= within the OSPF process configuration. With broader deployment of=0D IPv6, it's quite likely that OSPF networks will exist that have no=0D native IPv4 addressed interfaces. As a result, a 32-bit OSPF Router=0D= ID will either need to be explicitly specified, or derived in some=0D automatic manner that avoids collisions with other OSPF routers=0D within the local routing domain.=0D =0D Because this 32-bit value will not map to IPv4-addressed entities in=0D= the network, nor will it be DNS resolvable, it is considered=0D extremely desirable from an operational perspective that some=0D mechanism exist to map OSPF Router IDs to more easily interpreted=0D values, ideally, human-readable strings. This specification enables=0D= a mapping functionality which eases operational burdens that may=0D otherwise be introduced with native deployment of IPv6.=0D =0D =0D 5. Security Considerations=0D =0D Since the hostname-to-Router ID mapping relies on information=0D provided by the routers themselves, a misconfigured or compromised=0D router can inject false mapping information. Thus, this information=0D= needs to be treated with suspicion when, for example, doing=0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 7]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D diagnostics about a suspected security incident.=0D =0D This document raises no other new security issues for OSPF. Security=0D= considerations for the base OSPF protocol are covered in [RFC2328]=0D and [RFC5340]. The use of authentication for the OSPF routing=0D protocols is encouraged.=0D =0D =0D 6. IANA Considerations=0D =0D IANA maintains the "OSPF Router Information (RI) TLVs" registry=0D reachable at [IANA-RI]. An additional OSPF Router Information TLV=0D Type is defined in Section 3. It is required to be assigned by IANA=0D= from the Standards Action allocation range [RFC4970].=0D =0D =0D Registry Name: OSPF Router Information (RI) TLVs=0D =0D Type Value Capabilities Reference=0D ----------- -------------------------------------- ---------=0D TBD OSPF Dynamic Hostname This document=0D =0D =0D 7. Acknowledgments=0D =0D The authors of this document do not make any claims on the=0D originality of the ideas described. This document adapts format and=0D= text from similar work done in IS-IS [RFC2763]; we would like to=0D thank Naiming Shen and and Henk Smit, authors of that document.=0D =0D The authors would also like to thank Acee Lindem, Abhay Roy, Anton=0D Smirnov, and Dave Ward for their valuable comments and suggestions.=0D= =0D =0D 8. References=0D =0D 8.1. Normative References=0D =0D [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate=0D Requirement Levels", BCP 14, RFC 2119, March 1997.=0D =0D [RFC4970] Lindem, A., Shen, N., Vasseur, JP., Aggarwal, R., and S.=0D= Shaffer, "Extensions to OSPF for Advertising Optional=0D Router Capabilities", RFC 4970, July 2007.=0D =0D =0D =0D =0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 8]=0D= =0C=0D Internet-Draft Dynamic Hostnames for OSPF April 2009=0D= =0D =0D 8.2. Informative References=0D =0D [IANA-RI] Internet Assigned Numbers Authority, "OSPFv2 Parameters",=0D= July 2008,=0D .=0D =0D [RFC2328] Moy, J., "OSPF Version 2", STD 54, RFC 2328, April 1998.=0D= =0D [RFC2763] Shen, N. and H. Smit, "Dynamic Hostname Exchange Mechanism=0D= for IS-IS", RFC 2763, February 2000.=0D =0D [RFC3490] Faltstrom, P., Hoffman, P., and A. Costello,=0D "Internationalizing Domain Names in Applications (IDNA)",=0D= RFC 3490, March 2003.=0D =0D [RFC5340] Coltun, R., Ferguson, D., Moy, J., and A. Lindem, "OSPF=0D for IPv6", RFC 5340, July 2008.=0D =0D =0D Authors' Addresses=0D =0D Subbaiah Venkata=0D Google Inc.=0D =0D Email: svenkata@google.com=0D URI: http://www.google.com=0D =0D =0D Sanjay Harwani=0D Cisco Systems=0D =0D Email: sharwani@cisco.com=0D URI: http://www.cisco.com=0D =0D =0D Carlos Pignataro=0D Cisco Systems=0D =0D Email: cpignata@cisco.com=0D URI: http://www.cisco.com=0D =0D =0D Danny McPherson=0D Arbor Networks, Inc.=0D =0D Email: danny@arbor.net=0D =0D =0D =0D =0D =0D Venkata, et al. Expires October 5, 2009 [Page 9]=0D= =0C=0D --Apple-Mail-11-635391244 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1
Can you = send me a PROTO writeup?
Thanks, Ross
=A0Acee Lindem [mailto:acee@redback.com]=A0
Sent:=A003 April 2009 = 10:42
To:=A0Ross Callon; Adrian = Farrel
Cc:=A0OSPF List; iesg-secretary@ietf.org
= Subject:=A0Dynamic Hostname Exchange = Mechanism for OSPF - = draft-ietf-ospf-dynamic-hostname-03.txt
=A0
Ross, = Adrian,
We have = completed the WG last call for=A0Dynamic Hostname Exchange Mechanism for = OSPF and version=A0draft-ietf-ospf-dynamic-hostname-03.txt addresses a = couple minor comments. It is the desire of the WG that the document be = published as a standards track document. Here is an html link for your = convenience.
References: <8953272F-32B4-4380-90D5-EAA1E69196EC@redback.com> <3525C9833C09ED418C6FD6CD9514668C060CB487@emailwf1.jnpr.net> Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: multipart/alternative; boundary=Apple-Mail-15-637837605 Message-Id: <32E914E2-B224-48A0-8E7A-9DDF359BF994@redback.com> From: Acee Lindem Date: Fri, 3 Apr 2009 16:29:18 -0400 To: Ross Callon X-Mailer: Apple Mail (2.753.1) Cc: OSPF List , Adrian Farrel Subject: Re: [OSPF] Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-ospf-dynamic-hostname-03.txt X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Apr 2009 20:28:19 -0000 --Apple-Mail-15-637837605 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=UTF-8; delsp=yes; format=flowed Here is the proto write-up. I previously attached the wrong file. Thanks, Acee =EF=BF=BC On Apr 3, 2009, at 11:02 AM, Ross Callon wrote: > Can you send me a PROTO writeup? > > Thanks, Ross > > From: Acee Lindem [mailto:acee@redback.com] > Sent: 03 April 2009 10:42 > To: Ross Callon; Adrian Farrel > Cc: OSPF List; iesg-secretary@ietf.org > Subject: Dynamic Hostname Exchange Mechanism for OSPF - draft-ietf-=20 > ospf-dynamic-hostname-03.txt > > > Ross, Adrian, > We have completed the WG last call for Dynamic Hostname Exchange =20 > Mechanism for OSPF and version draft-ietf-ospf-dynamic-=20 > hostname-03.txt addresses a couple minor comments. It is the desire =20= > of the WG that the document be published as a standards track =20 > document. Here is an html link for your convenience. > > http://www.ietf.org/internet-drafts/draft-ietf-ospf-dynamic-=20 > hostname-03.txt. > > Thanks, > Acee and Abhay > --Apple-Mail-15-637837605 Content-Type: multipart/mixed; boundary=Apple-Mail-16-637837605 --Apple-Mail-16-637837605 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1 Here is the proto write-up. I = previously attached the wrong = file.
Thanks,
Acee
=A0
= --Apple-Mail-16-637837605 Content-Transfer-Encoding: 7bit Content-Type: text/plain; x-unix-mode=0644; name=ospf-dynamic-hostname-proto-writeup.txt Content-Disposition: attachment; filename=ospf-dynamic-hostname-proto-writeup.txt 1. Have the chairs personally reviewed this version of the Internet Draft (ID), and in particular, do they believe this ID is ready to forward to the IESG for publication? Yes 2. Has the document had adequate review from both key WG members and key non-WG members? Yes - I've reviewed it myself several times. Do you have any concerns about the depth or breadth of the reviews that have been performed? No 3. Do you have concerns that the document needs more review from a particular (broader) perspective (e.g., security, operational complexity, someone familiar with AAA, etc.)? No 4. Do you have any specific concerns/issues with this document that you believe the ADs and/or IESG should be aware of? For example, perhaps you are uncomfortable with certain parts of the document, or have concerns whether there really is a need for it. In any event, if your issues have been discussed in the WG and the WG has indicated it that it still wishes to advance the document, detail those concerns in the write-up. No 5. How solid is the WG consensus behind this document? Does it represent the strong concurrence of a few individuals, with others being silent, or does the WG as a whole understand and agree with it? The draft is already implemented and deployed in the ISIS protocol. Given this experience, I see no problems with standardization. 6. Has anyone threatened an appeal or otherwise indicated extreme discontent? If so, please summarise the areas of conflict in separate email to the Responsible Area Director. No 7. Have the chairs verified that the document adheres to all of the ID Checklist items? idnits 2.11.08 tmp/draft-ietf-ospf-dynamic-hostname-03.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see http://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to http://www.ietf.org/ietf/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to http://www.ietf.org/ID-Checklist.html: ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- No issues found here. Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) -- Obsolete informational reference (is this intentional?): RFC 2763 (Obsoleted by RFC 5301) Summary: 0 errors (**), 0 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 8. Is the document split into normative and informative references? Yes Are there normative references to IDs, where the IDs are not also ready for advancement or are otherwise in an unclear state? (note here that the RFC editor will not publish an RFC with normative references to IDs, it will delay publication until all such IDs are also ready for publication as RFCs.) No 9. What is the intended status of the document? (e.g., Proposed Standard, Informational?) Proposed Standard 10. For Standards Track and BCP documents, the IESG approval announcement includes a write-up section with the following sections: * Technical Summary This draft extends OSPF to enable advertisement of a hostname for human readable identification of OSPF routrs. * Working Group Summary There is no opposition to the draft and the consensus on adding the equivalent to the ISIS feature. * Protocol Quality This is a very straight forward draft and has already been implemented in ISIS. The draft has been reviewed several times and we see no issues. --Apple-Mail-16-637837605 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1
Can you = send me a PROTO writeup?
Thanks, Ross
=A0Acee Lindem [mailto:acee@redback.com]=A0
Sent:=A003 April 2009 = 10:42
To:=A0Ross Callon; Adrian = Farrel
Cc:=A0OSPF List; iesg-secretary@ietf.org
= Subject:=A0Dynamic Hostname Exchange = Mechanism for OSPF - = draft-ietf-ospf-dynamic-hostname-03.txt
=A0
Ross, = Adrian,
We have = completed the WG last call for=A0Dynamic Hostname Exchange Mechanism for = OSPF and version=A0draft-ietf-ospf-dynamic-hostname-03.txt addresses a = couple minor comments. It is the desire of the WG that the document be = published as a standards track document. Here is an html link for your = convenience.
To: Richard Graveman , "acee@redback.com" , "akr@cisco.com" Date: Wed, 15 Apr 2009 09:58:52 +0530 Thread-Topic: [OSPF] OSPF HMAC-SHA Cryptographic Authentication Thread-Index: Acm9Qr0WGh0KQeB4RWeTuOTP0B4+1gAPr9jA Message-ID: <7C362EEF9C7896468B36C9B79200D835BB114778@INBANSXCHMBSA1.in.alcatel-lucent.com> References: <45c8c21a0904141351p1a9c6d1at2ce527e77c96a6fa@mail.gmail.com> In-Reply-To: <45c8c21a0904141351p1a9c6d1at2ce527e77c96a6fa@mail.gmail.com> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: acceptlanguage: en-US Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 X-Scanned-By: MIMEDefang 2.57 on 155.132.188.84 Cc: "ospf@ietf.org" , "adrian.farrel@huawei.com" Subject: Re: [OSPF] OSPF HMAC-SHA Cryptographic Authentication X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 15 Apr 2009 04:28:14 -0000 Hi Richard, > The last argument is: >=20 > > He [The customer] may feel cheated and may not be so happy,=20 > when he learns that there was no value add. >=20 > Good. The document can then simply say that there is no security value > in the choice. This is easy, isn't it? Works for me!I can add some text to this effect in the Security Considerati= ons if the chairs and ADs find that acceptable. >=20 > Could the WG Chairs please declare consensus and ask the editor to set > a date to get a new draft to WG LC? Yes, I think its time we do that! The first version was posted in Jan 2007 = and there has been very little change since then. It has been discussed in = the WG list and everyone seems to be okay with what we're doing here. A similar draft in ISIS WG got published as RFC 5310 some time back. Cheers, Manav >=20 > Richard Graveman > = From prvs=349b02846=acee@redback.com Wed Apr 15 03:14:36 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id E08DE3A6AA8 for ; Wed, 15 Apr 2009 03:14:36 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.405 X-Spam-Level: X-Spam-Status: No, score=-2.405 tagged_above=-999 required=5 tests=[AWL=0.194, BAYES_00=-2.599] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Ps5-u08lo8g8 for ; Wed, 15 Apr 2009 03:14:36 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id 4279A3A6917 for ; Wed, 15 Apr 2009 03:14:36 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.40,191,1239001200"; d="scan'208";a="953341" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 15 Apr 2009 03:15:46 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id AAC49EDD84; Wed, 15 Apr 2009 03:15:46 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 27616-07; Wed, 15 Apr 2009 03:15:46 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id 2758FEDD83; Wed, 15 Apr 2009 03:15:45 -0700 (PDT) Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Acee Lindem Date: Wed, 15 Apr 2009 06:15:48 -0400 To: OSPF List X-Mailer: Apple Mail (2.753.1) Subject: [OSPF] Advertising a Router's Local Addresses in OSPF TE Extensions - draft-ietf-ospf-te-node-addr-05.txt X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 15 Apr 2009 10:14:51 -0000 As promised at the last two IETFs, we are now last calling the subject document. The Working Group Last Call will start today and end at 12:00 AM on April 30th, 2009. Please send any last call comments to this list. Thanks, Acee and Abhay From prvs=349b02846=acee@redback.com Wed Apr 15 03:21:56 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 582C43A693C for ; Wed, 15 Apr 2009 03:21:56 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.414 X-Spam-Level: X-Spam-Status: No, score=-2.414 tagged_above=-999 required=5 tests=[AWL=0.185, BAYES_00=-2.599] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id QstjMN-ykpZT for ; Wed, 15 Apr 2009 03:21:55 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id 56BF63A6D8F for ; Wed, 15 Apr 2009 03:20:39 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.40,191,1239001200"; d="scan'208";a="953426" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 15 Apr 2009 03:21:51 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id 97638EDD84; Wed, 15 Apr 2009 03:21:51 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 28395-07; Wed, 15 Apr 2009 03:21:51 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id 14368EDD83; Wed, 15 Apr 2009 03:21:50 -0700 (PDT) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <2E1E1884-32FE-4CCB-AD5B-4D812ACF15DA@redback.com> Content-Transfer-Encoding: 7bit From: Acee Lindem Date: Wed, 15 Apr 2009 06:21:52 -0400 To: OSPF List X-Mailer: Apple Mail (2.753.1) Subject: Re: [OSPF] Advertising a Router's Local Addresses in OSPF TE Extensions - draft-ietf-ospf-te-node-addr-05.txt X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 15 Apr 2009 10:21:56 -0000 Actually, I rechecked my E-mail folder on this draft and we completed a WG last call on this one prior in early December, 2008. So, we need not WG last call it again. We will go ahead and pass it to the ADs for advancement. If you do have any further comments, please speak now. Thanks, Acee On Apr 15, 2009, at 6:15 AM, Acee Lindem wrote: > As promised at the last two IETFs, we are now last calling the > subject document. The Working Group Last Call will start today and > end at 12:00 AM on April 30th, 2009. Please send any last call > comments to this list. > Thanks, > Acee and Abhay_______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf From prvs=349b02846=acee@redback.com Wed Apr 15 16:24:29 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7D6E73A6BCF for ; Wed, 15 Apr 2009 16:24:29 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.119 X-Spam-Level: X-Spam-Status: No, score=-2.119 tagged_above=-999 required=5 tests=[AWL=-0.120, BAYES_00=-2.599, J_CHICKENPOX_21=0.6] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id iMlxedGUIKGA for ; Wed, 15 Apr 2009 16:24:28 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id BC9483A6BE0 for ; Wed, 15 Apr 2009 16:24:28 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.40,194,1239001200"; d="scan'208";a="970877" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 15 Apr 2009 16:25:41 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id 74AD6A263A7; Wed, 15 Apr 2009 16:25:41 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 12656-09; Wed, 15 Apr 2009 16:25:41 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id 46875A263A2; Wed, 15 Apr 2009 16:25:40 -0700 (PDT) In-Reply-To: <7C362EEF9C7896468B36C9B79200D835BB114778@INBANSXCHMBSA1.in.alcatel-lucent.com> References: <45c8c21a0904141351p1a9c6d1at2ce527e77c96a6fa@mail.gmail.com> <7C362EEF9C7896468B36C9B79200D835BB114778@INBANSXCHMBSA1.in.alcatel-lucent.com> Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <7833B56B-07B2-4419-B233-53E9A31DAB7E@redback.com> Content-Transfer-Encoding: 7bit From: Acee Lindem Date: Wed, 15 Apr 2009 19:25:42 -0400 To: "Bhatia, Manav (Manav)" X-Mailer: Apple Mail (2.753.1) Cc: OSPF List , Adrian Farrel , Richard Graveman Subject: Re: [OSPF] OSPF HMAC-SHA Cryptographic Authentication X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 15 Apr 2009 23:24:29 -0000 In Minneapolis, Steve Blake presented Ran Atkinson's slides on the requirement to extend the HMAC-SHAx calculation to include a sequence of Apad bytes equal to the length of the digest (e.g., 20 bytes of Apad for HMAC-SHA1). Some of the main arguments were 1) customer preference 2) Consistency with ISIS and RIPv2 HMAC-SHAx (yes, I'm aware that it is not exactly the same case since OSPF doesn't have the HMAC-SHAx digest embedded in OSPF packet). At least in the room, the consensus was that this was a good idea. After the meeting, we had further discussions on this and I thought we were all ready to make the change and incorporate this into the WG document. I recall one issue was the number of authors but I believe we can handle that off list. Hence, my read of the WG is that we just go forward with this decision rather than continuing with the debate. Thanks, Acee On Apr 15, 2009, at 12:28 AM, Bhatia, Manav (Manav) wrote: > Hi Richard, > >> The last argument is: >> >>> He [The customer] may feel cheated and may not be so happy, >> when he learns that there was no value add. >> >> Good. The document can then simply say that there is no security >> value >> in the choice. This is easy, isn't it? > > Works for me!I can add some text to this effect in the Security > Considerations if the chairs and ADs find that acceptable. > >> >> Could the WG Chairs please declare consensus and ask the editor to >> set >> a date to get a new draft to WG LC? > > Yes, I think its time we do that! The first version was posted in > Jan 2007 and there has been very little change since then. It has > been discussed in the WG list and everyone seems to be okay with > what we're doing here. > > A similar draft in ISIS WG got published as RFC 5310 some time back. > > Cheers, Manav > >> >> Richard Graveman From DJOYAL@nortel.com Sun Apr 19 09:16:16 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id F3A3D28C284 for ; Sun, 19 Apr 2009 09:16:15 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -4.104 X-Spam-Level: X-Spam-Status: No, score=-4.104 tagged_above=-999 required=5 tests=[AWL=0.405, BAYES_05=-1.11, HTML_MESSAGE=0.001, J_CHICKENPOX_56=0.6, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id jTsAZWfDTFd3 for ; Sun, 19 Apr 2009 09:16:15 -0700 (PDT) Received: from zrtps0kp.nortel.com (zrtps0kp.nortel.com [47.140.192.56]) by core3.amsl.com (Postfix) with ESMTP id D2CDD28C272 for ; Sun, 19 Apr 2009 09:16:14 -0700 (PDT) Received: from zrtphxm1.corp.nortel.com (zrtphxm1.corp.nortel.com [47.140.202.50]) by zrtps0kp.nortel.com (Switch-2.2.6/Switch-2.2.0) with ESMTP id n3JGHQp22213; Sun, 19 Apr 2009 16:17:26 GMT X-MimeOLE: Produced By Microsoft Exchange V6.5 Content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----_=_NextPart_001_01C9C10A.52D29720" Date: Sun, 19 Apr 2009 12:17:24 -0400 Message-ID: <4B7DAC3FEFD35D4A96BDD0116990501413B25160@zrtphxm1.corp.nortel.com> X-MS-Has-Attach: X-MS-TNEF-Correlator: thread-topic: Changes to OSPFv3 MIB I-D from MIB doctor review thread-index: AcnBCNQbXz58ccuXRI2ExsnZCiy9FQ== From: "Daniel Joyal" To: Subject: [OSPF] Changes to OSPFv3 MIB I-D from MIB doctor review X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 19 Apr 2009 16:16:16 -0000 This is a multi-part message in MIME format. ------_=_NextPart_001_01C9C10A.52D29720 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable The following changes were made to draft-ietf-ospf-ospfv3-mib between revision 12 and revision 14 based on comments from the MIB doctor review. 1. Updated boilerplate to latest version. 2. Renamed ospfv3Compliance to ospfv3FullCompliance and added ospfv3ReadOnlyCompliance. 3. Expanded description of ospfv3CfgNbrTable 4. Move restrictions on InetAddress and InetAddressType to the Conformance Section. 5. Remove StorageType object from ospfv3HostTable. 6. Rename Ospfv3LsidTC to Ospfv3LsIdTC for consistency with other TCs. 7. Added DEFVAL clause to more writable objects. 8. Added more Reference clauses. 9. Changed ospfv3AreaNssaTranslatorStabInt to ospfv3AreaNssaTranslatorStabInterval. 10. Added reference to RFC 3413 for fine-grain control of notification enabling and disabling in description of ospfv3NotificationEnable object. 11. For objects with a syntax of TruthValue, added text in the Description clause to descibed what a value of true(1) means. 12. Changed objects with syntax of RowStatus to have RowStatus in the object name. 13. Changed ospfv3AdminStat to ospfv3AdminStatus. 14. Some object's syntax changed from signed integer to unsigned integer. 15. Removed last sentence in Description clause for ospfv3PacketSrc object. 16. Use DISPLAY-HINT of "d" rather than "d-0" throughout the MIB. 17. In graceful restart objects, changed "ExitRc" to "ExitReason". -Dan ------_=_NextPart_001_01C9C10A.52D29720 Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable Changes to OSPFv3 MIB I-D from MIB doctor review

The following changes were made to = draft-ietf-ospf-ospfv3-mib
between revision 12 and revision 14 = based on comments
from the MIB doctor review.

1. Updated boilerplate to latest = version.
2. Renamed ospfv3Compliance to = ospfv3FullCompliance
    and added = ospfv3ReadOnlyCompliance.
3. Expanded description of = ospfv3CfgNbrTable
4. Move restrictions on InetAddress = and InetAddressType
    to the Conformance = Section.
5. Remove StorageType object from = ospfv3HostTable.
6. Rename Ospfv3LsidTC to Ospfv3LsIdTC = for
    consistency with = other TCs.
7. Added DEFVAL clause to more = writable objects.
8. Added more Reference = clauses.
9. Changed = ospfv3AreaNssaTranslatorStabInt to
    = ospfv3AreaNssaTranslatorStabInterval.
10. Added reference to RFC 3413 for = fine-grain
     control of = notification enabling and disabling
     in = description of ospfv3NotificationEnable object.
11. For objects with a syntax of = TruthValue, added
     text in the = Description clause to descibed what
     a value of = true(1) means.
12. Changed objects with syntax of = RowStatus to
     have = RowStatus in the object name.
13. Changed ospfv3AdminStat to = ospfv3AdminStatus.
14. Some object's syntax changed from = signed integer
      to = unsigned integer.
15. Removed last sentence in = Description clause for
      = ospfv3PacketSrc object.
16. Use DISPLAY-HINT of "d" = rather than "d-0" throughout
      the = MIB.
17. In graceful restart objects, = changed "ExitRc" to
      = "ExitReason".

-Dan

------_=_NextPart_001_01C9C10A.52D29720-- From prvs=3548f8c5e=acee@redback.com Mon Apr 20 05:57:33 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 81E983A6880 for ; Mon, 20 Apr 2009 05:57:33 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.115 X-Spam-Level: X-Spam-Status: No, score=-2.115 tagged_above=-999 required=5 tests=[AWL=-0.117, BAYES_00=-2.599, HTML_MESSAGE=0.001, J_CHICKENPOX_56=0.6] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 9LDZko62Olpv for ; Mon, 20 Apr 2009 05:57:32 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id 9EE313A6359 for ; Mon, 20 Apr 2009 05:57:32 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.40,217,1239001200"; d="scan'208,217";a="1064310" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 20 Apr 2009 05:58:48 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id 60E452667F2; Mon, 20 Apr 2009 05:58:48 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 29243-05; Mon, 20 Apr 2009 05:58:48 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id F03A42667F1; Mon, 20 Apr 2009 05:58:46 -0700 (PDT) In-Reply-To: <4B7DAC3FEFD35D4A96BDD0116990501413B25160@zrtphxm1.corp.nortel.com> References: <4B7DAC3FEFD35D4A96BDD0116990501413B25160@zrtphxm1.corp.nortel.com> Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: multipart/alternative; boundary=Apple-Mail-63--67876907 Message-Id: <21F35F03-53CB-4B0F-9397-685BC4BAB1D7@redback.com> From: Acee Lindem Date: Mon, 20 Apr 2009 08:58:48 -0400 To: Daniel Joyal X-Mailer: Apple Mail (2.753.1) Cc: OSPF List Subject: Re: [OSPF] Changes to OSPFv3 MIB I-D from MIB doctor review X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 20 Apr 2009 12:57:46 -0000 --Apple-Mail-63--67876907 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed OSPF WG, Dan has provided a list of changes to the OSPFv3 MIB during the MIB doctor review. Dave, Abhay, and I decided on a short review period for these changes before we move forward. Given the amount of work and review we've put into this document, we're going to be very careful as to what comments we accept. Please post any comments on the latest version (-14) to this list by Tuesday, April 28th, 2009. Thanks Joan, Dan, and Vishwas for all the hard work during the MIB doctor review, Acee On Apr 19, 2009, at 12:17 PM, Daniel Joyal wrote: > The following changes were made to draft-ietf-ospf-ospfv3-mib > between revision 12 and revision 14 based on comments > from the MIB doctor review. > > 1. Updated boilerplate to latest version. > 2. Renamed ospfv3Compliance to ospfv3FullCompliance > and added ospfv3ReadOnlyCompliance. > 3. Expanded description of ospfv3CfgNbrTable > 4. Move restrictions on InetAddress and InetAddressType > to the Conformance Section. > 5. Remove StorageType object from ospfv3HostTable. > 6. Rename Ospfv3LsidTC to Ospfv3LsIdTC for > consistency with other TCs. > 7. Added DEFVAL clause to more writable objects. > 8. Added more Reference clauses. > 9. Changed ospfv3AreaNssaTranslatorStabInt to > ospfv3AreaNssaTranslatorStabInterval. > 10. Added reference to RFC 3413 for fine-grain > control of notification enabling and disabling > in description of ospfv3NotificationEnable object. > 11. For objects with a syntax of TruthValue, added > text in the Description clause to descibed what > a value of true(1) means. > 12. Changed objects with syntax of RowStatus to > have RowStatus in the object name. > 13. Changed ospfv3AdminStat to ospfv3AdminStatus. > 14. Some object's syntax changed from signed integer > to unsigned integer. > 15. Removed last sentence in Description clause for > ospfv3PacketSrc object. > 16. Use DISPLAY-HINT of "d" rather than "d-0" throughout > the MIB. > 17. In graceful restart objects, changed "ExitRc" to > "ExitReason". > > -Dan > --Apple-Mail-63--67876907 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1
OSPF WG,=A0
Dan has provided a list of changes to the = OSPFv3 MIB during the MIB doctor review. Dave, Abhay, and I decided on a = short review period for these changes before we move forward. Given the = amount of work and review we've put into this document, we're going to = be very careful as to what comments we accept. Please post any comments = on the latest version (-14) to this list by Tuesday, April 28th, = 2009.=A0

Thanks Joan, Dan, and Vishwas for all the = hard work during the MIB doctor review,
Acee=A0

On = Apr 19, 2009, at 12:17 PM, Daniel Joyal wrote:

The = following changes were made to draft-ietf-ospf-ospfv3-mib =
between revision 12 and revision 14 = based on comments
from the = MIB doctor review.

1. = Updated boilerplate to latest version.
2. Renamed ospfv3Compliance to = ospfv3FullCompliance
=A0=A0=A0 = and added ospfv3ReadOnlyCompliance.
3. Expanded description of ospfv3CfgNbrTable =
4. Move restrictions on InetAddress = and InetAddressType
=A0=A0=A0 = to the Conformance Section.
5. = Remove StorageType object from ospfv3HostTable.
6. Rename Ospfv3LsidTC to Ospfv3LsIdTC = for
=A0=A0=A0 consistency = with other TCs.
7. Added = DEFVAL clause to more writable objects.
8. Added more Reference clauses.
9. Changed ospfv3AreaNssaTranslatorStabInt = to
=A0=A0=A0 = ospfv3AreaNssaTranslatorStabInterval.
10. Added reference to RFC 3413 for fine-grain =
=A0=A0=A0=A0 control of notification = enabling and disabling
=A0=A0=A0= =A0 in description of ospfv3NotificationEnable object.
11. For objects with a syntax of TruthValue, = added
=A0=A0=A0=A0 text in = the Description clause to descibed what
=A0=A0=A0=A0 a value of true(1) means.
12. Changed objects with syntax of RowStatus = to
=A0=A0=A0=A0 have = RowStatus in the object name.
13. Changed ospfv3AdminStat to ospfv3AdminStatus. =
14. Some object's syntax changed = from signed integer
=A0=A0=A0=A0= =A0 to unsigned integer.
15. = Removed last sentence in Description clause for
=A0=A0=A0=A0=A0 ospfv3PacketSrc object. =
16. Use DISPLAY-HINT of "d" rather = than "d-0" throughout
=A0=A0=A0= =A0=A0 the MIB.
17. In = graceful restart objects, changed "ExitRc" to
=A0=A0=A0=A0=A0 "ExitReason".

-Dan

=

= --Apple-Mail-63--67876907-- From erblichs@earthlink.net Mon Apr 20 19:09:04 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7ACCB3A6832 for ; Mon, 20 Apr 2009 19:09:04 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.609 X-Spam-Level: X-Spam-Status: No, score=-0.609 tagged_above=-999 required=5 tests=[AWL=-1.990, BAYES_05=-1.11, FRT_ESTABLISH2=2.492] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id vBFaA-eLCCiG for ; Mon, 20 Apr 2009 19:08:58 -0700 (PDT) Received: from elasmtp-kukur.atl.sa.earthlink.net (elasmtp-kukur.atl.sa.earthlink.net [209.86.89.65]) by core3.amsl.com (Postfix) with ESMTP id F372E3A6AB7 for ; Mon, 20 Apr 2009 19:08:57 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net; b=JmDQA/QJ++hm3sSlrZgyIReoM9kPl2ay0kvgMYDDeguVnr3Zeukq+5TVpLeu/nVy; h=Received:Message-Id:From:To:Content-Type:Content-Transfer-Encoding:Mime-Version:Subject:Date:X-Mailer:X-ELNK-Trace:X-Originating-IP; Received: from [98.234.127.54] (helo=[10.0.1.4]) by elasmtp-kukur.atl.sa.earthlink.net with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.67) (envelope-from ) id 1Lw5R7-0002tM-Rj for ospf@ietf.org; Mon, 20 Apr 2009 22:10:14 -0400 Message-Id: From: Mitchell Erblich To: ospf@ietf.org Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit Mime-Version: 1.0 (Apple Message framework v930.3) Date: Mon, 20 Apr 2009 19:10:12 -0700 X-Mailer: Apple Mail (2.930.3) X-ELNK-Trace: 074f60c55517ea841aa676d7e74259b7b3291a7d08dfec798d8b4b685cf8e863e524678866d95322350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 98.234.127.54 Subject: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 21 Apr 2009 02:09:04 -0000 Group, General thoughts... After a conversation about OSPF's standard database exchange and how inefficient it is when two routers have large LSDBs and them almost being the same, we concluded that the use of Checksums might significantly decrease the amount of overhead and achieve faster convergence. Moy first (I think) suggested the LSDB checksum support to identify identical LSDBs. Sometime later the MIB supported the ospfExternLsaChksumSum, ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. Thus, why not support the ability to have OSPF routers first exchange checksums with LSA counts. If the LSDBs have more than XYZ entries in their LSDBs, then the counts which are above specific thresholds then support the checksum on each X count of LSA type. Only the groups with non-matching Checksums would then go thru database exchange. Thus, before database exchange, support some form of triple that as routers have almost the same LSDBs, the checksum exchange would remove unnecessary LSA header exchange. This, IMO, would also be a big win if say a DRother was adjacent, lost adjacency, and re-exstablished it a short time later while keeping a past snapshot. It could remove its own originated LSAs and compare checksums.. In this scenario, within BMA areas, I think only DR / DRother and BDR/ DRother need to be making this Checksum exchange.. Wouldn't the exchange of this data benefit each router receiving the information and allow both routers to know the number of LSAs to be possibly exchanged? Yes, a new capability would probably have to be added. So, the first question is whether the checksums are unique enough with a count and type to determine whether that section of the LSDB matches between two routers? And then what savings would normally exist in this hybrid database exchange and is it worth the effort? Mitchell Erblich From asmirnov@cisco.com Tue Apr 21 10:28:30 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id BAB1C3A6B2D for ; Tue, 21 Apr 2009 10:28:30 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.107 X-Spam-Level: X-Spam-Status: No, score=-0.107 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, FRT_ESTABLISH2=2.492] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id UcnWQV85y88x for ; Tue, 21 Apr 2009 10:28:30 -0700 (PDT) Received: from av-tac-bru.cisco.com (odd-brew.cisco.com [144.254.15.119]) by core3.amsl.com (Postfix) with ESMTP id A2EAE3A69CF for ; Tue, 21 Apr 2009 10:28:29 -0700 (PDT) X-TACSUNS: Virus Scanned Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3LHThc3020627; Tue, 21 Apr 2009 19:29:43 +0200 (CEST) Received: from [144.254.14.196] (dhcp-peg2-vl21-144-254-14-196.cisco.com [144.254.14.196]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3LHTg3p020428; Tue, 21 Apr 2009 19:29:43 +0200 (CEST) Message-ID: <49EE0286.9070903@cisco.com> Date: Tue, 21 Apr 2009 19:29:42 +0200 From: Anton Smirnov Organization: Cisco Systems, Inc. User-Agent: Thunderbird 2.0.0.19 (X11/20081227) MIME-Version: 1.0 To: Mitchell Erblich References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: ospf@ietf.org Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 21 Apr 2009 17:28:30 -0000 Mitchell, LSDBs may be very close but they are seldom EXACTLY the same. The thing is that change in adjacency triggers LSA [re]build, so depending on timing router and/or network LSAs may change by the time adjacency restores. Also, in moderately sized modern OSPF network there is constant background noise of LSAs being refreshed. So again, chances that refreshed LSA arrived to one side but not another while adjacency is down are relatively high. These considerations make precomputed checksums over large number of LSAs rather bad for the purpose. If you want to explore venue of checksumming database exchange then I recommend to look for more flexible approach giving routers possibility to choose range of LSAs to compute checksum over. For example, if both routers support the checksum exchange then router sending DBD could specify the first, the last LSA and value of some kind of hash of LSA headers between them (LSAs in LSDB can be viewed as ordered sequence as per RFC 5243). In this way router preparing DBD packet may place start and end points of checksumming to avoid including into the range LSAs with ages around 0 and MAXAGE (because these are LSAs which are less likely to match between routers). So database exchange will consist of series of single LSA headers (the 'old' way) and large LSA sequences. Anton Mitchell Erblich wrote: > Group, > > General thoughts... > > After a conversation about OSPF's standard database exchange and > how inefficient it > is when two routers have large LSDBs and them almost being the > same, we concluded > that the use of Checksums might significantly decrease the > amount of overhead and > achieve faster convergence. > > Moy first (I think) suggested the LSDB checksum support to > identify identical > LSDBs. Sometime later the MIB supported the > ospfExternLsaChksumSum, > ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. > > Thus, why not support the ability to have OSPF routers first > exchange > checksums with LSA counts. If the LSDBs have more than XYZ > entries > in their LSDBs, then the counts which are above specific > thresholds then > support the checksum on each X count of LSA type. Only the > groups > with non-matching Checksums would then go thru database > exchange. > > Thus, before database exchange, support some form of triple > that > as routers have almost the same LSDBs, the checksum exchange > would remove unnecessary LSA header exchange. > > This, IMO, would also be a big win if say a DRother was > adjacent, lost > adjacency, and re-exstablished it a short time later while > keeping a > past snapshot. It could remove its own originated LSAs and > compare > checksums.. > > In this scenario, within BMA areas, I think only DR / > DRother and BDR/ DRother > need to be making this Checksum exchange.. > > Wouldn't the exchange of this data benefit each router > receiving the > information and allow both routers to know the number of > LSAs to > be possibly exchanged? > > Yes, a new capability would probably have to be added. > > > So, the first question is whether the checksums are unique > enough with a > count and type to determine whether that section of the LSDB > matches > between two routers? > > And then what savings would normally exist in this hybrid > database exchange > and is it worth the effort? > > > Mitchell Erblich > > > > > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf From pranjalchakravarty@yahoo.co.in Wed Apr 22 22:11:43 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id F03B33A68CD for ; Wed, 22 Apr 2009 22:11:42 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: 0 X-Spam-Level: X-Spam-Status: No, score=x tagged_above=-999 required=5 tests=[] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id XjbskPMgPzRj for ; Wed, 22 Apr 2009 22:11:42 -0700 (PDT) Received: from web8401.mail.in.yahoo.com (web8401.mail.in.yahoo.com [202.43.219.149]) by core3.amsl.com (Postfix) with SMTP id 101433A68B2 for ; Wed, 22 Apr 2009 22:11:39 -0700 (PDT) Received: (qmail 970 invoked by uid 60001); 23 Apr 2009 05:12:56 -0000 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=yahoo.co.in; s=s1024; t=1240463576; bh=tvppcgx2UnIEt1Us/JJxBB71fLfF2d5Byc/pizIfJig=; h=Message-ID:X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:MIME-Version:Content-Type; b=pJyRvnz6963FBXgtlb+/SFRnK5EDOSTq4dNc4qqVoMEw25+mHPqrd63RFxnd898rqyyIhlzOvf989Rgs2z2mngTiuMSe/XtM2qkDvKt8he6eBfYQEDqembWGz1J1/hSnqGj97cwoQS9RuJQARBd5SsSmQxJKDcUxhOKcekJym58= DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.co.in; h=Message-ID:X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:MIME-Version:Content-Type; b=zAtwlxinbErYLtUfCoMxjAMqX1lFbwno5lGZHQlfRuYa3kY8H2ijsu3kzMXg9t1HSx1BZc1MmcaRwUSeMkiiSRrlX1DQjHySIHIx9cez24HwnrfvVOQYZ+pvlLh0Ums4ouHEQ7pPM0cPHKuvjuGc6PAfmt6EtBYImu1dDH3o1RE=; Message-ID: <243380.97846.qm@web8401.mail.in.yahoo.com> X-YMail-OSG: SuYqOWMVM1mHcOpM6QI0yHFpq9RYSwXcxAFrpCfPxLjanBZb5HIvRuNCtrmRstV91hvgL0jyze57muQeTGr1aq2RIhcXNgQcQ822Bnr.pI17mpL3msJgASLqTlA994tKTKo2IFYTstC35HV3UTgp_iWGrlLsdk8OtNOxw6s3mlou14D.fKzE1vcdLD.yXWjQaw1FiI_Z Received: from [210.94.41.89] by web8401.mail.in.yahoo.com via HTTP; Thu, 23 Apr 2009 10:42:56 IST X-Mailer: YahooMailClassic/5.2.19 YahooMailWebService/0.7.289.1 Date: Thu, 23 Apr 2009 10:42:56 +0530 (IST) From: Pranjal Chakravarty To: ospf@ietf.org MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="0-793066149-1240463576=:97846" Subject: [OSPF] OSPF Multi Process ECMP Prblem X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 23 Apr 2009 05:11:43 -0000 --0-793066149-1240463576=:97846 Content-Type: multipart/alternative; boundary="0-711202833-1240463576=:97846" --0-711202833-1240463576=:97846 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Dear all, =A0 I have one doubt regarding OSPF Multi-Process. =A0 R1-------R2-------R3 =A0 In Router R1 there is a static route 10.1.1.0/24 ,I have redistribute stati= c into ospf in R1,OSPF Process is 100 =A0 In Router R3 the same static route is there ,i have redistribute static int= o ospf in Router R3 also.OSPF Process is 200 =A0 In Router R2 ,there is two process router ospf 100 and router ospf 200, Now when we do show ip route there should have=A0seen two next hop showing = for the route 10.1.1.0/24 because it is learnt from both R1 and R3 via OSPF= . But in actualy only one route is showing . =A0 My question is why it is not giving ECMP for the route though it is learnt = from two different ospf process. =A0 I am attaching one diagram and configuration for reference. =A0 Also i would be delighted if someone points to some useful link about multi= -procees ospf. =A0 Regards Pranjal=0A=0A=0A Own a website.Get an unlimited package.Pay next to no= thing.*Go to http://in.business.yahoo.com/ --0-711202833-1240463576=:97846 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable
Dear all,
 
I have one doubt regarding OSPF Multi-Process.
 
R1-------R2-------R3
 
In Router R1 there is a static route 10.1.1.0/24 ,I have redistribute = static into ospf in R1,OSPF Process is 100
 
In Router R3 the same static route is there ,i have redistribute stati= c into ospf in Router R3 also.OSPF Process is 200
 
In Router R2 ,there is two process router ospf 100 and router ospf 200= ,
Now when we do show ip route there should have seen two next hop = showing for the route 10.1.1.0/24 because it is learnt from both R1 and R3 = via OSPF.
But in actualy only one route is showing .
 
My question is why it is not giving ECMP for the route though it is le= arnt from two different ospf process.
 
I am attaching one diagram and configuration for reference.
 
Also i would be delighted if someone points to some useful link about = multi-procees ospf.
 
Regards
Pranjal

=0A=0A=0A
Own a website.Get an unlimited package.Pay next to nothing.* Click here!. --0-711202833-1240463576=:97846-- --0-793066149-1240463576=:97846 Content-Type: application/octet-stream; name=OSPF Content-Transfer-Encoding: base64 Content-Disposition: attachment; filename="OSPF Multi Process ECMP problem.vsd" 0M8R4KGxGuEAAAAAAAAAAAAAAAAAAAAAPgADAP7/CQAGAAAAAAAAAAAAAAAF AAAAAgAAAAAAAAAAEAAABQAAAAEAAAD+////AAAAAAMAAAAHAAAACAAAAFsB AADVAQAA//////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ///////////////////////9/////v////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////1IAbwBvAHQAIABF AG4AdAByAHkAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAWAAUA////////////////AAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAA/v///wAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAD///////////////8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAP// /////////////wAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA//////////// ////AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAUgBvAG8AdAAgAEUAbgB0AHIAeQAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABYABQD//////////wIAAAATGgIA AAAAAMAAAAAAAABGAAAAAAAAAAAAAAAAUDA6/tDDyQEGAAAAQAAAAAAAAABW AGkAcwBpAG8ARABvAGMAdQBtAGUAbgB0AAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAHAACAf////8EAAAA/////wAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAANYBAACweAAAAAAAAAUAUwB1AG0A bQBhAHIAeQBJAG4AZgBvAHIAbQBhAHQAaQBvAG4AAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAoAAIBAQAAAAMAAAD/////AAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAXAEAAIDwAAAAAAAABQBEAG8AYwB1AG0AZQBu AHQAUwB1AG0AbQBhAHIAeQBJAG4AZgBvAHIAbQBhAHQAaQBvAG4AAAAAAAAA AAAAADgAAgH///////////////8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAJAAAARKICAAAAAAD//////////wQAAAD9/////v////7/ ///+/////f////3///8KAAAACwAAAAwAAAANAAAADgAAAA8AAAAQAAAAEQAA ABIAAAATAAAAFAAAABUAAAAWAAAAFwAAABgAAAAZAAAAGgAAABsAAAAcAAAA HQAAAB4AAAAfAAAAIAAAACEAAAAiAAAAIwAAACQAAAAlAAAAJgAAACcAAAAo AAAAKQAAACoAAAArAAAALAAAAC0AAAAuAAAALwAAADAAAAAxAAAAMgAAADMA AAA0AAAANQAAADYAAAA3AAAAOAAAADkAAAA6AAAAOwAAADwAAAA9AAAAPgAA AD8AAABAAAAAQQAAAEIAAABDAAAARAAAAEUAAABGAAAARwAAAEgAAABJAAAA SgAAAEsAAABMAAAATQAAAE4AAABPAAAAUAAAAFEAAABSAAAAUwAAAFQAAABV AAAAVgAAAFcAAABYAAAAWQAAAFoAAABbAAAAXAAAAF0AAABeAAAAXwAAAGAA AABhAAAAYgAAAGMAAABkAAAAZQAAAGYAAABnAAAAaAAAAGkAAABqAAAAawAA AGwAAABtAAAAbgAAAG8AAABwAAAAcQAAAHIAAABzAAAAdAAAAHUAAAB2AAAA dwAAAHgAAAB5AAAAegAAAHsAAAB8AAAAfQAAAH4AAAB/AAAAgAAAAFYAaQBz AGkAbwBJAG4AZgBvAHIAbQBhAHQAaQBvAG4AAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAiAAIA////////////////AAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABwAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAD///////////////8AAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAP///////////////wAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA//// ////////////AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAA/v////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ///////////////////////////////////////+/wAABAACAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AIEAAACCAAAAgwAAAIQAAACFAAAAhgAAAIcAAACIAAAAiQAAAIoAAACLAAAA jAAAAI0AAACOAAAAjwAAAJAAAACRAAAAkgAAAJMAAACUAAAAlQAAAJYAAACX AAAAmAAAAJkAAACaAAAAmwAAAJwAAACdAAAAngAAAJ8AAACgAAAAoQAAAKIA AACjAAAApAAAAKUAAACmAAAApwAAAKgAAACpAAAAqgAAAKsAAACsAAAArQAA AK4AAACvAAAAsAAAALEAAACyAAAAswAAALQAAAC1AAAAtgAAALcAAAC4AAAA uQAAALoAAAC7AAAAvAAAAL0AAAC+AAAAvwAAAMAAAADBAAAAwgAAAMMAAADE AAAAxQAAAMYAAADHAAAAyAAAAMkAAADKAAAAywAAAMwAAADNAAAAzgAAAM8A AADQAAAA0QAAANIAAADTAAAA1AAAANUAAADWAAAA1wAAANgAAADZAAAA2gAA ANsAAADcAAAA3QAAAN4AAADfAAAA4AAAAOEAAADiAAAA4wAAAOQAAADlAAAA 5gAAAOcAAADoAAAA6QAAAOoAAADrAAAA7AAAAO0AAADuAAAA7wAAAPAAAADx AAAA8gAAAPMAAAD0AAAA9QAAAPYAAAD3AAAA+AAAAPkAAAD6AAAA+wAAAPwA AAD9AAAA/gAAAP8AAAAAAQAAAQEAAAIBAAADAQAABAEAAAUBAAAGAQAABwEA AAgBAAAJAQAACgEAAAsBAAAMAQAADQEAAA4BAAAPAQAAEAEAABEBAAASAQAA EwEAABQBAAAVAQAAFgEAABcBAAAYAQAAGQEAABoBAAAbAQAAHAEAAB0BAAAe AQAAHwEAACABAAAhAQAAIgEAACMBAAAkAQAAJQEAACYBAAAnAQAAKAEAACkB AAAqAQAAKwEAACwBAAAtAQAALgEAAC8BAAAwAQAAMQEAADIBAAAzAQAANAEA ADUBAAA2AQAANwEAADgBAAA5AQAAOgEAADsBAAA8AQAAPQEAAD4BAAA/AQAA QAEAAEEBAABCAQAAQwEAAEQBAABFAQAARgEAAEcBAABIAQAASQEAAEoBAABL AQAATAEAAE0BAABOAQAATwEAAFABAABRAQAAUgEAAFMBAABUAQAAVQEAAFYB AABXAQAAWAEAAFkBAABaAQAA/v////3///9dAQAAXgEAAF8BAABgAQAAYQEA AGIBAABjAQAAZAEAAGUBAABmAQAAZwEAAGgBAABpAQAAagEAAGsBAABsAQAA bQEAAG4BAABvAQAAcAEAAHEBAAByAQAAcwEAAHQBAAB1AQAAdgEAAHcBAAB4 AQAAeQEAAHoBAAB7AQAAfAEAAH0BAAB+AQAAfwEAAIABAAD+/wAABAACAAAA AAAAAAAAAAAAAAAAAAACAAAAAtXN1ZwuGxCTlwgAKyz5rkQAAAAF1c3VnC4b EJOXCAArLPmuJAEAAOAAAAAHAAAAAQAAAEAAAAAAAACASAAAAA4AAABQAAAA DwAAAFwAAAACAAAAbAAAAAwAAAB4AAAADQAAALAAAAACAAAAtQMAABMAAAAJ BAAAHgAAAAEAAAAAAAAAHgAAAAgAAABTQU1TVU5HAB4AAAABAAAAAAAAAAwQ AAAEAAAAHgAAAAYAAABQYWdlcwADAAAAAQAAAB4AAAAIAAAATWFzdGVycwAD AAAAAgAAAAAAHhAAAAMAAAAIAAAAUGFnZS0xAAAIAAAAUm91dGVyAAAMAAAA V1dXIENsdXN0ZXIAIKECAAYAAAAAAAAAOAAAAAEAAACMAAAAAAAAgJQAAAAC AAAAnAAAAAMAAACoAAAABAAAALQAAAADAAAABAAAABAAAABfVlBJRF9QUkVW SUVXUwA/AwAAABgAAABfVlBJRF9BTFRFUk5BVEVOQU1FUwBBdz8CAAAAEAAA AF9QSURfTElOS0JBU0UAmT8CAAAAtQMAABMAAAAJBAAAQQAAAAQAAAAAAAAA HgAAAAEAAAAAAAAADBAAAAIAAAADAAAAAAAAAEcAAABUoAIA/////w4AAAAB AAAAbAAAAOX///+UAAAAPAMAACMDAAAAAAAAAAAAAKVUAABzbQAAIEVNRgAA AQBMoAIAdgkAAAgAAAAAAAAAAAAAAAAAAAAABQAAAAQAAFIBAAAOAQAAAAAA AAAAAAAAAAAAUCgFALAeBABGAAAALAAAACAAAABFTUYrAUABABwAAAAQAAAA ARDA2wEAAAB4AAAAeAAAAEYAAACUNgAAiDYAAEVNRiswQAIAEAAAAAQAAAAA AIA/H0AEAAwAAAAAAAAAHkABAAwAAAAAAAAAHUAAABQAAAAIAAAAAQAAACUE AAAdQAAAFAAAAAgAAAACAAAAJQQAACFABwAMAAAAAAAAACVAAAAQAAAABAAA AAAAAAAsQAAAJAAAABgAAADOjdM9AAAAgAAAAIAxHHs9oL6hQ53YukMIQAAF yDUAALw1AAABEMDbAgAAAAMAAACsNQAAAQAAAGwAAAABAAAAAQAAANoAAACB AAAAAAAAAAAAAAChFgAAaQ0AACBFTUYAAAEArDUAAPoAAAAGAAAAAAAAAAAA AAAAAAAAAAUAAAAEAABSAQAADgEAAAAAAAAAAAAAAAAAAFAoBQCwHgQARgAA AEwXAAA+FwAAR0RJQwEAAIAAAwAArZ+w8wAAAAAmFwAAAQAJAAADkwsAAAYA hgEAAAAABQAAAAwCaA2gFgUAAAAMAmgNoBYHAAAAEAQBAAEAAQABAAMAAAAe AAQAAAAEAQ0ABwAAAPwCAQAAAAAAAAAEAAAALQEAAAkAAAD6AgAAAAAAAAAA AAAiAAQAAAAtAQEACQAAAPoCBgAeAAAAAAAAACIABAAAAC0BAgAEAAAALQEA AAcAAAD8AgAAv7+/AAAABAAAAC0BAwAEAAAABgEBAAYBAAAkA4EAjxb3A4sW xAOAFpEDbhZfA1UWLgM0Fv0CDhbOAuAVnwKtFXICcxVGAjMVGgLuFPEBoxTI AVMUoQH9E3sBoxNXAUQTNAHgEhMBdxLzAAoS1gCZEboAJRGgAKwQiAAwEHIA sA9eAC0PTACoDjwAHw4vAJQNIwAGDRsAdgwUAOMLEABPCw8AuwoQACgKFACY CRsACgkjAH8ILwD2BzwAcQdMAO4GXgBuBnIA8gWIAHkFoAAFBboAlATWACcE 8wC+AxMBWgM0AfsCVwGhAnsBSwKhAfsByAGwAfEBawEaAisBRgLxAHICvgCf ApAAzgJqAP0CSQAuAzAAXwMeAJEDEwDEAw8A9wMTACoEHgBdBDAAjwRJAMAE agDxBJAAIAW+AE8F8QB8BSsBqAVrAdQFsAH9BfsBJgZLAk0GoQJzBvsClwZa A7oGvgPbBicE+waUBBgHBQU0B3kFTgfyBWYHbgZ8B+4GkAdxB6IH9geyB38I vwcKCcsHmAnTBygK2ge7Ct4HTwvfB+ML3gd2DNoHBg3TB5QNywcfDr8HqA6y By0PogewD5AHMBB8B6wQZgclEU4HmRE0BwoSGAd3EvsG4BLbBkQTugajE5cG /RNzBlMUTQajFCYG7hT9BTMV1AVzFagFrRV8BeAVTwUOFiAFNBbxBFUWwARu Fo8EgBZdBIsWKgSPFvcDBAAAAC0BAAAEAAAA8AEDAAQAAAAEAQ0ABgEAACUD gQCPFvcDixbEA4AWkQNuFl8DVRYuAzQW/QIOFs4C4BWfAq0VcgJzFUYCMxUa Au4U8QGjFMgBUxShAf0TewGjE1cBRBM0AeASEwF3EvMAChLWAJkRugAlEaAA rBCIADAQcgCwD14ALQ9MAKgOPAAfDi8AlA0jAAYNGwB2DBQA4wsQAE8LDwC7 ChAAKAoUAJgJGwAKCSMAfwgvAPYHPABxB0wA7gZeAG4GcgDyBYgAeQWgAAUF ugCUBNYAJwTzAL4DEwFaAzQB+wJXAaECewFLAqEB+wHIAbAB8QFrARoCKwFG AvEAcgK+AJ8CkADOAmoA/QJJAC4DMABfAx4AkQMTAMQDDwD3AxMAKgQeAF0E MACPBEkAwARqAPEEkAAgBb4ATwXxAHwFKwGoBWsB1AWwAf0F+wEmBksCTQah AnMG+wKXBloDuga+A9sGJwT7BpQEGAcFBTQHeQVOB/IFZgduBnwH7gaQB3EH ogf2B7IHfwi/BwoJyweYCdMHKAraB7sK3gdPC98H4wveB3YM2gcGDdMHlA3L Bx8OvweoDrIHLQ+iB7APkAcwEHwHrBBmByURTgeZETQHChIYB3cS+wbgEtsG RBO6BqMTlwb9E3MGUxRNBqMUJgbuFP0FMxXUBXMVqAWtFXwF4BVPBQ4WIAU0 FvEEVRbABG4WjwSAFl0EixYqBI8W9wMEAAAABAENAAQAAAAtAQAABwAAAPwC AAC/v78AAAAEAAAALQEDAAQAAAAGAQEAhgEAACQDwQCPFvcDixYqBIAWXQRu Fo8EVBbABDQW8AQNFiAF4BVOBa0VewVzFagFMxXTBe4U/QWjFCUGUxRMBv0T cgajE5YGQxO5Bt8S2gZ3EvoGChIYB5kRNAckEU4HrBBmBy8QfAewD5AHLQ+i B6cOsgcfDr8Hkw3LBwUN0wd1DNoH4wveB08L3we7Ct4HKQraB5kJ0wcLCcsH fwi/B/cHsgdxB6IH7gaQB28GfAfyBWYHegVOBwUFNAeUBBgHJwT6Br8D2gZb A7kG+wKWBqECcgZLAkwG+wElBrAB/QVrAdMFKwGoBfEAewW+AE4FkQAgBWoA 8ARKAMAEMACPBB4AXQQTACoEDwD3Aw8A+wMPAAcEDwAaBA8AMwQPAFMEDwB4 BA8AowQPANIEDwAFBQ8APAUPAHYFDwCyBQ8A8AUPADAGDwBxBg8AswYPAPUG DwA2Bw8AdgcPALQHDwDwBw8AKggPAGEIDwCUCA8AwwgPAO4IDwATCQ8AMwkP AEwJDwBfCQ8AawkPAG8JEwCiCR4A1QkwAAcKSgA4CmoAaAqRAJgKvgDGCvEA 8worASALawFLC7ABdQv7AZ0LSwLEC6EC6gv7Ag4MWwMxDL8DUgwnBHIMlASQ DAUFrAx6BcYM8gXeDG8G9AzuBggNcQcaDfcHKg1/CDcNCwlDDZkJSw0pClIN uwpWDU8LVw3jC1YNdQxSDQUNSw2TDUMNHw43DacOKg0tDxoNsA8IDS8Q9Ays EN4MJBHGDJkRrAwKEpAMdxJyDN8SUgxDEzEMoxMODP0T6gtTFMQLoxSdC+4U dQszFUsLcxUgC60V8wrgFcYKDRaYCjQWaApUFjgKbhYHCoAW1QmLFqIJjxZv CY8WawmPFl8JjxZMCY8WMwmPFhMJjxbuCI8WwwiPFpQIjxZhCI8WKgiPFvAH jxa0B48WdgePFjYHjxb1Bo8WswaPFnEGjxYwBo8W8AWPFrIFjxZ2BY8WPAWP FgUFjxbSBI8WowSPFngEjxZTBI8WMwSPFhoEjxYHBI8W+wOPFvcDBAAAAC0B AAAEAAAA8AEDAAQAAAAEAQ0AhgEAACUDwQCPFvcDixYqBIAWXQRuFo8EVBbA BDQW8AQNFiAF4BVOBa0VewVzFagFMxXTBe4U/QWjFCUGUxRMBv0TcgajE5YG QxO5Bt8S2gZ3EvoGChIYB5kRNAckEU4HrBBmBy8QfAewD5AHLQ+iB6cOsgcf Dr8Hkw3LBwUN0wd1DNoH4wveB08L3we7Ct4HKQraB5kJ0wcLCcsHfwi/B/cH sgdxB6IH7gaQB28GfAfyBWYHegVOBwUFNAeUBBgHJwT6Br8D2gZbA7kG+wKW BqECcgZLAkwG+wElBrAB/QVrAdMFKwGoBfEAewW+AE4FkQAgBWoA8ARKAMAE MACPBB4AXQQTACoEDwD3Aw8A+wMPAAcEDwAaBA8AMwQPAFMEDwB4BA8AowQP ANIEDwAFBQ8APAUPAHYFDwCyBQ8A8AUPADAGDwBxBg8AswYPAPUGDwA2Bw8A dgcPALQHDwDwBw8AKggPAGEIDwCUCA8AwwgPAO4IDwATCQ8AMwkPAEwJDwBf CQ8AawkPAG8JEwCiCR4A1QkwAAcKSgA4CmoAaAqRAJgKvgDGCvEA8worASAL awFLC7ABdQv7AZ0LSwLEC6EC6gv7Ag4MWwMxDL8DUgwnBHIMlASQDAUFrAx6 BcYM8gXeDG8G9AzuBggNcQcaDfcHKg1/CDcNCwlDDZkJSw0pClINuwpWDU8L Vw3jC1YNdQxSDQUNSw2TDUMNHw43DacOKg0tDxoNsA8IDS8Q9AysEN4MJBHG DJkRrAwKEpAMdxJyDN8SUgxDEzEMoxMODP0T6gtTFMQLoxSdC+4UdQszFUsL cxUgC60V8wrgFcYKDRaYCjQWaApUFjgKbhYHCoAW1QmLFqIJjxZvCY8WawmP Fl8JjxZMCY8WMwmPFhMJjxbuCI8WwwiPFpQIjxZhCI8WKgiPFvAHjxa0B48W dgePFjYHjxb1Bo8WswaPFnEGjxYwBo8W8AWPFrIFjxZ2BY8WPAWPFgUFjxbS BI8WowSPFngEjxZTBI8WMwSPFhoEjxYHBI8W+wOPFvcDBAAAAAQBDQAJAAAA +gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQIABAAAAC0BAAAHAAAA/AIA AAAAAAAAAAQAAAAtAQIABAAAAAYBAQAUAAAAJAMIAJoLBwM5DY4DOhHvAQET ewIMEhgBYQ0YAUEPgQGaCwcDBAAAAC0BAAAEAAAABAENAAkAAAD6AgAABQAA AP///wAiAAQAAAAtAQQABAAAAPABAwAUAAAAJQMIAJoLBwM5DY4DOhHvAQET ewIMEhgBYQ0YAUEPgQGaCwcDBAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAE AAAALQEDAAQAAADwAQQABAAAAC0BAAAEAAAALQECAAQAAAAGAQEAFAAAACQD CACaCwcDOQ2OAzoR7wEBE3sCDBIYAWENGAFBD4EBmgsHAwQAAAAtAQAABAAA AAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAALQEEAAQAAADwAQMAFAAAACUD CACaCwcDOQ2OAzoR7wEBE3sCDBIYAWENGAFBD4EBmgsHAwQAAAAEAQ0ACQAA APoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA8AEEAAQAAAAtAQAABAAAAC0B AgAEAAAABgEBABQAAAAkAwgA4QqDBEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6 BeEKgwQEAAAALQEAAAQAAAAEAQ0ACQAAAPoCAAAFAAAA////ACIABAAAAC0B BAAEAAAA8AEDABQAAAAlAwgA4QqDBEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6 BeEKgwQEAAAABAENAAkAAAD6AgYABQAAAAAAAAAiAAQAAAAtAQMABAAAAPAB BAAEAAAALQEAAAQAAAAtAQIABAAAAAYBAQAUAAAAJAMIAOEKgwRMCfwDbgWb BX8DDwV5BIsGTAmLBjoH+gXhCoMEBAAAAC0BAAAEAAAABAENAAkAAAD6AgAA BQAAAP///wAiAAQAAAAtAQQABAAAAPABAwAUAAAAJQMIAOEKgwRMCfwDbgWb BX8DDwV5BIsGTAmLBjoH+gXhCoMEBAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAA IgAEAAAALQEDAAQAAADwAQQABAAAAC0BAAAEAAAALQECAAQAAAAGAQEAFAAA ACQDCADjA4EBjAX6AI0JewJZCw0CXwprA68FawOjBwcD4wOBAQQAAAAtAQAA BAAAAAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAALQEEAAQAAADwAQMAFAAA ACUDCADjA4EBjAX6AI0JewJZCw0CXwprA68FawOjBwcD4wOBAQQAAAAEAQ0A CQAAAPoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA8AEEAAQAAAAtAQAABAAA AC0BAgAEAAAABgEBABQAAAAkAwgA4wOBAYwF+gCNCXsCWQsNAl8KawOvBWsD owcHA+MDgQEEAAAALQEAAAQAAAAEAQ0ACQAAAPoCAAAFAAAA////ACIABAAA AC0BBAAEAAAA8AEDABQAAAAlAwgA4wOBAYwF+gCNCXsCWQsNAl8KawOvBWsD owcHA+MDgQEEAAAABAENAAkAAAD6AgYABQAAAAAAAAAiAAQAAAAtAQMABAAA APABBAAEAAAALQEAAAQAAAAtAQIABAAAAAYBAQAUAAAAJAMIALsSJwYcEa4G OQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGBAAAAC0BAAAEAAAABAENAAkAAAD6 AgAABQAAAP///wAiAAQAAAAtAQQABAAAAPABAwAUAAAAJQMIALsSJwYcEa4G OQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGBAAAAAQBDQAJAAAA+gIGAAUAAAAA AAAAIgAEAAAALQEDAAQAAADwAQQABAAAAC0BAAAEAAAALQECAAQAAAAGAQEA FAAAACQDCAC7EicGHBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxInBgQAAAAt AQAABAAAAAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAALQEEAAQAAADwAQMA FAAAACUDCAC7EicGHBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxInBgQAAAAE AQ0ACQAAAPoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA8AEEAAQAAAAtAQAA BwAAAPwCAAD///8AAAAEAAAALQEEAAQAAAAGAQEAFAAAACQDCAC9CyUDYQ2s A10RDQIfE5kCLxJAAX8NQAF9D58BvQslAwQAAAAtAQAABAAAAAQBDQAJAAAA +gIAAAUAAAD///8AIgAEAAAALQEFAAQAAADwAQMAFAAAACUDCAC9CyUDYQ2s A10RDQIfE5kCLxJAAX8NQAF9D58BvQslAwQAAAAEAQ0ACQAAAPoCBgAFAAAA AAAAACIABAAAAC0BAwAEAAAA8AEFAAQAAAAtAQAABAAAAC0BBAAEAAAABgEB ABQAAAAkAwgAvQslA2ENrANdEQ0CHxOZAi8SQAF/DUABfQ+fAb0LJQMEAAAA LQEAAAQAAAAEAQ0AFAAAACUDCAC9CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9 D58BvQslAwQAAAAEAQ0ABAAAAC0BAAAEAAAALQEEAAQAAAAGAQEAFAAAACQD CAAOC6EEbwkaBIwFuQWiAy0FnASuBm8JrgZdBycGDguhBAQAAAAtAQAABAAA AAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAALQEFAAQAAADwAQMAFAAAACUD CAAOC6EEbwkaBIwFuQWiAy0FnASuBm8JrgZdBycGDguhBAQAAAAEAQ0ACQAA APoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA8AEFAAQAAAAtAQAABAAAAC0B BAAEAAAABgEBABQAAAAkAwgADguhBG8JGgSMBbkFogMtBZwErgZvCa4GXQcn Bg4LoQQEAAAALQEAAAQAAAAEAQ0AFAAAACUDCAAOC6EEbwkaBIwFuQWiAy0F nASuBm8JrgZdBycGDguhBAQAAAAEAQ0ABAAAAC0BAAAEAAAALQEEAAQAAAAG AQEAFAAAACQDCAAaBJ8BrwUYAbAJmQJ8CysCggqOA9wFjgPBByUDGgSfAQQA AAAtAQAABAAAAAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAALQEFAAQAAADw AQMAFAAAACUDCAAaBJ8BrwUYAbAJmQJ8CysCggqOA9wFjgPBByUDGgSfAQQA AAAEAQ0ACQAAAPoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA8AEFAAQAAAAt AQAABAAAAC0BBAAEAAAABgEBABQAAAAkAwgAGgSfAa8FGAGwCZkCfAsrAoIK jgPcBY4DwQclAxoEnwEEAAAALQEAAAQAAAAEAQ0AFAAAACUDCAAaBJ8BrwUY AbAJmQJ8CysCggqOA9wFjgPBByUDGgSfAQQAAAAEAQ0ABAAAAC0BAAAEAAAA LQEEAAQAAAAGAQEAFAAAACQDCADjEkUGOhHMBmENLQV8C7kFbAxgBDoRYAQj D78E4xJFBgQAAAAtAQAABAAAAAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAA LQEFAAQAAADwAQMAFAAAACUDCADjEkUGOhHMBmENLQV8C7kFbAxgBDoRYAQj D78E4xJFBgQAAAAEAQ0ACQAAAPoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA 8AEFAAQAAAAtAQAABAAAAC0BBAAEAAAABgEBABQAAAAkAwgA4xJFBjoRzAZh DS0FfAu5BWwMYAQ6EWAEIw+/BOMSRQYEAAAALQEAAAQAAAAEAQ0AFAAAACUD CADjEkUGOhHMBmENLQV8C7kFbAxgBDoRYAQjD78E4xJFBgQAAAAEAQ0ABAAA ACcB//8eAAAAJgYPADEAQ29weXJpZ2h0IKkgMTk5NiBNaWNyb3N5c3RlbXMg RW5naW5lZXJpbmcgQ29tcGFueQAPAAAAJgYPABMAQWxsIFJpZ2h0cyBSZXNl cnZlZAAEAAAA8AEBAAMAAAAAAAAAEQAAAAwAAAAIAAAACwAAABAAAADbAAAA ggAAAAkAAAAQAAAA2wAAAIIAAAAJAAAAEAAAAKAWAABoDQAACQAAABAAAACg FgAAaA0AACAAAAAYAAAAAQAAAAEAAAABAAAAAQAAACEAAAAIAAAAFAAAAAwA AAANAAAAJQAAAAwAAAAFAACAJgAAABwAAAABAAAAAAAAAAAAAAAAAAAAAAAA ACUAAAAMAAAAAQAAACYAAAAcAAAAAgAAAAYAAAAeAAAAAAAAAAAAAAAlAAAA DAAAAAIAAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAAv7+/AAAAAAAl AAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAIAIAAAEAAAABAAAA2gAAAEwA AACBAAAAjxb3A4sWxAOAFpEDbhZfA1UWLgM0Fv0CDhbOAuAVnwKtFXICcxVG AjMVGgLuFPEBoxTIAVMUoQH9E3sBoxNXAUQTNAHgEhMBdxLzAAoS1gCZEboA JRGgAKwQiAAwEHIAsA9eAC0PTACoDjwAHw4vAJQNIwAGDRsAdgwUAOMLEABP Cw8AuwoQACgKFACYCRsACgkjAH8ILwD2BzwAcQdMAO4GXgBuBnIA8gWIAHkF oAAFBboAlATWACcE8wC+AxMBWgM0AfsCVwGhAnsBSwKhAfsByAGwAfEBawEa AisBRgLxAHICvgCfApAAzgJqAP0CSQAuAzAAXwMeAJEDEwDEAw8A9wMTACoE HgBdBDAAjwRJAMAEagDxBJAAIAW+AE8F8QB8BSsBqAVrAdQFsAH9BfsBJgZL Ak0GoQJzBvsClwZaA7oGvgPbBicE+waUBBgHBQU0B3kFTgfyBWYHbgZ8B+4G kAdxB6IH9geyB38IvwcKCcsHmAnTBygK2ge7Ct4HTwvfB+ML3gd2DNoHBg3T B5QNywcfDr8HqA6yBy0PogewD5AHMBB8B6wQZgclEU4HmRE0BwoSGAd3EvsG 4BLbBkQTugajE5cG/RNzBlMUTQajFCYG7hT9BTMV1AVzFagFrRV8BeAVTwUO FiAFNBbxBFUWwARuFo8EgBZdBIsWKgSPFvcDJQAAAAwAAAAFAACAKAAAAAwA AAADAAAAFAAAAAwAAAANAAAAVwAAACACAAABAAAAAQAAANoAAABMAAAAgQAA AI8W9wOLFsQDgBaRA24WXwNVFi4DNBb9Ag4WzgLgFZ8CrRVyAnMVRgIzFRoC 7hTxAaMUyAFTFKEB/RN7AaMTVwFEEzQB4BITAXcS8wAKEtYAmRG6ACURoACs EIgAMBByALAPXgAtD0wAqA48AB8OLwCUDSMABg0bAHYMFADjCxAATwsPALsK EAAoChQAmAkbAAoJIwB/CC8A9gc8AHEHTADuBl4AbgZyAPIFiAB5BaAABQW6 AJQE1gAnBPMAvgMTAVoDNAH7AlcBoQJ7AUsCoQH7AcgBsAHxAWsBGgIrAUYC 8QByAr4AnwKQAM4CagD9AkkALgMwAF8DHgCRAxMAxAMPAPcDEwAqBB4AXQQw AI8ESQDABGoA8QSQACAFvgBPBfEAfAUrAagFawHUBbAB/QX7ASYGSwJNBqEC cwb7ApcGWgO6Br4D2wYnBPsGlAQYBwUFNAd5BU4H8gVmB24GfAfuBpAHcQei B/YHsgd/CL8HCgnLB5gJ0wcoCtoHuwreB08L3wfjC94HdgzaBwYN0weUDcsH Hw6/B6gOsgctD6IHsA+QBzAQfAesEGYHJRFOB5kRNAcKEhgHdxL7BuAS2wZE E7oGoxOXBv0TcwZTFE0GoxQmBu4U/QUzFdQFcxWoBa0VfAXgFU8FDhYgBTQW 8QRVFsAEbhaPBIAWXQSLFioEjxb3AxQAAAAMAAAADQAAACUAAAAMAAAABQAA gCcAAAAYAAAAAwAAAAAAAAC/v78AAAAAACUAAAAMAAAAAwAAABMAAAAMAAAA AQAAAFYAAAAgAwAAAQAAACYAAADaAAAAgQAAAMEAAACPFvcDixYqBIAWXQRu Fo8EVBbABDQW8AQNFiAF4BVOBa0VewVzFagFMxXTBe4U/QWjFCUGUxRMBv0T cgajE5YGQxO5Bt8S2gZ3EvoGChIYB5kRNAckEU4HrBBmBy8QfAewD5AHLQ+i B6cOsgcfDr8Hkw3LBwUN0wd1DNoH4wveB08L3we7Ct4HKQraB5kJ0wcLCcsH fwi/B/cHsgdxB6IH7gaQB28GfAfyBWYHegVOBwUFNAeUBBgHJwT6Br8D2gZb A7kG+wKWBqECcgZLAkwG+wElBrAB/QVrAdMFKwGoBfEAewW+AE4FkQAgBWoA 8ARKAMAEMACPBB4AXQQTACoEDwD3Aw8A+wMPAAcEDwAaBA8AMwQPAFMEDwB4 BA8AowQPANIEDwAFBQ8APAUPAHYFDwCyBQ8A8AUPADAGDwBxBg8AswYPAPUG DwA2Bw8AdgcPALQHDwDwBw8AKggPAGEIDwCUCA8AwwgPAO4IDwATCQ8AMwkP AEwJDwBfCQ8AawkPAG8JEwCiCR4A1QkwAAcKSgA4CmoAaAqRAJgKvgDGCvEA 8worASALawFLC7ABdQv7AZ0LSwLEC6EC6gv7Ag4MWwMxDL8DUgwnBHIMlASQ DAUFrAx6BcYM8gXeDG8G9AzuBggNcQcaDfcHKg1/CDcNCwlDDZkJSw0pClIN uwpWDU8LVw3jC1YNdQxSDQUNSw2TDUMNHw43DacOKg0tDxoNsA8IDS8Q9Ays EN4MJBHGDJkRrAwKEpAMdxJyDN8SUgxDEzEMoxMODP0T6gtTFMQLoxSdC+4U dQszFUsLcxUgC60V8wrgFcYKDRaYCjQWaApUFjgKbhYHCoAW1QmLFqIJjxZv CY8WawmPFl8JjxZMCY8WMwmPFhMJjxbuCI8WwwiPFpQIjxZhCI8WKgiPFvAH jxa0B48WdgePFjYHjxb1Bo8WswaPFnEGjxYwBo8W8AWPFrIFjxZ2BY8WPAWP FgUFjxbSBI8WowSPFngEjxZTBI8WMwSPFhoEjxYHBI8W+wOPFvcDJQAAAAwA AAAFAACAKAAAAAwAAAADAAAAFAAAAAwAAAANAAAAVwAAACADAAABAAAAJgAA ANoAAACBAAAAwQAAAI8W9wOLFioEgBZdBG4WjwRUFsAENBbwBA0WIAXgFU4F rRV7BXMVqAUzFdMF7hT9BaMUJQZTFEwG/RNyBqMTlgZDE7kG3xLaBncS+gYK EhgHmRE0ByQRTgesEGYHLxB8B7APkActD6IHpw6yBx8OvweTDcsHBQ3TB3UM 2gfjC94HTwvfB7sK3gcpCtoHmQnTBwsJywd/CL8H9weyB3EHogfuBpAHbwZ8 B/IFZgd6BU4HBQU0B5QEGAcnBPoGvwPaBlsDuQb7ApYGoQJyBksCTAb7ASUG sAH9BWsB0wUrAagF8QB7Bb4ATgWRACAFagDwBEoAwAQwAI8EHgBdBBMAKgQP APcDDwD7Aw8ABwQPABoEDwAzBA8AUwQPAHgEDwCjBA8A0gQPAAUFDwA8BQ8A dgUPALIFDwDwBQ8AMAYPAHEGDwCzBg8A9QYPADYHDwB2Bw8AtAcPAPAHDwAq CA8AYQgPAJQIDwDDCA8A7ggPABMJDwAzCQ8ATAkPAF8JDwBrCQ8AbwkTAKIJ HgDVCTAABwpKADgKagBoCpEAmAq+AMYK8QDzCisBIAtrAUsLsAF1C/sBnQtL AsQLoQLqC/sCDgxbAzEMvwNSDCcEcgyUBJAMBQWsDHoFxgzyBd4Mbwb0DO4G CA1xBxoN9wcqDX8INw0LCUMNmQlLDSkKUg27ClYNTwtXDeMLVg11DFINBQ1L DZMNQw0fDjcNpw4qDS0PGg2wDwgNLxD0DKwQ3gwkEcYMmRGsDAoSkAx3EnIM 3xJSDEMTMQyjEw4M/RPqC1MUxAujFJ0L7hR1CzMVSwtzFSALrRXzCuAVxgoN FpgKNBZoClQWOApuFgcKgBbVCYsWogmPFm8JjxZrCY8WXwmPFkwJjxYzCY8W EwmPFu4IjxbDCI8WlAiPFmEIjxYqCI8W8AePFrQHjxZ2B48WNgePFvUGjxaz Bo8WcQaPFjAGjxbwBY8WsgWPFnYFjxY8BY8WBQWPFtIEjxajBI8WeASPFlME jxYzBI8WGgSPFgcEjxb7A48W9wMUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAG AAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAACAAAAJQAAAAwA AAAFAACAJwAAABgAAAACAAAAAAAAAAAAAAAAAAAAJQAAAAwAAAACAAAAEwAA AAwAAAABAAAAVgAAADwAAABwAAAACwAAALgAAAAjAAAACAAAAJoLBwM5DY4D OhHvAQETewIMEhgBYQ0YAUEPgQGaCwcDJQAAAAwAAAAFAACAFAAAAAwAAAAN AAAAJgAAABwAAAAEAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABAAAACgA AAAMAAAAAwAAAFcAAAA8AAAAcAAAAAsAAAC4AAAAIgAAAAgAAACaCwcDOQ2O AzoR7wEBE3sCDBIYAWENGAFBD4EBmgsHAxQAAAAMAAAADQAAACYAAAAcAAAA AwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAQAAAAl AAAADAAAAAUAAIAlAAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAPAAAAHAA AAALAAAAuAAAACMAAAAIAAAAmgsHAzkNjgM6Ee8BARN7AgwSGAFhDRgBQQ+B AZoLBwMlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAAAAAA BQAAAAAAAAD///8AJQAAAAwAAAAEAAAAKAAAAAwAAAADAAAAVwAAADwAAABw AAAACwAAALgAAAAiAAAACAAAAJoLBwM5DY4DOhHvAQETewIMEhgBYQ0YAUEP gQGaCwcDFAAAAAwAAAANAAAAJgAAABwAAAADAAAABgAAAAUAAAAAAAAAAAAA ACUAAAAMAAAAAwAAACgAAAAMAAAABAAAACUAAAAMAAAABQAAgCUAAAAMAAAA AgAAABMAAAAMAAAAAQAAAFYAAAA8AAAAIgAAACcAAABpAAAAPwAAAAgAAADh CoMETAn8A24FmwV/Aw8FeQSLBkwJiwY6B/oF4QqDBCUAAAAMAAAABQAAgBQA AAAMAAAADQAAACYAAAAcAAAABAAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAA AAQAAAAoAAAADAAAAAMAAABXAAAAPAAAACIAAAAnAAAAaQAAAD8AAAAIAAAA 4QqDBEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6BeEKgwQUAAAADAAAAA0AAAAm AAAAHAAAAAMAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwA AAAEAAAAJQAAAAwAAAAFAACAJQAAAAwAAAACAAAAEwAAAAwAAAABAAAAVgAA ADwAAAAiAAAAJwAAAGkAAAA/AAAACAAAAOEKgwRMCfwDbgWbBX8DDwV5BIsG TAmLBjoH+gXhCoMEJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAE AAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABAAAACgAAAAMAAAAAwAAAFcA AAA8AAAAIgAAACcAAABpAAAAPwAAAAgAAADhCoMETAn8A24FmwV/Aw8FeQSL BkwJiwY6B/oF4QqDBBQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYAAAAFAAAA AAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAQAAAAlAAAADAAAAAUAAIAl AAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAPAAAACYAAAAKAAAAbgAAACEA AAAIAAAA4wOBAYwF+gCNCXsCWQsNAl8KawOvBWsDowcHA+MDgQElAAAADAAA AAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAAAAAABQAAAAAAAAD///8A JQAAAAwAAAAEAAAAKAAAAAwAAAADAAAAVwAAADwAAAAmAAAACQAAAG4AAAAh AAAACAAAAOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EBFAAAAAwA AAANAAAAJgAAABwAAAADAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAA ACgAAAAMAAAABAAAACUAAAAMAAAABQAAgCUAAAAMAAAAAgAAABMAAAAMAAAA AQAAAFYAAAA8AAAAJgAAAAoAAABuAAAAIQAAAAgAAADjA4EBjAX6AI0JewJZ Cw0CXwprA68FawOjBwcD4wOBASUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYA AAAcAAAABAAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAQAAAAoAAAADAAA AAMAAABXAAAAPAAAACYAAAAJAAAAbgAAACEAAAAIAAAA4wOBAYwF+gCNCXsC WQsNAl8KawOvBWsDowcHA+MDgQEUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAG AAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAAEAAAAJQAAAAwA AAAFAACAJQAAAAwAAAACAAAAEwAAAAwAAAABAAAAVgAAADwAAABuAAAAKQAA ALUAAABBAAAACAAAALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicG JQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAEAAAAAAAAAAUAAAAA AAAA////ACUAAAAMAAAABAAAACgAAAAMAAAAAwAAAFcAAAA8AAAAbgAAACkA AAC1AAAAQQAAAAgAAAC7EicGHBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxIn BhQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAA DAAAAAMAAAAoAAAADAAAAAQAAAAlAAAADAAAAAUAAIAlAAAADAAAAAIAAAAT AAAADAAAAAEAAABWAAAAPAAAAG4AAAApAAAAtQAAAEEAAAAIAAAAuxInBhwR rgY5DQ8FWQubBT8MOAQcETgEAA+hBLsSJwYlAAAADAAAAAUAAIAUAAAADAAA AA0AAAAmAAAAHAAAAAQAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAEAAAA KAAAAAwAAAADAAAAVwAAADwAAABuAAAAKQAAALUAAABBAAAACAAAALsSJwYc Ea4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGFAAAAAwAAAANAAAAJgAAABwA AAADAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABAAA ACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAD///8AAAAAACUAAAAMAAAA BAAAABMAAAAMAAAAAQAAAFYAAAA8AAAAcgAAAAwAAAC5AAAAJAAAAAgAAAC9 CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9D58BvQslAyUAAAAMAAAABQAAgBQA AAAMAAAADQAAACYAAAAcAAAABQAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAA AAUAAAAoAAAADAAAAAMAAABXAAAAPAAAAHIAAAAMAAAAuQAAACQAAAAIAAAA vQslA2ENrANdEQ0CHxOZAi8SQAF/DUABfQ+fAb0LJQMUAAAADAAAAA0AAAAm AAAAHAAAAAMAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwA AAAFAAAAJQAAAAwAAAAFAACAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAA ADwAAAByAAAADAAAALkAAAAkAAAACAAAAL0LJQNhDawDXRENAh8TmQIvEkAB fw1AAX0PnwG9CyUDJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAADwAAABy AAAADAAAALkAAAAkAAAACAAAAL0LJQNhDawDXRENAh8TmQIvEkABfw1AAX0P nwG9CyUDFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJQAAAAwAAAAEAAAAEwAA AAwAAAABAAAAVgAAADwAAAAjAAAAKAAAAGsAAABBAAAACAAAAA4LoQRvCRoE jAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEJQAAAAwAAAAFAACAFAAAAAwAAAAN AAAAJgAAABwAAAAFAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABQAAACgA AAAMAAAAAwAAAFcAAAA8AAAAIwAAACgAAABrAAAAQQAAAAgAAAAOC6EEbwka BIwFuQWiAy0FnASuBm8JrgZdBycGDguhBBQAAAAMAAAADQAAACYAAAAcAAAA AwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAUAAAAl AAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAAAAEAAABWAAAAPAAAACMA AAAoAAAAawAAAEEAAAAIAAAADguhBG8JGgSMBbkFogMtBZwErgZvCa4GXQcn Bg4LoQQlAAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAPAAAACMAAAAoAAAA awAAAEEAAAAIAAAADguhBG8JGgSMBbkFogMtBZwErgZvCa4GXQcnBg4LoQQU AAAADAAAAA0AAAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAAAAEA AABWAAAAPAAAACgAAAALAAAAbwAAACMAAAAIAAAAGgSfAa8FGAGwCZkCfAsr AoIKjgPcBY4DwQclAxoEnwElAAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAA HAAAAAUAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAFAAAAKAAAAAwAAAAD AAAAVwAAADwAAAAoAAAACwAAAG8AAAAiAAAACAAAABoEnwGvBRgBsAmZAnwL KwKCCo4D3AWOA8EHJQMaBJ8BFAAAAAwAAAANAAAAJgAAABwAAAADAAAABgAA AAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABQAAACUAAAAMAAAA BQAAgCUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAA8AAAAKAAAAAsAAABv AAAAIwAAAAgAAAAaBJ8BrwUYAbAJmQJ8CysCggqOA9wFjgPBByUDGgSfASUA AAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAA8AAAAKAAAAAsAAABvAAAAIgAA AAgAAAAaBJ8BrwUYAbAJmQJ8CysCggqOA9wFjgPBByUDGgSfARQAAAAMAAAA DQAAACUAAAAMAAAABQAAgCUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAA8 AAAAbwAAACoAAAC3AAAAQgAAAAgAAADjEkUGOhHMBmENLQV8C7kFbAxgBDoR YAQjD78E4xJFBiUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAABQAA AAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAUAAAAoAAAADAAAAAMAAABXAAAA PAAAAG8AAAAqAAAAtwAAAEIAAAAIAAAA4xJFBjoRzAZhDS0FfAu5BWwMYAQ6 EWAEIw+/BOMSRQYUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAABQAAAAAA AAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAAFAAAAJQAAAAwAAAAFAACAJQAA AAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADwAAABvAAAAKgAAALcAAABCAAAA CAAAAOMSRQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMPvwTjEkUGJQAAAAwAAAAF AACAFAAAAAwAAAANAAAAVwAAADwAAABvAAAAKgAAALcAAABCAAAACAAAAOMS RQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMPvwTjEkUGFAAAAAwAAAANAAAAIgAA AAwAAAD/////KAAAAAwAAAABAAAAJQAAAAwAAAAHAACAJQAAAAwAAAAAAACA MAAAAAwAAAAPAACASwAAABAAAAAAAAAABQAAACgAAAAMAAAAAgAAACgAAAAM AAAAAwAAACgAAAAMAAAABAAAAA4AAAAUAAAAAAAAABAAAAAUAAAAGkAAQCwA AAAgAAAA/////wIAAAAAAAAAAAAAAD9hXEMfMwNDAAAAAOgD6AMhAAAACAAA AGIAAAAMAAAAAQAAACEAAAAIAAAAYgAAAAwAAAABAAAAIQAAAAgAAAAhAAAA CAAAAHMAAAAMAAAAAAAAABwAAAAIAAAAJQAAAAwAAAAAAACAJQAAAAwAAAAH AACAJQAAAAwAAAAOAACAMAAAAAwAAAAPAACAGQAAAAwAAAD///8AGAAAAAwA AAAAAAAAEgAAAAwAAAACAAAAEwAAAAwAAAABAAAAFAAAAAwAAAANAAAAFQAA AAwAAAABAAAAFgAAAAwAAAAAAAAADQAAABAAAAAAAAAAAAAAADoAAAAMAAAA CgAAABsAAAAQAAAAAAAAAAAAAAAjAAAAIAAAAIUN7j4AAAAAAAAAAPvx6z4A AKJDAAC7QyQAAAAkAAAAhQ3uPgAAAAAAAAAA+/HrPgAAokMAALtDBAAAAHMA AAAMAAAAAAAAAA0AAAAQAAAARAEAAHYBAABSAAAAcAEAAAEAAAAUAAAACgAA AAAAAAAAAAAAkAEAAAAAAIEHAgIiUwB5AHMAdABlAG0AAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAH DwAApw8AAAcPAACTDwAABw8AAHoPAAAHDwAAaw8AABEPAABcDwAAFg8AAEgP AADQJGQDOQ8AANAkZAMmAIoBBBoh3gsAiAEWDwAAAg8AABEPAADzDgAABw8A AN8OAAAHDwAAyw4AAAcPAAC3DgAABw8AAKMOAAARDwAAAAAAABYPAACADgAA Kg8AAHYOAAA5DwAAZw4AAFIPAABTDgAAaw8AAE4OAACEDwAAPw4AAIQPAAA1 DgAAhA8AAAAAAACEDwAAHA4AAI4PAAANDgAAmA8AAP4NAACsDwAA+Q0AALsP AADvDQAAzw8AAO8NAADoDwAA7w0AAPwPAABkdgAIAAAAACUAAAAMAAAAAQAA AEYAAAAoAAAAHAAAAEdESUMCAAAAAAAAAAAAAADbAAAAggAAAAAAAAAjAAAA IAAAAIUN7j4AAAAAAAAAAPvx6z4AAKJDAAC7QyQAAAAkAAAAhQ3uPgAAAAAA AAAA+/HrPgAAokMAALtDBAAAACMAAAAgAAAAkaXLQgAAAAAAAAAAw6FvQgAA okMAALtDJAAAACQAAACRpctCAAAAAAAAAADDoW9CAACiQwAAu0MEAAAAIwAA ACAAAACFDe4+AAAAAAAAAAD78es+AACiQwAAu0MkAAAAJAAAAIUN7j4AAAAA AAAAAPvx6z4AAKJDAAC7QwQAAAAjAAAAIAAAAO8DkDwAAAAAAAAAAEP/jjwA AKJDAAC7QyQAAAAkAAAA7wOQPAAAAAAAAAAAQ/+OPAAAokMAALtDBAAAACMA AAAgAAAA7wOQPAAAAAAAAAAAQ/+OPAAAokMAALtDJAAAACQAAADvA5A8AAAA AAAAAABD/448AACiQwAAu0MEAAAAIwAAACAAAADvA5A8AAAAAAAAAABD/448 AACiQwAAu0MkAAAAJAAAAO8DkDwAAAAAAAAAAEP/jjwAAKJDAAC7QwQAAAAh AAAACAAAABQAAAAMAAAADQAAACUAAAAMAAAABQAAgCYAAAAcAAAAAgAAAAAA AAAAAAAAAAAAAAAAAAAlAAAADAAAAAIAAAAmAAAAHAAAAAMAAAAGAAAAHgAA AAAAAAAAAAAAJQAAAAwAAAADAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAA AAAAAL+/vwAAAAAAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAACACAABE AQAAdgEAAKoBAACaAQAAgQAAAI8W9wOLFsQDgBaRA24WXwNVFi4DNBb9Ag4W zgLgFZ8CrRVyAnMVRgIzFRoC7hTxAaMUyAFTFKEB/RN7AaMTVwFEEzQB4BIT AXcS8wAKEtYAmRG6ACURoACsEIgAMBByALAPXgAtD0wAqA48AB8OLwCUDSMA Bg0bAHYMFADjCxAATwsPALsKEAAoChQAmAkbAAoJIwB/CC8A9gc8AHEHTADu Bl4AbgZyAPIFiAB5BaAABQW6AJQE1gAnBPMAvgMTAVoDNAH7AlcBoQJ7AUsC oQH7AcgBsAHxAWsBGgIrAUYC8QByAr4AnwKQAM4CagD9AkkALgMwAF8DHgCR AxMAxAMPAPcDEwAqBB4AXQQwAI8ESQDABGoA8QSQACAFvgBPBfEAfAUrAagF awHUBbAB/QX7ASYGSwJNBqECcwb7ApcGWgO6Br4D2wYnBPsGlAQYBwUFNAd5 BU4H8gVmB24GfAfuBpAHcQeiB/YHsgd/CL8HCgnLB5gJ0wcoCtoHuwreB08L 3wfjC94HdgzaBwYN0weUDcsHHw6/B6gOsgctD6IHsA+QBzAQfAesEGYHJRFO B5kRNAcKEhgHdxL7BuAS2wZEE7oGoxOXBv0TcwZTFE0GoxQmBu4U/QUzFdQF cxWoBa0VfAXgFU8FDhYgBTQW8QRVFsAEbhaPBIAWXQSLFioEjxb3AyUAAAAM AAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAgAgAARAEAAHYB AACqAQAAmgEAAIEAAACPFvcDixbEA4AWkQNuFl8DVRYuAzQW/QIOFs4C4BWf Aq0VcgJzFUYCMxUaAu4U8QGjFMgBUxShAf0TewGjE1cBRBM0AeASEwF3EvMA ChLWAJkRugAlEaAArBCIADAQcgCwD14ALQ9MAKgOPAAfDi8AlA0jAAYNGwB2 DBQA4wsQAE8LDwC7ChAAKAoUAJgJGwAKCSMAfwgvAPYHPABxB0wA7gZeAG4G cgDyBYgAeQWgAAUFugCUBNYAJwTzAL4DEwFaAzQB+wJXAaECewFLAqEB+wHI AbAB8QFrARoCKwFGAvEAcgK+AJ8CkADOAmoA/QJJAC4DMABfAx4AkQMTAMQD DwD3AxMAKgQeAF0EMACPBEkAwARqAPEEkAAgBb4ATwXxAHwFKwGoBWsB1AWw Af0F+wEmBksCTQahAnMG+wKXBloDuga+A9sGJwT7BpQEGAcFBTQHeQVOB/IF ZgduBnwH7gaQB3EHogf2B7IHfwi/BwoJyweYCdMHKAraB7sK3gdPC98H4wve B3YM2gcGDdMHlA3LBx8OvweoDrIHLQ+iB7APkAcwEHwHrBBmByURTgeZETQH ChIYB3cS+wbgEtsGRBO6BqMTlwb9E3MGUxRNBqMUJgbuFP0FMxXUBXMVqAWt FXwF4BVPBQ4WIAU0FvEEVRbABG4WjwSAFl0EixYqBI8W9wMUAAAADAAAAA0A AAAlAAAADAAAAAUAAIAnAAAAGAAAAAQAAAAAAAAAv7+/AAAAAAAlAAAADAAA AAQAAAATAAAADAAAAAEAAABWAAAAIAMAAEQBAACHAQAAqgEAALIBAADBAAAA jxb3A4sWKgSAFl0EbhaPBFQWwAQ0FvAEDRYgBeAVTgWtFXsFcxWoBTMV0wXu FP0FoxQlBlMUTAb9E3IGoxOWBkMTuQbfEtoGdxL6BgoSGAeZETQHJBFOB6wQ ZgcvEHwHsA+QBy0PogenDrIHHw6/B5MNywcFDdMHdQzaB+ML3gdPC98Huwre BykK2geZCdMHCwnLB38Ivwf3B7IHcQeiB+4GkAdvBnwH8gVmB3oFTgcFBTQH lAQYBycE+ga/A9oGWwO5BvsClgahAnIGSwJMBvsBJQawAf0FawHTBSsBqAXx AHsFvgBOBZEAIAVqAPAESgDABDAAjwQeAF0EEwAqBA8A9wMPAPsDDwAHBA8A GgQPADMEDwBTBA8AeAQPAKMEDwDSBA8ABQUPADwFDwB2BQ8AsgUPAPAFDwAw Bg8AcQYPALMGDwD1Bg8ANgcPAHYHDwC0Bw8A8AcPACoIDwBhCA8AlAgPAMMI DwDuCA8AEwkPADMJDwBMCQ8AXwkPAGsJDwBvCRMAogkeANUJMAAHCkoAOApq AGgKkQCYCr4AxgrxAPMKKwEgC2sBSwuwAXUL+wGdC0sCxAuhAuoL+wIODFsD MQy/A1IMJwRyDJQEkAwFBawMegXGDPIF3gxvBvQM7gYIDXEHGg33ByoNfwg3 DQsJQw2ZCUsNKQpSDbsKVg1PC1cN4wtWDXUMUg0FDUsNkw1DDR8ONw2nDioN LQ8aDbAPCA0vEPQMrBDeDCQRxgyZEawMChKQDHcScgzfElIMQxMxDKMTDgz9 E+oLUxTEC6MUnQvuFHULMxVLC3MVIAutFfMK4BXGCg0WmAo0FmgKVBY4Cm4W BwqAFtUJixaiCY8WbwmPFmsJjxZfCY8WTAmPFjMJjxYTCY8W7giPFsMIjxaU CI8WYQiPFioIjxbwB48WtAePFnYHjxY2B48W9QaPFrMGjxZxBo8WMAaPFvAF jxayBY8WdgWPFjwFjxYFBY8W0gSPFqMEjxZ4BI8WUwSPFjMEjxYaBI8WBwSP FvsDjxb3AyUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcA AAAgAwAARAEAAIcBAACqAQAAsgEAAMEAAACPFvcDixYqBIAWXQRuFo8EVBbA BDQW8AQNFiAF4BVOBa0VewVzFagFMxXTBe4U/QWjFCUGUxRMBv0TcgajE5YG QxO5Bt8S2gZ3EvoGChIYB5kRNAckEU4HrBBmBy8QfAewD5AHLQ+iB6cOsgcf Dr8Hkw3LBwUN0wd1DNoH4wveB08L3we7Ct4HKQraB5kJ0wcLCcsHfwi/B/cH sgdxB6IH7gaQB28GfAfyBWYHegVOBwUFNAeUBBgHJwT6Br8D2gZbA7kG+wKW BqECcgZLAkwG+wElBrAB/QVrAdMFKwGoBfEAewW+AE4FkQAgBWoA8ARKAMAE MACPBB4AXQQTACoEDwD3Aw8A+wMPAAcEDwAaBA8AMwQPAFMEDwB4BA8AowQP ANIEDwAFBQ8APAUPAHYFDwCyBQ8A8AUPADAGDwBxBg8AswYPAPUGDwA2Bw8A dgcPALQHDwDwBw8AKggPAGEIDwCUCA8AwwgPAO4IDwATCQ8AMwkPAEwJDwBf CQ8AawkPAG8JEwCiCR4A1QkwAAcKSgA4CmoAaAqRAJgKvgDGCvEA8worASAL awFLC7ABdQv7AZ0LSwLEC6EC6gv7Ag4MWwMxDL8DUgwnBHIMlASQDAUFrAx6 BcYM8gXeDG8G9AzuBggNcQcaDfcHKg1/CDcNCwlDDZkJSw0pClINuwpWDU8L Vw3jC1YNdQxSDQUNSw2TDUMNHw43DacOKg0tDxoNsA8IDS8Q9AysEN4MJBHG DJkRrAwKEpAMdxJyDN8SUgxDEzEMoxMODP0T6gtTFMQLoxSdC+4UdQszFUsL cxUgC60V8wrgFcYKDRaYCjQWaApUFjgKbhYHCoAW1QmLFqIJjxZvCY8WawmP Fl8JjxZMCY8WMwmPFhMJjxbuCI8WwwiPFpQIjxZhCI8WKgiPFvAHjxa0B48W dgePFjYHjxb1Bo8WswaPFnEGjxYwBo8W8AWPFrIFjxZ2BY8WPAWPFgUFjxbS BI8WowSPFngEjxZTBI8WMwSPFhoEjxYHBI8W+wOPFvcDFAAAAAwAAAANAAAA JgAAABwAAAAEAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAABAAAACgAAAAM AAAAAwAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAAAAAAAAAAAACUA AAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAA8AAAAeAEAAHoBAACaAQAAhgEA AAgAAACaCwcDOQ2OAzoR7wEBE3sCDBIYAWENGAFBD4EBmgsHAyUAAAAMAAAA BQAAgBQAAAAMAAAADQAAACYAAAAcAAAABQAAAAAAAAAFAAAAAAAAAP///wAl AAAADAAAAAUAAAAoAAAADAAAAAQAAABXAAAAPAAAAHgBAAB6AQAAmgEAAIYB AAAIAAAAmgsHAzkNjgM6Ee8BARN7AgwSGAFhDRgBQQ+BAZoLBwMUAAAADAAA AA0AAAAmAAAAHAAAAAQAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAAEAAAA KAAAAAwAAAAFAAAAJQAAAAwAAAAFAACAJQAAAAwAAAADAAAAEwAAAAwAAAAB AAAAVgAAADwAAAB4AQAAegEAAJoBAACGAQAACAAAAJoLBwM5DY4DOhHvAQET ewIMEhgBYQ0YAUEPgQGaCwcDJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAA ABwAAAAFAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABQAAACgAAAAMAAAA BAAAAFcAAAA8AAAAeAEAAHoBAACaAQAAhgEAAAgAAACaCwcDOQ2OAzoR7wEB E3sCDBIYAWENGAFBD4EBmgsHAxQAAAAMAAAADQAAACYAAAAcAAAABAAAAAYA AAAFAAAAAAAAAAAAAAAlAAAADAAAAAQAAAAoAAAADAAAAAUAAAAlAAAADAAA AAUAAIAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAPAAAAFMBAACHAQAA dQEAAJQBAAAIAAAA4QqDBEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6BeEKgwQl AAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAUAAAAAAAAABQAAAAAA AAD///8AJQAAAAwAAAAFAAAAKAAAAAwAAAAEAAAAVwAAADwAAABTAQAAhwEA AHUBAACUAQAACAAAAOEKgwRMCfwDbgWbBX8DDwV5BIsGTAmLBjoH+gXhCoME FAAAAAwAAAANAAAAJgAAABwAAAAEAAAABgAAAAUAAAAAAAAAAAAAACUAAAAM AAAABAAAACgAAAAMAAAABQAAACUAAAAMAAAABQAAgCUAAAAMAAAAAwAAABMA AAAMAAAAAQAAAFYAAAA8AAAAUwEAAIcBAAB1AQAAlAEAAAgAAADhCoMETAn8 A24FmwV/Aw8FeQSLBkwJiwY6B/oF4QqDBCUAAAAMAAAABQAAgBQAAAAMAAAA DQAAACYAAAAcAAAABQAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAUAAAAo AAAADAAAAAQAAABXAAAAPAAAAFMBAACHAQAAdQEAAJQBAAAIAAAA4QqDBEwJ /ANuBZsFfwMPBXkEiwZMCYsGOgf6BeEKgwQUAAAADAAAAA0AAAAmAAAAHAAA AAQAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAAEAAAAKAAAAAwAAAAFAAAA JQAAAAwAAAAFAACAJQAAAAwAAAADAAAAEwAAAAwAAAABAAAAVgAAADwAAABV AQAAegEAAHgBAACGAQAACAAAAOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MH BwPjA4EBJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAFAAAAAAAA AAUAAAAAAAAA////ACUAAAAMAAAABQAAACgAAAAMAAAABAAAAFcAAAA8AAAA VQEAAHoBAAB4AQAAhgEAAAgAAADjA4EBjAX6AI0JewJZCw0CXwprA68FawOj BwcD4wOBARQAAAAMAAAADQAAACYAAAAcAAAABAAAAAYAAAAFAAAAAAAAAAAA AAAlAAAADAAAAAQAAAAoAAAADAAAAAUAAAAlAAAADAAAAAUAAIAlAAAADAAA AAMAAAATAAAADAAAAAEAAABWAAAAPAAAAFUBAAB6AQAAeAEAAIYBAAAIAAAA 4wOBAYwF+gCNCXsCWQsNAl8KawOvBWsDowcHA+MDgQElAAAADAAAAAUAAIAU AAAADAAAAA0AAAAmAAAAHAAAAAUAAAAAAAAABQAAAAAAAAD///8AJQAAAAwA AAAFAAAAKAAAAAwAAAAEAAAAVwAAADwAAABVAQAAegEAAHgBAACGAQAACAAA AOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EBFAAAAAwAAAANAAAA JgAAABwAAAAEAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAABAAAACgAAAAM AAAABQAAACUAAAAMAAAABQAAgCUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYA AAA8AAAAdwEAAIgBAACZAQAAlAEAAAgAAAC7EicGHBGuBjkNDwVZC5sFPww4 BBwROAQAD6EEuxInBiUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAA BQAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAUAAAAoAAAADAAAAAQAAABX AAAAPAAAAHcBAACIAQAAmQEAAJQBAAAIAAAAuxInBhwRrgY5DQ8FWQubBT8M OAQcETgEAA+hBLsSJwYUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAGAAAABQAA AAAAAAAAAAAAJQAAAAwAAAAEAAAAKAAAAAwAAAAFAAAAJQAAAAwAAAAFAACA JQAAAAwAAAADAAAAEwAAAAwAAAABAAAAVgAAADwAAAB3AQAAiAEAAJkBAACU AQAACAAAALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGJQAAAAwA AAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAFAAAAAAAAAAUAAAAAAAAA//// ACUAAAAMAAAABQAAACgAAAAMAAAABAAAAFcAAAA8AAAAdwEAAIgBAACZAQAA lAEAAAgAAAC7EicGHBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxInBhQAAAAM AAAADQAAACYAAAAcAAAABAAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAQA AAAoAAAADAAAAAUAAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAA//// AAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAPAAAAHgBAAB7AQAA mwEAAIcBAAAIAAAAvQslA2ENrANdEQ0CHxOZAi8SQAF/DUABfQ+fAb0LJQMl AAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAYAAAAAAAAABQAAAAAA AAD///8AJQAAAAwAAAAGAAAAKAAAAAwAAAAEAAAAVwAAADwAAAB4AQAAewEA AJsBAACHAQAACAAAAL0LJQNhDawDXRENAh8TmQIvEkABfw1AAX0PnwG9CyUD FAAAAAwAAAANAAAAJgAAABwAAAAEAAAABgAAAAUAAAAAAAAAAAAAACUAAAAM AAAABAAAACgAAAAMAAAABgAAACUAAAAMAAAABQAAgCUAAAAMAAAABQAAABMA AAAMAAAAAQAAAFYAAAA8AAAAeAEAAHsBAACbAQAAhwEAAAgAAAC9CyUDYQ2s A10RDQIfE5kCLxJAAX8NQAF9D58BvQslAyUAAAAMAAAABQAAgBQAAAAMAAAA DQAAAFcAAAA8AAAAeAEAAHsBAACbAQAAhwEAAAgAAAC9CyUDYQ2sA10RDQIf E5kCLxJAAX8NQAF9D58BvQslAxQAAAAMAAAADQAAACUAAAAMAAAABQAAgCUA AAAMAAAABQAAABMAAAAMAAAAAQAAAFYAAAA8AAAAVAEAAIgBAAB2AQAAlAEA AAgAAAAOC6EEbwkaBIwFuQWiAy0FnASuBm8JrgZdBycGDguhBCUAAAAMAAAA BQAAgBQAAAAMAAAADQAAACYAAAAcAAAABgAAAAAAAAAFAAAAAAAAAP///wAl AAAADAAAAAYAAAAoAAAADAAAAAQAAABXAAAAPAAAAFQBAACIAQAAdgEAAJQB AAAIAAAADguhBG8JGgSMBbkFogMtBZwErgZvCa4GXQcnBg4LoQQUAAAADAAA AA0AAAAmAAAAHAAAAAQAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAAEAAAA KAAAAAwAAAAGAAAAJQAAAAwAAAAFAACAJQAAAAwAAAAFAAAAEwAAAAwAAAAB AAAAVgAAADwAAABUAQAAiAEAAHYBAACUAQAACAAAAA4LoQRvCRoEjAW5BaID LQWcBK4GbwmuBl0HJwYOC6EEJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAA ADwAAABUAQAAiAEAAHYBAACUAQAACAAAAA4LoQRvCRoEjAW5BaIDLQWcBK4G bwmuBl0HJwYOC6EEFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJQAAAAwAAAAF AAAAEwAAAAwAAAABAAAAVgAAADwAAABWAQAAegEAAHgBAACGAQAACAAAABoE nwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EHJQMaBJ8BJQAAAAwAAAAFAACAFAAA AAwAAAANAAAAJgAAABwAAAAGAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAA BgAAACgAAAAMAAAABAAAAFcAAAA8AAAAVgEAAHoBAAB4AQAAhgEAAAgAAAAa BJ8BrwUYAbAJmQJ8CysCggqOA9wFjgPBByUDGgSfARQAAAAMAAAADQAAACYA AAAcAAAABAAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAQAAAAoAAAADAAA AAYAAAAlAAAADAAAAAUAAIAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAA PAAAAFYBAAB6AQAAeAEAAIYBAAAIAAAAGgSfAa8FGAGwCZkCfAsrAoIKjgPc BY4DwQclAxoEnwElAAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAPAAAAFYB AAB6AQAAeAEAAIYBAAAIAAAAGgSfAa8FGAGwCZkCfAsrAoIKjgPcBY4DwQcl AxoEnwEUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAlAAAADAAAAAUAAAATAAAA DAAAAAEAAABWAAAAPAAAAHcBAACJAQAAmQEAAJUBAAAIAAAA4xJFBjoRzAZh DS0FfAu5BWwMYAQ6EWAEIw+/BOMSRQYlAAAADAAAAAUAAIAUAAAADAAAAA0A AAAmAAAAHAAAAAYAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAGAAAAKAAA AAwAAAAEAAAAVwAAADwAAAB3AQAAiQEAAJkBAACVAQAACAAAAOMSRQY6EcwG YQ0tBXwLuQVsDGAEOhFgBCMPvwTjEkUGFAAAAAwAAAANAAAAJgAAABwAAAAE AAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAABAAAACgAAAAMAAAABgAAACUA AAAMAAAABQAAgCUAAAAMAAAABQAAABMAAAAMAAAAAQAAAFYAAAA8AAAAdwEA AIkBAACZAQAAlQEAAAgAAADjEkUGOhHMBmENLQV8C7kFbAxgBDoRYAQjD78E 4xJFBiUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAA8AAAAdwEAAIkBAACZ AQAAlQEAAAgAAADjEkUGOhHMBmENLQV8C7kFbAxgBDoRYAQjD78E4xJFBhQA AAAMAAAADQAAACIAAAAMAAAA/////ygAAAAMAAAAAgAAACUAAAAMAAAABwAA gCUAAAAMAAAAAAAAgEsAAAAQAAAAAAAAAAUAAAAoAAAADAAAAAMAAAAoAAAA DAAAAAQAAAAoAAAADAAAAAUAAABGAAAAFAAAAAgAAABHRElDAwAAACIAAAAM AAAA/////yIAAAAMAAAA/////yUAAAAMAAAADQAAgCgAAAAMAAAAAQAAACIA AAAMAAAA/////0YAAABINgAAPDYAAEVNRismQAAAEAAAAAQAAAAAAAAAJUAA ABAAAAAEAAAAAAAAACxAAAAkAAAAGAAAAM6N0z0AAACAAAAAgDEcez0/DwFD ndi6QwhAAQXINQAAvDUAAAEQwNsCAAAAAwAAAKw1AAABAAAAbAAAAAEAAAAB AAAA2gAAAIEAAAAAAAAAAAAAAKEWAABpDQAAIEVNRgAAAQCsNQAA+gAAAAYA AAAAAAAAAAAAAAAAAAAABQAAAAQAAFIBAAAOAQAAAAAAAAAAAAAAAAAAUCgF ALAeBABGAAAATBcAAD4XAABHRElDAQAAgAADAACtn7DzAAAAACYXAAABAAkA AAOTCwAABgCGAQAAAAAFAAAADAJoDaAWBQAAAAwCaA2gFgcAAAAQBAEAAQAB AAEAAwAAAB4ABAAAAAQBDQAHAAAA/AIBAAAAAAAAAAQAAAAtAQAACQAAAPoC AAAAAAAAAAAAACIABAAAAC0BAQAJAAAA+gIGAB4AAAAAAAAAIgAEAAAALQEC AAQAAAAtAQAABwAAAPwCAAC/v78AAAAEAAAALQEDAAQAAAAGAQEABgEAACQD gQCPFvcDixbEA4AWkQNuFl8DVRYuAzQW/QIOFs4C4BWfAq0VcgJzFUYCMxUa Au4U8QGjFMgBUxShAf0TewGjE1cBRBM0AeASEwF3EvMAChLWAJkRugAlEaAA rBCIADAQcgCwD14ALQ9MAKgOPAAfDi8AlA0jAAYNGwB2DBQA4wsQAE8LDwC7 ChAAKAoUAJgJGwAKCSMAfwgvAPYHPABxB0wA7gZeAG4GcgDyBYgAeQWgAAUF ugCUBNYAJwTzAL4DEwFaAzQB+wJXAaECewFLAqEB+wHIAbAB8QFrARoCKwFG AvEAcgK+AJ8CkADOAmoA/QJJAC4DMABfAx4AkQMTAMQDDwD3AxMAKgQeAF0E MACPBEkAwARqAPEEkAAgBb4ATwXxAHwFKwGoBWsB1AWwAf0F+wEmBksCTQah AnMG+wKXBloDuga+A9sGJwT7BpQEGAcFBTQHeQVOB/IFZgduBnwH7gaQB3EH ogf2B7IHfwi/BwoJyweYCdMHKAraB7sK3gdPC98H4wveB3YM2gcGDdMHlA3L Bx8OvweoDrIHLQ+iB7APkAcwEHwHrBBmByURTgeZETQHChIYB3cS+wbgEtsG RBO6BqMTlwb9E3MGUxRNBqMUJgbuFP0FMxXUBXMVqAWtFXwF4BVPBQ4WIAU0 FvEEVRbABG4WjwSAFl0EixYqBI8W9wMEAAAALQEAAAQAAADwAQMABAAAAAQB DQAGAQAAJQOBAI8W9wOLFsQDgBaRA24WXwNVFi4DNBb9Ag4WzgLgFZ8CrRVy AnMVRgIzFRoC7hTxAaMUyAFTFKEB/RN7AaMTVwFEEzQB4BITAXcS8wAKEtYA mRG6ACURoACsEIgAMBByALAPXgAtD0wAqA48AB8OLwCUDSMABg0bAHYMFADj CxAATwsPALsKEAAoChQAmAkbAAoJIwB/CC8A9gc8AHEHTADuBl4AbgZyAPIF iAB5BaAABQW6AJQE1gAnBPMAvgMTAVoDNAH7AlcBoQJ7AUsCoQH7AcgBsAHx AWsBGgIrAUYC8QByAr4AnwKQAM4CagD9AkkALgMwAF8DHgCRAxMAxAMPAPcD EwAqBB4AXQQwAI8ESQDABGoA8QSQACAFvgBPBfEAfAUrAagFawHUBbAB/QX7 ASYGSwJNBqECcwb7ApcGWgO6Br4D2wYnBPsGlAQYBwUFNAd5BU4H8gVmB24G fAfuBpAHcQeiB/YHsgd/CL8HCgnLB5gJ0wcoCtoHuwreB08L3wfjC94Hdgza BwYN0weUDcsHHw6/B6gOsgctD6IHsA+QBzAQfAesEGYHJRFOB5kRNAcKEhgH dxL7BuAS2wZEE7oGoxOXBv0TcwZTFE0GoxQmBu4U/QUzFdQFcxWoBa0VfAXg FU8FDhYgBTQW8QRVFsAEbhaPBIAWXQSLFioEjxb3AwQAAAAEAQ0ABAAAAC0B AAAHAAAA/AIAAL+/vwAAAAQAAAAtAQMABAAAAAYBAQCGAQAAJAPBAI8W9wOL FioEgBZdBG4WjwRUFsAENBbwBA0WIAXgFU4FrRV7BXMVqAUzFdMF7hT9BaMU JQZTFEwG/RNyBqMTlgZDE7kG3xLaBncS+gYKEhgHmRE0ByQRTgesEGYHLxB8 B7APkActD6IHpw6yBx8OvweTDcsHBQ3TB3UM2gfjC94HTwvfB7sK3gcpCtoH mQnTBwsJywd/CL8H9weyB3EHogfuBpAHbwZ8B/IFZgd6BU4HBQU0B5QEGAcn BPoGvwPaBlsDuQb7ApYGoQJyBksCTAb7ASUGsAH9BWsB0wUrAagF8QB7Bb4A TgWRACAFagDwBEoAwAQwAI8EHgBdBBMAKgQPAPcDDwD7Aw8ABwQPABoEDwAz BA8AUwQPAHgEDwCjBA8A0gQPAAUFDwA8BQ8AdgUPALIFDwDwBQ8AMAYPAHEG DwCzBg8A9QYPADYHDwB2Bw8AtAcPAPAHDwAqCA8AYQgPAJQIDwDDCA8A7ggP ABMJDwAzCQ8ATAkPAF8JDwBrCQ8AbwkTAKIJHgDVCTAABwpKADgKagBoCpEA mAq+AMYK8QDzCisBIAtrAUsLsAF1C/sBnQtLAsQLoQLqC/sCDgxbAzEMvwNS DCcEcgyUBJAMBQWsDHoFxgzyBd4Mbwb0DO4GCA1xBxoN9wcqDX8INw0LCUMN mQlLDSkKUg27ClYNTwtXDeMLVg11DFINBQ1LDZMNQw0fDjcNpw4qDS0PGg2w DwgNLxD0DKwQ3gwkEcYMmRGsDAoSkAx3EnIM3xJSDEMTMQyjEw4M/RPqC1MU xAujFJ0L7hR1CzMVSwtzFSALrRXzCuAVxgoNFpgKNBZoClQWOApuFgcKgBbV CYsWogmPFm8JjxZrCY8WXwmPFkwJjxYzCY8WEwmPFu4IjxbDCI8WlAiPFmEI jxYqCI8W8AePFrQHjxZ2B48WNgePFvUGjxazBo8WcQaPFjAGjxbwBY8WsgWP FnYFjxY8BY8WBQWPFtIEjxajBI8WeASPFlMEjxYzBI8WGgSPFgcEjxb7A48W 9wMEAAAALQEAAAQAAADwAQMABAAAAAQBDQCGAQAAJQPBAI8W9wOLFioEgBZd BG4WjwRUFsAENBbwBA0WIAXgFU4FrRV7BXMVqAUzFdMF7hT9BaMUJQZTFEwG /RNyBqMTlgZDE7kG3xLaBncS+gYKEhgHmRE0ByQRTgesEGYHLxB8B7APkAct D6IHpw6yBx8OvweTDcsHBQ3TB3UM2gfjC94HTwvfB7sK3gcpCtoHmQnTBwsJ ywd/CL8H9weyB3EHogfuBpAHbwZ8B/IFZgd6BU4HBQU0B5QEGAcnBPoGvwPa BlsDuQb7ApYGoQJyBksCTAb7ASUGsAH9BWsB0wUrAagF8QB7Bb4ATgWRACAF agDwBEoAwAQwAI8EHgBdBBMAKgQPAPcDDwD7Aw8ABwQPABoEDwAzBA8AUwQP AHgEDwCjBA8A0gQPAAUFDwA8BQ8AdgUPALIFDwDwBQ8AMAYPAHEGDwCzBg8A 9QYPADYHDwB2Bw8AtAcPAPAHDwAqCA8AYQgPAJQIDwDDCA8A7ggPABMJDwAz CQ8ATAkPAF8JDwBrCQ8AbwkTAKIJHgDVCTAABwpKADgKagBoCpEAmAq+AMYK 8QDzCisBIAtrAUsLsAF1C/sBnQtLAsQLoQLqC/sCDgxbAzEMvwNSDCcEcgyU BJAMBQWsDHoFxgzyBd4Mbwb0DO4GCA1xBxoN9wcqDX8INw0LCUMNmQlLDSkK Ug27ClYNTwtXDeMLVg11DFINBQ1LDZMNQw0fDjcNpw4qDS0PGg2wDwgNLxD0 DKwQ3gwkEcYMmRGsDAoSkAx3EnIM3xJSDEMTMQyjEw4M/RPqC1MUxAujFJ0L 7hR1CzMVSwtzFSALrRXzCuAVxgoNFpgKNBZoClQWOApuFgcKgBbVCYsWogmP Fm8JjxZrCY8WXwmPFkwJjxYzCY8WEwmPFu4IjxbDCI8WlAiPFmEIjxYqCI8W 8AePFrQHjxZ2B48WNgePFvUGjxazBo8WcQaPFjAGjxbwBY8WsgWPFnYFjxY8 BY8WBQWPFtIEjxajBI8WeASPFlMEjxYzBI8WGgSPFgcEjxb7A48W9wMEAAAA BAENAAkAAAD6AgYABQAAAAAAAAAiAAQAAAAtAQMABAAAAPABAgAEAAAALQEA AAcAAAD8AgAAAAAAAAAABAAAAC0BAgAEAAAABgEBABQAAAAkAwgAmgsHAzkN jgM6Ee8BARN7AgwSGAFhDRgBQQ+BAZoLBwMEAAAALQEAAAQAAAAEAQ0ACQAA APoCAAAFAAAA////ACIABAAAAC0BBAAEAAAA8AEDABQAAAAlAwgAmgsHAzkN jgM6Ee8BARN7AgwSGAFhDRgBQQ+BAZoLBwMEAAAABAENAAkAAAD6AgYABQAA AAAAAAAiAAQAAAAtAQMABAAAAPABBAAEAAAALQEAAAQAAAAtAQIABAAAAAYB AQAUAAAAJAMIAJoLBwM5DY4DOhHvAQETewIMEhgBYQ0YAUEPgQGaCwcDBAAA AC0BAAAEAAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQQABAAAAPAB AwAUAAAAJQMIAJoLBwM5DY4DOhHvAQETewIMEhgBYQ0YAUEPgQGaCwcDBAAA AAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQQABAAAAC0B AAAEAAAALQECAAQAAAAGAQEAFAAAACQDCADhCoMETAn8A24FmwV/Aw8FeQSL BkwJiwY6B/oF4QqDBAQAAAAtAQAABAAAAAQBDQAJAAAA+gIAAAUAAAD///8A IgAEAAAALQEEAAQAAADwAQMAFAAAACUDCADhCoMETAn8A24FmwV/Aw8FeQSL BkwJiwY6B/oF4QqDBAQAAAAEAQ0ACQAAAPoCBgAFAAAAAAAAACIABAAAAC0B AwAEAAAA8AEEAAQAAAAtAQAABAAAAC0BAgAEAAAABgEBABQAAAAkAwgA4QqD BEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6BeEKgwQEAAAALQEAAAQAAAAEAQ0A CQAAAPoCAAAFAAAA////ACIABAAAAC0BBAAEAAAA8AEDABQAAAAlAwgA4QqD BEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6BeEKgwQEAAAABAENAAkAAAD6AgYA BQAAAAAAAAAiAAQAAAAtAQMABAAAAPABBAAEAAAALQEAAAQAAAAtAQIABAAA AAYBAQAUAAAAJAMIAOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EB BAAAAC0BAAAEAAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQQABAAA APABAwAUAAAAJQMIAOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EB BAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQQABAAA AC0BAAAEAAAALQECAAQAAAAGAQEAFAAAACQDCADjA4EBjAX6AI0JewJZCw0C XwprA68FawOjBwcD4wOBAQQAAAAtAQAABAAAAAQBDQAJAAAA+gIAAAUAAAD/ //8AIgAEAAAALQEEAAQAAADwAQMAFAAAACUDCADjA4EBjAX6AI0JewJZCw0C XwprA68FawOjBwcD4wOBAQQAAAAEAQ0ACQAAAPoCBgAFAAAAAAAAACIABAAA AC0BAwAEAAAA8AEEAAQAAAAtAQAABAAAAC0BAgAEAAAABgEBABQAAAAkAwgA uxInBhwRrgY5DQ8FWQubBT8MOAQcETgEAA+hBLsSJwYEAAAALQEAAAQAAAAE AQ0ACQAAAPoCAAAFAAAA////ACIABAAAAC0BBAAEAAAA8AEDABQAAAAlAwgA uxInBhwRrgY5DQ8FWQubBT8MOAQcETgEAA+hBLsSJwYEAAAABAENAAkAAAD6 AgYABQAAAAAAAAAiAAQAAAAtAQMABAAAAPABBAAEAAAALQEAAAQAAAAtAQIA BAAAAAYBAQAUAAAAJAMIALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7 EicGBAAAAC0BAAAEAAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQQA BAAAAPABAwAUAAAAJQMIALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7 EicGBAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQQA BAAAAC0BAAAHAAAA/AIAAP///wAAAAQAAAAtAQQABAAAAAYBAQAUAAAAJAMI AL0LJQNhDawDXRENAh8TmQIvEkABfw1AAX0PnwG9CyUDBAAAAC0BAAAEAAAA BAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQUABAAAAPABAwAUAAAAJQMI AL0LJQNhDawDXRENAh8TmQIvEkABfw1AAX0PnwG9CyUDBAAAAAQBDQAJAAAA +gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQUABAAAAC0BAAAEAAAALQEE AAQAAAAGAQEAFAAAACQDCAC9CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9D58B vQslAwQAAAAtAQAABAAAAAQBDQAUAAAAJQMIAL0LJQNhDawDXRENAh8TmQIv EkABfw1AAX0PnwG9CyUDBAAAAAQBDQAEAAAALQEAAAQAAAAtAQQABAAAAAYB AQAUAAAAJAMIAA4LoQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEBAAA AC0BAAAEAAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQUABAAAAPAB AwAUAAAAJQMIAA4LoQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEBAAA AAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQUABAAAAC0B AAAEAAAALQEEAAQAAAAGAQEAFAAAACQDCAAOC6EEbwkaBIwFuQWiAy0FnASu Bm8JrgZdBycGDguhBAQAAAAtAQAABAAAAAQBDQAUAAAAJQMIAA4LoQRvCRoE jAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEBAAAAAQBDQAEAAAALQEAAAQAAAAt AQQABAAAAAYBAQAUAAAAJAMIABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EH JQMaBJ8BBAAAAC0BAAAEAAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAt AQUABAAAAPABAwAUAAAAJQMIABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EH JQMaBJ8BBAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADw AQUABAAAAC0BAAAEAAAALQEEAAQAAAAGAQEAFAAAACQDCAAaBJ8BrwUYAbAJ mQJ8CysCggqOA9wFjgPBByUDGgSfAQQAAAAtAQAABAAAAAQBDQAUAAAAJQMI ABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EHJQMaBJ8BBAAAAAQBDQAEAAAA LQEAAAQAAAAtAQQABAAAAAYBAQAUAAAAJAMIAOMSRQY6EcwGYQ0tBXwLuQVs DGAEOhFgBCMPvwTjEkUGBAAAAC0BAAAEAAAABAENAAkAAAD6AgAABQAAAP// /wAiAAQAAAAtAQUABAAAAPABAwAUAAAAJQMIAOMSRQY6EcwGYQ0tBXwLuQVs DGAEOhFgBCMPvwTjEkUGBAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAA LQEDAAQAAADwAQUABAAAAC0BAAAEAAAALQEEAAQAAAAGAQEAFAAAACQDCADj EkUGOhHMBmENLQV8C7kFbAxgBDoRYAQjD78E4xJFBgQAAAAtAQAABAAAAAQB DQAUAAAAJQMIAOMSRQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMPvwTjEkUGBAAA AAQBDQAEAAAAJwH//x4AAAAmBg8AMQBDb3B5cmlnaHQgqSAxOTk2IE1pY3Jv c3lzdGVtcyBFbmdpbmVlcmluZyBDb21wYW55AA8AAAAmBg8AEwBBbGwgUmln aHRzIFJlc2VydmVkAAQAAADwAQEAAwAAAAAAAAARAAAADAAAAAgAAAALAAAA EAAAANsAAACCAAAACQAAABAAAADbAAAAggAAAAkAAAAQAAAAoBYAAGgNAAAJ AAAAEAAAAKAWAABoDQAAIAAAABgAAAABAAAAAQAAAAEAAAABAAAAIQAAAAgA AAAUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAmAAAAHAAAAAEAAAAAAAAAAAAA AAAAAAAAAAAAJQAAAAwAAAABAAAAJgAAABwAAAACAAAABgAAAB4AAAAAAAAA AAAAACUAAAAMAAAAAgAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAC/ v78AAAAAACUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAAgAgAAAQAAAAEA AADaAAAATAAAAIEAAACPFvcDixbEA4AWkQNuFl8DVRYuAzQW/QIOFs4C4BWf Aq0VcgJzFUYCMxUaAu4U8QGjFMgBUxShAf0TewGjE1cBRBM0AeASEwF3EvMA ChLWAJkRugAlEaAArBCIADAQcgCwD14ALQ9MAKgOPAAfDi8AlA0jAAYNGwB2 DBQA4wsQAE8LDwC7ChAAKAoUAJgJGwAKCSMAfwgvAPYHPABxB0wA7gZeAG4G cgDyBYgAeQWgAAUFugCUBNYAJwTzAL4DEwFaAzQB+wJXAaECewFLAqEB+wHI AbAB8QFrARoCKwFGAvEAcgK+AJ8CkADOAmoA/QJJAC4DMABfAx4AkQMTAMQD DwD3AxMAKgQeAF0EMACPBEkAwARqAPEEkAAgBb4ATwXxAHwFKwGoBWsB1AWw Af0F+wEmBksCTQahAnMG+wKXBloDuga+A9sGJwT7BpQEGAcFBTQHeQVOB/IF ZgduBnwH7gaQB3EHogf2B7IHfwi/BwoJyweYCdMHKAraB7sK3gdPC98H4wve B3YM2gcGDdMHlA3LBx8OvweoDrIHLQ+iB7APkAcwEHwHrBBmByURTgeZETQH ChIYB3cS+wbgEtsGRBO6BqMTlwb9E3MGUxRNBqMUJgbuFP0FMxXUBXMVqAWt FXwF4BVPBQ4WIAU0FvEEVRbABG4WjwSAFl0EixYqBI8W9wMlAAAADAAAAAUA AIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAIAIAAAEAAAABAAAA2gAA AEwAAACBAAAAjxb3A4sWxAOAFpEDbhZfA1UWLgM0Fv0CDhbOAuAVnwKtFXIC cxVGAjMVGgLuFPEBoxTIAVMUoQH9E3sBoxNXAUQTNAHgEhMBdxLzAAoS1gCZ EboAJRGgAKwQiAAwEHIAsA9eAC0PTACoDjwAHw4vAJQNIwAGDRsAdgwUAOML EABPCw8AuwoQACgKFACYCRsACgkjAH8ILwD2BzwAcQdMAO4GXgBuBnIA8gWI AHkFoAAFBboAlATWACcE8wC+AxMBWgM0AfsCVwGhAnsBSwKhAfsByAGwAfEB awEaAisBRgLxAHICvgCfApAAzgJqAP0CSQAuAzAAXwMeAJEDEwDEAw8A9wMT ACoEHgBdBDAAjwRJAMAEagDxBJAAIAW+AE8F8QB8BSsBqAVrAdQFsAH9BfsB JgZLAk0GoQJzBvsClwZaA7oGvgPbBicE+waUBBgHBQU0B3kFTgfyBWYHbgZ8 B+4GkAdxB6IH9geyB38IvwcKCcsHmAnTBygK2ge7Ct4HTwvfB+ML3gd2DNoH Bg3TB5QNywcfDr8HqA6yBy0PogewD5AHMBB8B6wQZgclEU4HmRE0BwoSGAd3 EvsG4BLbBkQTugajE5cG/RNzBlMUTQajFCYG7hT9BTMV1AVzFagFrRV8BeAV TwUOFiAFNBbxBFUWwARuFo8EgBZdBIsWKgSPFvcDFAAAAAwAAAANAAAAJQAA AAwAAAAFAACAJwAAABgAAAADAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAADAAAA EwAAAAwAAAABAAAAVgAAACADAAABAAAAJgAAANoAAACBAAAAwQAAAI8W9wOL FioEgBZdBG4WjwRUFsAENBbwBA0WIAXgFU4FrRV7BXMVqAUzFdMF7hT9BaMU JQZTFEwG/RNyBqMTlgZDE7kG3xLaBncS+gYKEhgHmRE0ByQRTgesEGYHLxB8 B7APkActD6IHpw6yBx8OvweTDcsHBQ3TB3UM2gfjC94HTwvfB7sK3gcpCtoH mQnTBwsJywd/CL8H9weyB3EHogfuBpAHbwZ8B/IFZgd6BU4HBQU0B5QEGAcn BPoGvwPaBlsDuQb7ApYGoQJyBksCTAb7ASUGsAH9BWsB0wUrAagF8QB7Bb4A TgWRACAFagDwBEoAwAQwAI8EHgBdBBMAKgQPAPcDDwD7Aw8ABwQPABoEDwAz BA8AUwQPAHgEDwCjBA8A0gQPAAUFDwA8BQ8AdgUPALIFDwDwBQ8AMAYPAHEG DwCzBg8A9QYPADYHDwB2Bw8AtAcPAPAHDwAqCA8AYQgPAJQIDwDDCA8A7ggP ABMJDwAzCQ8ATAkPAF8JDwBrCQ8AbwkTAKIJHgDVCTAABwpKADgKagBoCpEA mAq+AMYK8QDzCisBIAtrAUsLsAF1C/sBnQtLAsQLoQLqC/sCDgxbAzEMvwNS DCcEcgyUBJAMBQWsDHoFxgzyBd4Mbwb0DO4GCA1xBxoN9wcqDX8INw0LCUMN mQlLDSkKUg27ClYNTwtXDeMLVg11DFINBQ1LDZMNQw0fDjcNpw4qDS0PGg2w DwgNLxD0DKwQ3gwkEcYMmRGsDAoSkAx3EnIM3xJSDEMTMQyjEw4M/RPqC1MU xAujFJ0L7hR1CzMVSwtzFSALrRXzCuAVxgoNFpgKNBZoClQWOApuFgcKgBbV CYsWogmPFm8JjxZrCY8WXwmPFkwJjxYzCY8WEwmPFu4IjxbDCI8WlAiPFmEI jxYqCI8W8AePFrQHjxZ2B48WNgePFvUGjxazBo8WcQaPFjAGjxbwBY8WsgWP FnYFjxY8BY8WBQWPFtIEjxajBI8WeASPFlMEjxYzBI8WGgSPFgcEjxb7A48W 9wMlAAAADAAAAAUAAIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAIAMA AAEAAAAmAAAA2gAAAIEAAADBAAAAjxb3A4sWKgSAFl0EbhaPBFQWwAQ0FvAE DRYgBeAVTgWtFXsFcxWoBTMV0wXuFP0FoxQlBlMUTAb9E3IGoxOWBkMTuQbf EtoGdxL6BgoSGAeZETQHJBFOB6wQZgcvEHwHsA+QBy0PogenDrIHHw6/B5MN ywcFDdMHdQzaB+ML3gdPC98HuwreBykK2geZCdMHCwnLB38Ivwf3B7IHcQei B+4GkAdvBnwH8gVmB3oFTgcFBTQHlAQYBycE+ga/A9oGWwO5BvsClgahAnIG SwJMBvsBJQawAf0FawHTBSsBqAXxAHsFvgBOBZEAIAVqAPAESgDABDAAjwQe AF0EEwAqBA8A9wMPAPsDDwAHBA8AGgQPADMEDwBTBA8AeAQPAKMEDwDSBA8A BQUPADwFDwB2BQ8AsgUPAPAFDwAwBg8AcQYPALMGDwD1Bg8ANgcPAHYHDwC0 Bw8A8AcPACoIDwBhCA8AlAgPAMMIDwDuCA8AEwkPADMJDwBMCQ8AXwkPAGsJ DwBvCRMAogkeANUJMAAHCkoAOApqAGgKkQCYCr4AxgrxAPMKKwEgC2sBSwuw AXUL+wGdC0sCxAuhAuoL+wIODFsDMQy/A1IMJwRyDJQEkAwFBawMegXGDPIF 3gxvBvQM7gYIDXEHGg33ByoNfwg3DQsJQw2ZCUsNKQpSDbsKVg1PC1cN4wtW DXUMUg0FDUsNkw1DDR8ONw2nDioNLQ8aDbAPCA0vEPQMrBDeDCQRxgyZEawM ChKQDHcScgzfElIMQxMxDKMTDgz9E+oLUxTEC6MUnQvuFHULMxVLC3MVIAut FfMK4BXGCg0WmAo0FmgKVBY4Cm4WBwqAFtUJixaiCY8WbwmPFmsJjxZfCY8W TAmPFjMJjxYTCY8W7giPFsMIjxaUCI8WYQiPFioIjxbwB48WtAePFnYHjxY2 B48W9QaPFrMGjxZxBo8WMAaPFvAFjxayBY8WdgWPFjwFjxYFBY8W0gSPFqME jxZ4BI8WUwSPFjMEjxYaBI8WBwSPFvsDjxb3AxQAAAAMAAAADQAAACYAAAAc AAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAIA AAAlAAAADAAAAAUAAIAnAAAAGAAAAAIAAAAAAAAAAAAAAAAAAAAlAAAADAAA AAIAAAATAAAADAAAAAEAAABWAAAAPAAAAHAAAAALAAAAuAAAACMAAAAIAAAA mgsHAzkNjgM6Ee8BARN7AgwSGAFhDRgBQQ+BAZoLBwMlAAAADAAAAAUAAIAU AAAADAAAAA0AAAAmAAAAHAAAAAQAAAAAAAAABQAAAAAAAAD///8AJQAAAAwA AAAEAAAAKAAAAAwAAAADAAAAVwAAADwAAABwAAAACwAAALgAAAAiAAAACAAA AJoLBwM5DY4DOhHvAQETewIMEhgBYQ0YAUEPgQGaCwcDFAAAAAwAAAANAAAA JgAAABwAAAADAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAM AAAABAAAACUAAAAMAAAABQAAgCUAAAAMAAAAAgAAABMAAAAMAAAAAQAAAFYA AAA8AAAAcAAAAAsAAAC4AAAAIwAAAAgAAACaCwcDOQ2OAzoR7wEBE3sCDBIY AWENGAFBD4EBmgsHAyUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAA BAAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAQAAAAoAAAADAAAAAMAAABX AAAAPAAAAHAAAAALAAAAuAAAACIAAAAIAAAAmgsHAzkNjgM6Ee8BARN7AgwS GAFhDRgBQQ+BAZoLBwMUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAABQAA AAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAAEAAAAJQAAAAwAAAAFAACA JQAAAAwAAAACAAAAEwAAAAwAAAABAAAAVgAAADwAAAAiAAAAJwAAAGkAAAA/ AAAACAAAAOEKgwRMCfwDbgWbBX8DDwV5BIsGTAmLBjoH+gXhCoMEJQAAAAwA AAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAEAAAAAAAAAAUAAAAAAAAA//// ACUAAAAMAAAABAAAACgAAAAMAAAAAwAAAFcAAAA8AAAAIgAAACcAAABpAAAA PwAAAAgAAADhCoMETAn8A24FmwV/Aw8FeQSLBkwJiwY6B/oF4QqDBBQAAAAM AAAADQAAACYAAAAcAAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMA AAAoAAAADAAAAAQAAAAlAAAADAAAAAUAAIAlAAAADAAAAAIAAAATAAAADAAA AAEAAABWAAAAPAAAACIAAAAnAAAAaQAAAD8AAAAIAAAA4QqDBEwJ/ANuBZsF fwMPBXkEiwZMCYsGOgf6BeEKgwQlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAm AAAAHAAAAAQAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAEAAAAKAAAAAwA AAADAAAAVwAAADwAAAAiAAAAJwAAAGkAAAA/AAAACAAAAOEKgwRMCfwDbgWb BX8DDwV5BIsGTAmLBjoH+gXhCoMEFAAAAAwAAAANAAAAJgAAABwAAAADAAAA BgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABAAAACUAAAAM AAAABQAAgCUAAAAMAAAAAgAAABMAAAAMAAAAAQAAAFYAAAA8AAAAJgAAAAoA AABuAAAAIQAAAAgAAADjA4EBjAX6AI0JewJZCw0CXwprA68FawOjBwcD4wOB ASUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAABAAAAAAAAAAFAAAA AAAAAP///wAlAAAADAAAAAQAAAAoAAAADAAAAAMAAABXAAAAPAAAACYAAAAJ AAAAbgAAACEAAAAIAAAA4wOBAYwF+gCNCXsCWQsNAl8KawOvBWsDowcHA+MD gQEUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAABQAAAAAAAAAAAAAAJQAA AAwAAAADAAAAKAAAAAwAAAAEAAAAJQAAAAwAAAAFAACAJQAAAAwAAAACAAAA EwAAAAwAAAABAAAAVgAAADwAAAAmAAAACgAAAG4AAAAhAAAACAAAAOMDgQGM BfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EBJQAAAAwAAAAFAACAFAAAAAwA AAANAAAAJgAAABwAAAAEAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABAAA ACgAAAAMAAAAAwAAAFcAAAA8AAAAJgAAAAkAAABuAAAAIQAAAAgAAADjA4EB jAX6AI0JewJZCw0CXwprA68FawOjBwcD4wOBARQAAAAMAAAADQAAACYAAAAc AAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAQA AAAlAAAADAAAAAUAAIAlAAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAPAAA AG4AAAApAAAAtQAAAEEAAAAIAAAAuxInBhwRrgY5DQ8FWQubBT8MOAQcETgE AA+hBLsSJwYlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAA AAAABQAAAAAAAAD///8AJQAAAAwAAAAEAAAAKAAAAAwAAAADAAAAVwAAADwA AABuAAAAKQAAALUAAABBAAAACAAAALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4 BAAPoQS7EicGFAAAAAwAAAANAAAAJgAAABwAAAADAAAABgAAAAUAAAAAAAAA AAAAACUAAAAMAAAAAwAAACgAAAAMAAAABAAAACUAAAAMAAAABQAAgCUAAAAM AAAAAgAAABMAAAAMAAAAAQAAAFYAAAA8AAAAbgAAACkAAAC1AAAAQQAAAAgA AAC7EicGHBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxInBiUAAAAMAAAABQAA gBQAAAAMAAAADQAAACYAAAAcAAAABAAAAAAAAAAFAAAAAAAAAP///wAlAAAA DAAAAAQAAAAoAAAADAAAAAMAAABXAAAAPAAAAG4AAAApAAAAtQAAAEEAAAAI AAAAuxInBhwRrgY5DQ8FWQubBT8MOAQcETgEAA+hBLsSJwYUAAAADAAAAA0A AAAmAAAAHAAAAAMAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAA AAwAAAAEAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAP///wAAAAAA JQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADwAAAByAAAADAAAALkAAAAk AAAACAAAAL0LJQNhDawDXRENAh8TmQIvEkABfw1AAX0PnwG9CyUDJQAAAAwA AAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAFAAAAAAAAAAUAAAAAAAAA//// ACUAAAAMAAAABQAAACgAAAAMAAAAAwAAAFcAAAA8AAAAcgAAAAwAAAC5AAAA JAAAAAgAAAC9CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9D58BvQslAxQAAAAM AAAADQAAACYAAAAcAAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMA AAAoAAAADAAAAAUAAAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAA AAEAAABWAAAAPAAAAHIAAAAMAAAAuQAAACQAAAAIAAAAvQslA2ENrANdEQ0C HxOZAi8SQAF/DUABfQ+fAb0LJQMlAAAADAAAAAUAAIAUAAAADAAAAA0AAABX AAAAPAAAAHIAAAAMAAAAuQAAACQAAAAIAAAAvQslA2ENrANdEQ0CHxOZAi8S QAF/DUABfQ+fAb0LJQMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAlAAAADAAA AAQAAAATAAAADAAAAAEAAABWAAAAPAAAACMAAAAoAAAAawAAAEEAAAAIAAAA DguhBG8JGgSMBbkFogMtBZwErgZvCa4GXQcnBg4LoQQlAAAADAAAAAUAAIAU AAAADAAAAA0AAAAmAAAAHAAAAAUAAAAAAAAABQAAAAAAAAD///8AJQAAAAwA AAAFAAAAKAAAAAwAAAADAAAAVwAAADwAAAAjAAAAKAAAAGsAAABBAAAACAAA AA4LoQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEFAAAAAwAAAANAAAA JgAAABwAAAADAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAM AAAABQAAACUAAAAMAAAABQAAgCUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYA AAA8AAAAIwAAACgAAABrAAAAQQAAAAgAAAAOC6EEbwkaBIwFuQWiAy0FnASu Bm8JrgZdBycGDguhBCUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAA8AAAA IwAAACgAAABrAAAAQQAAAAgAAAAOC6EEbwkaBIwFuQWiAy0FnASuBm8JrgZd BycGDguhBBQAAAAMAAAADQAAACUAAAAMAAAABQAAgCUAAAAMAAAABAAAABMA AAAMAAAAAQAAAFYAAAA8AAAAKAAAAAsAAABvAAAAIwAAAAgAAAAaBJ8BrwUY AbAJmQJ8CysCggqOA9wFjgPBByUDGgSfASUAAAAMAAAABQAAgBQAAAAMAAAA DQAAACYAAAAcAAAABQAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAUAAAAo AAAADAAAAAMAAABXAAAAPAAAACgAAAALAAAAbwAAACIAAAAIAAAAGgSfAa8F GAGwCZkCfAsrAoIKjgPcBY4DwQclAxoEnwEUAAAADAAAAA0AAAAmAAAAHAAA AAMAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAAFAAAA JQAAAAwAAAAFAACAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADwAAAAo AAAACwAAAG8AAAAjAAAACAAAABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EH JQMaBJ8BJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAADwAAAAoAAAACwAA AG8AAAAiAAAACAAAABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EHJQMaBJ8B FAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJQAAAAwAAAAEAAAAEwAAAAwAAAAB AAAAVgAAADwAAABvAAAAKgAAALcAAABCAAAACAAAAOMSRQY6EcwGYQ0tBXwL uQVsDGAEOhFgBCMPvwTjEkUGJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAA ABwAAAAFAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABQAAACgAAAAMAAAA AwAAAFcAAAA8AAAAbwAAACoAAAC3AAAAQgAAAAgAAADjEkUGOhHMBmENLQV8 C7kFbAxgBDoRYAQjD78E4xJFBhQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYA AAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAUAAAAlAAAADAAA AAUAAIAlAAAADAAAAAQAAAATAAAADAAAAAEAAABWAAAAPAAAAG8AAAAqAAAA twAAAEIAAAAIAAAA4xJFBjoRzAZhDS0FfAu5BWwMYAQ6EWAEIw+/BOMSRQYl AAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAPAAAAG8AAAAqAAAAtwAAAEIA AAAIAAAA4xJFBjoRzAZhDS0FfAu5BWwMYAQ6EWAEIw+/BOMSRQYUAAAADAAA AA0AAAAiAAAADAAAAP////8oAAAADAAAAAEAAAAlAAAADAAAAAcAAIAlAAAA DAAAAAAAAIAwAAAADAAAAA8AAIBLAAAAEAAAAAAAAAAFAAAAKAAAAAwAAAAC AAAAKAAAAAwAAAADAAAAKAAAAAwAAAAEAAAADgAAABQAAAAAAAAAEAAAABQA AAAaQAFALAAAACAAAAD/////AgAAAAAAAAAAAAAAP2FcQx8zA0MAAAAA6APo AyEAAAAIAAAAYgAAAAwAAAABAAAAIQAAAAgAAAAhAAAACAAAAHMAAAAMAAAA AAAAABwAAAAIAAAAJQAAAAwAAAAAAACAJQAAAAwAAAAHAACAJQAAAAwAAAAO AACAMAAAAAwAAAAPAACAGQAAAAwAAAD///8AGAAAAAwAAAAAAAAAEgAAAAwA AAACAAAAEwAAAAwAAAABAAAAFAAAAAwAAAANAAAAFQAAAAwAAAABAAAAFgAA AAwAAAAAAAAADQAAABAAAAAAAAAAAAAAADoAAAAMAAAACgAAABsAAAAQAAAA AAAAAAAAAAAjAAAAIAAAAIUN7j4AAAAAAAAAAPvx6z4AAAJDAAC7QyQAAAAk AAAAhQ3uPgAAAAAAAAAA+/HrPgAAAkMAALtDBAAAAHMAAAAMAAAAAAAAAA0A AAAQAAAAggAAAHYBAABSAAAAcAEAAAEAAAAUAAAACgAAAAAAAAAAAAAAkAEA AAAAAIEHAgIiUwB5AHMAdABlAG0AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACAAAAAA AAAAAAAAAEhl8XfcZPF3CQAAACEAigENAAAAlN8SAIU08nfQJGQDBBoh3tAk ZAMmAIoBBBoh3gsAiAH//wEABAAAANzcEgCk3RIA0N0SABjukHxwBZF8//// /20FkXxamYB8AAAUAAAAAABtmYB8AAAQxZBIWQMgMPV3JQAAALBGFQAAABAA sEYVAOTdEgCbKPJ3nMchALBGFQDQJGQDR0RJQwQaId4hAIoBCAAAAAAAEgBQ S1kD+L9sA9B4GgATAAAA0HgaANB4GgD4v2wD+L9sAyjeEgDBEvJ30HgaABMA AABQS1kD0HgaADjeEgBkdgAIAAAAACUAAAAMAAAAAQAAAEYAAAAoAAAAHAAA AEdESUMCAAAAAAAAAAAAAADbAAAAggAAAAAAAAAjAAAAIAAAAIUN7j4AAAAA AAAAAPvx6z4AAAJDAAC7QyQAAAAkAAAAhQ3uPgAAAAAAAAAA+/HrPgAAAkMA ALtDBAAAACMAAAAgAAAAkaXLQgAAAAAAAAAAw6FvQgAAAkMAALtDJAAAACQA AACRpctCAAAAAAAAAADDoW9CAAACQwAAu0MEAAAAIwAAACAAAACFDe4+AAAA AAAAAAD78es+AAACQwAAu0MkAAAAJAAAAIUN7j4AAAAAAAAAAPvx6z4AAAJD AAC7QwQAAAAjAAAAIAAAAO8DkDwAAAAAAAAAAEP/jjwAAAJDAAC7QyQAAAAk AAAA7wOQPAAAAAAAAAAAQ/+OPAAAAkMAALtDBAAAACMAAAAgAAAA7wOQPAAA AAAAAAAAQ/+OPAAAAkMAALtDJAAAACQAAADvA5A8AAAAAAAAAABD/448AAAC QwAAu0MEAAAAIwAAACAAAADvA5A8AAAAAAAAAABD/448AAACQwAAu0MkAAAA JAAAAO8DkDwAAAAAAAAAAEP/jjwAAAJDAAC7QwQAAAAhAAAACAAAABQAAAAM AAAADQAAACUAAAAMAAAABQAAgCYAAAAcAAAABQAAAAAAAAAAAAAAAAAAAAAA AAAlAAAADAAAAAUAAAAmAAAAHAAAAAQAAAAGAAAAHgAAAAAAAAAAAAAAJQAA AAwAAAAEAAAAJQAAAAwAAAAFAACAJwAAABgAAAADAAAAAAAAAL+/vwAAAAAA JQAAAAwAAAADAAAAEwAAAAwAAAABAAAAVgAAACACAACCAAAAdgEAAOgAAACa AQAAgQAAAI8W9wOLFsQDgBaRA24WXwNVFi4DNBb9Ag4WzgLgFZ8CrRVyAnMV RgIzFRoC7hTxAaMUyAFTFKEB/RN7AaMTVwFEEzQB4BITAXcS8wAKEtYAmRG6 ACURoACsEIgAMBByALAPXgAtD0wAqA48AB8OLwCUDSMABg0bAHYMFADjCxAA TwsPALsKEAAoChQAmAkbAAoJIwB/CC8A9gc8AHEHTADuBl4AbgZyAPIFiAB5 BaAABQW6AJQE1gAnBPMAvgMTAVoDNAH7AlcBoQJ7AUsCoQH7AcgBsAHxAWsB GgIrAUYC8QByAr4AnwKQAM4CagD9AkkALgMwAF8DHgCRAxMAxAMPAPcDEwAq BB4AXQQwAI8ESQDABGoA8QSQACAFvgBPBfEAfAUrAagFawHUBbAB/QX7ASYG SwJNBqECcwb7ApcGWgO6Br4D2wYnBPsGlAQYBwUFNAd5BU4H8gVmB24GfAfu BpAHcQeiB/YHsgd/CL8HCgnLB5gJ0wcoCtoHuwreB08L3wfjC94HdgzaBwYN 0weUDcsHHw6/B6gOsgctD6IHsA+QBzAQfAesEGYHJRFOB5kRNAcKEhgHdxL7 BuAS2wZEE7oGoxOXBv0TcwZTFE0GoxQmBu4U/QUzFdQFcxWoBa0VfAXgFU8F DhYgBTQW8QRVFsAEbhaPBIAWXQSLFioEjxb3AyUAAAAMAAAABQAAgCgAAAAM AAAAAwAAABQAAAAMAAAADQAAAFcAAAAgAgAAggAAAHYBAADoAAAAmgEAAIEA AACPFvcDixbEA4AWkQNuFl8DVRYuAzQW/QIOFs4C4BWfAq0VcgJzFUYCMxUa Au4U8QGjFMgBUxShAf0TewGjE1cBRBM0AeASEwF3EvMAChLWAJkRugAlEaAA rBCIADAQcgCwD14ALQ9MAKgOPAAfDi8AlA0jAAYNGwB2DBQA4wsQAE8LDwC7 ChAAKAoUAJgJGwAKCSMAfwgvAPYHPABxB0wA7gZeAG4GcgDyBYgAeQWgAAUF ugCUBNYAJwTzAL4DEwFaAzQB+wJXAaECewFLAqEB+wHIAbAB8QFrARoCKwFG AvEAcgK+AJ8CkADOAmoA/QJJAC4DMABfAx4AkQMTAMQDDwD3AxMAKgQeAF0E MACPBEkAwARqAPEEkAAgBb4ATwXxAHwFKwGoBWsB1AWwAf0F+wEmBksCTQah AnMG+wKXBloDuga+A9sGJwT7BpQEGAcFBTQHeQVOB/IFZgduBnwH7gaQB3EH ogf2B7IHfwi/BwoJyweYCdMHKAraB7sK3gdPC98H4wveB3YM2gcGDdMHlA3L Bx8OvweoDrIHLQ+iB7APkAcwEHwHrBBmByURTgeZETQHChIYB3cS+wbgEtsG RBO6BqMTlwb9E3MGUxRNBqMUJgbuFP0FMxXUBXMVqAWtFXwF4BVPBQ4WIAU0 FvEEVRbABG4WjwSAFl0EixYqBI8W9wMUAAAADAAAAA0AAAAlAAAADAAAAAUA AIAnAAAAGAAAAAMAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAMAAAATAAAADAAA AAEAAABWAAAAIAMAAIIAAACHAQAA6AAAALIBAADBAAAAjxb3A4sWKgSAFl0E bhaPBFQWwAQ0FvAEDRYgBeAVTgWtFXsFcxWoBTMV0wXuFP0FoxQlBlMUTAb9 E3IGoxOWBkMTuQbfEtoGdxL6BgoSGAeZETQHJBFOB6wQZgcvEHwHsA+QBy0P ogenDrIHHw6/B5MNywcFDdMHdQzaB+ML3gdPC98HuwreBykK2geZCdMHCwnL B38Ivwf3B7IHcQeiB+4GkAdvBnwH8gVmB3oFTgcFBTQHlAQYBycE+ga/A9oG WwO5BvsClgahAnIGSwJMBvsBJQawAf0FawHTBSsBqAXxAHsFvgBOBZEAIAVq APAESgDABDAAjwQeAF0EEwAqBA8A9wMPAPsDDwAHBA8AGgQPADMEDwBTBA8A eAQPAKMEDwDSBA8ABQUPADwFDwB2BQ8AsgUPAPAFDwAwBg8AcQYPALMGDwD1 Bg8ANgcPAHYHDwC0Bw8A8AcPACoIDwBhCA8AlAgPAMMIDwDuCA8AEwkPADMJ DwBMCQ8AXwkPAGsJDwBvCRMAogkeANUJMAAHCkoAOApqAGgKkQCYCr4Axgrx APMKKwEgC2sBSwuwAXUL+wGdC0sCxAuhAuoL+wIODFsDMQy/A1IMJwRyDJQE kAwFBawMegXGDPIF3gxvBvQM7gYIDXEHGg33ByoNfwg3DQsJQw2ZCUsNKQpS DbsKVg1PC1cN4wtWDXUMUg0FDUsNkw1DDR8ONw2nDioNLQ8aDbAPCA0vEPQM rBDeDCQRxgyZEawMChKQDHcScgzfElIMQxMxDKMTDgz9E+oLUxTEC6MUnQvu FHULMxVLC3MVIAutFfMK4BXGCg0WmAo0FmgKVBY4Cm4WBwqAFtUJixaiCY8W bwmPFmsJjxZfCY8WTAmPFjMJjxYTCY8W7giPFsMIjxaUCI8WYQiPFioIjxbw B48WtAePFnYHjxY2B48W9QaPFrMGjxZxBo8WMAaPFvAFjxayBY8WdgWPFjwF jxYFBY8W0gSPFqMEjxZ4BI8WUwSPFjMEjxYaBI8WBwSPFvsDjxb3AyUAAAAM AAAABQAAgCgAAAAMAAAAAwAAABQAAAAMAAAADQAAAFcAAAAgAwAAggAAAIcB AADoAAAAsgEAAMEAAACPFvcDixYqBIAWXQRuFo8EVBbABDQW8AQNFiAF4BVO Ba0VewVzFagFMxXTBe4U/QWjFCUGUxRMBv0TcgajE5YGQxO5Bt8S2gZ3EvoG ChIYB5kRNAckEU4HrBBmBy8QfAewD5AHLQ+iB6cOsgcfDr8Hkw3LBwUN0wd1 DNoH4wveB08L3we7Ct4HKQraB5kJ0wcLCcsHfwi/B/cHsgdxB6IH7gaQB28G fAfyBWYHegVOBwUFNAeUBBgHJwT6Br8D2gZbA7kG+wKWBqECcgZLAkwG+wEl BrAB/QVrAdMFKwGoBfEAewW+AE4FkQAgBWoA8ARKAMAEMACPBB4AXQQTACoE DwD3Aw8A+wMPAAcEDwAaBA8AMwQPAFMEDwB4BA8AowQPANIEDwAFBQ8APAUP AHYFDwCyBQ8A8AUPADAGDwBxBg8AswYPAPUGDwA2Bw8AdgcPALQHDwDwBw8A KggPAGEIDwCUCA8AwwgPAO4IDwATCQ8AMwkPAEwJDwBfCQ8AawkPAG8JEwCi CR4A1QkwAAcKSgA4CmoAaAqRAJgKvgDGCvEA8worASALawFLC7ABdQv7AZ0L SwLEC6EC6gv7Ag4MWwMxDL8DUgwnBHIMlASQDAUFrAx6BcYM8gXeDG8G9Azu BggNcQcaDfcHKg1/CDcNCwlDDZkJSw0pClINuwpWDU8LVw3jC1YNdQxSDQUN Sw2TDUMNHw43DacOKg0tDxoNsA8IDS8Q9AysEN4MJBHGDJkRrAwKEpAMdxJy DN8SUgxDEzEMoxMODP0T6gtTFMQLoxSdC+4UdQszFUsLcxUgC60V8wrgFcYK DRaYCjQWaApUFjgKbhYHCoAW1QmLFqIJjxZvCY8WawmPFl8JjxZMCY8WMwmP FhMJjxbuCI8WwwiPFpQIjxZhCI8WKgiPFvAHjxa0B48WdgePFjYHjxb1Bo8W swaPFnEGjxYwBo8W8AWPFrIFjxZ2BY8WPAWPFgUFjxbSBI8WowSPFngEjxZT BI8WMwSPFhoEjxYHBI8W+wOPFvcDFAAAAAwAAAANAAAAJgAAABwAAAADAAAA BgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABAAAACUAAAAM AAAABQAAgCcAAAAYAAAABAAAAAAAAAAAAAAAAAAAACUAAAAMAAAABAAAABMA AAAMAAAAAQAAAFYAAAA8AAAAtgAAAHoBAADYAAAAhgEAAAgAAACaCwcDOQ2O AzoR7wEBE3sCDBIYAWENGAFBD4EBmgsHAyUAAAAMAAAABQAAgBQAAAAMAAAA DQAAACYAAAAcAAAAAgAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAIAAAAo AAAADAAAAAMAAABXAAAAPAAAALYAAAB6AQAA2AAAAIYBAAAIAAAAmgsHAzkN jgM6Ee8BARN7AgwSGAFhDRgBQQ+BAZoLBwMUAAAADAAAAA0AAAAmAAAAHAAA AAMAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAACAAAA JQAAAAwAAAAFAACAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADwAAAC2 AAAAegEAANgAAACGAQAACAAAAJoLBwM5DY4DOhHvAQETewIMEhgBYQ0YAUEP gQGaCwcDJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAACAAAAAAAA AAUAAAAAAAAA////ACUAAAAMAAAAAgAAACgAAAAMAAAAAwAAAFcAAAA8AAAA tgAAAHoBAADYAAAAhgEAAAgAAACaCwcDOQ2OAzoR7wEBE3sCDBIYAWENGAFB D4EBmgsHAxQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYAAAAFAAAAAAAAAAAA AAAlAAAADAAAAAMAAAAoAAAADAAAAAIAAAAlAAAADAAAAAUAAIAlAAAADAAA AAQAAAATAAAADAAAAAEAAABWAAAAPAAAAJEAAACHAQAAswAAAJQBAAAIAAAA 4QqDBEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6BeEKgwQlAAAADAAAAAUAAIAU AAAADAAAAA0AAAAmAAAAHAAAAAIAAAAAAAAABQAAAAAAAAD///8AJQAAAAwA AAACAAAAKAAAAAwAAAADAAAAVwAAADwAAACRAAAAhwEAALMAAACUAQAACAAA AOEKgwRMCfwDbgWbBX8DDwV5BIsGTAmLBjoH+gXhCoMEFAAAAAwAAAANAAAA JgAAABwAAAADAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAM AAAAAgAAACUAAAAMAAAABQAAgCUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYA AAA8AAAAkQAAAIcBAACzAAAAlAEAAAgAAADhCoMETAn8A24FmwV/Aw8FeQSL BkwJiwY6B/oF4QqDBCUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAA AgAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAIAAAAoAAAADAAAAAMAAABX AAAAPAAAAJEAAACHAQAAswAAAJQBAAAIAAAA4QqDBEwJ/ANuBZsFfwMPBXkE iwZMCYsGOgf6BeEKgwQUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAABQAA AAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAACAAAAJQAAAAwAAAAFAACA JQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADwAAACTAAAAegEAALYAAACG AQAACAAAAOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EBJQAAAAwA AAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAACAAAAAAAAAAUAAAAAAAAA//// ACUAAAAMAAAAAgAAACgAAAAMAAAAAwAAAFcAAAA8AAAAkwAAAHoBAAC2AAAA hgEAAAgAAADjA4EBjAX6AI0JewJZCw0CXwprA68FawOjBwcD4wOBARQAAAAM AAAADQAAACYAAAAcAAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMA AAAoAAAADAAAAAIAAAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAA AAEAAABWAAAAPAAAAJMAAAB6AQAAtgAAAIYBAAAIAAAA4wOBAYwF+gCNCXsC WQsNAl8KawOvBWsDowcHA+MDgQElAAAADAAAAAUAAIAUAAAADAAAAA0AAAAm AAAAHAAAAAIAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAACAAAAKAAAAAwA AAADAAAAVwAAADwAAACTAAAAegEAALYAAACGAQAACAAAAOMDgQGMBfoAjQl7 AlkLDQJfCmsDrwVrA6MHBwPjA4EBFAAAAAwAAAANAAAAJgAAABwAAAADAAAA BgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAAAgAAACUAAAAM AAAABQAAgCUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAA8AAAAtQAAAIgB AADXAAAAlAEAAAgAAAC7EicGHBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxIn BiUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAAAgAAAAAAAAAFAAAA AAAAAP///wAlAAAADAAAAAIAAAAoAAAADAAAAAMAAABXAAAAPAAAALUAAACI AQAA1wAAAJQBAAAIAAAAuxInBhwRrgY5DQ8FWQubBT8MOAQcETgEAA+hBLsS JwYUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAABQAAAAAAAAAAAAAAJQAA AAwAAAADAAAAKAAAAAwAAAACAAAAJQAAAAwAAAAFAACAJQAAAAwAAAAEAAAA EwAAAAwAAAABAAAAVgAAADwAAAC1AAAAiAEAANcAAACUAQAACAAAALsSJwYc Ea4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGJQAAAAwAAAAFAACAFAAAAAwA AAANAAAAJgAAABwAAAACAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAAAgAA ACgAAAAMAAAAAwAAAFcAAAA8AAAAtQAAAIgBAADXAAAAlAEAAAgAAAC7EicG HBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxInBhQAAAAMAAAADQAAACYAAAAc AAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAIA AAAlAAAADAAAAAUAAIAnAAAAGAAAAAIAAAAAAAAA////AAAAAAAlAAAADAAA AAIAAAATAAAADAAAAAEAAABWAAAAPAAAALYAAAB7AQAA2QAAAIcBAAAIAAAA vQslA2ENrANdEQ0CHxOZAi8SQAF/DUABfQ+fAb0LJQMlAAAADAAAAAUAAIAU AAAADAAAAA0AAAAmAAAAHAAAAAYAAAAAAAAABQAAAAAAAAD///8AJQAAAAwA AAAGAAAAKAAAAAwAAAADAAAAVwAAADwAAAC2AAAAewEAANkAAACHAQAACAAA AL0LJQNhDawDXRENAh8TmQIvEkABfw1AAX0PnwG9CyUDFAAAAAwAAAANAAAA JgAAABwAAAADAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAM AAAABgAAACUAAAAMAAAABQAAgCUAAAAMAAAAAgAAABMAAAAMAAAAAQAAAFYA AAA8AAAAtgAAAHsBAADZAAAAhwEAAAgAAAC9CyUDYQ2sA10RDQIfE5kCLxJA AX8NQAF9D58BvQslAyUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAA8AAAA tgAAAHsBAADZAAAAhwEAAAgAAAC9CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9 D58BvQslAxQAAAAMAAAADQAAACUAAAAMAAAABQAAgCUAAAAMAAAAAgAAABMA AAAMAAAAAQAAAFYAAAA8AAAAkgAAAIgBAAC0AAAAlAEAAAgAAAAOC6EEbwka BIwFuQWiAy0FnASuBm8JrgZdBycGDguhBCUAAAAMAAAABQAAgBQAAAAMAAAA DQAAACYAAAAcAAAABgAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAYAAAAo AAAADAAAAAMAAABXAAAAPAAAAJIAAACIAQAAtAAAAJQBAAAIAAAADguhBG8J GgSMBbkFogMtBZwErgZvCa4GXQcnBg4LoQQUAAAADAAAAA0AAAAmAAAAHAAA AAMAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAAGAAAA JQAAAAwAAAAFAACAJQAAAAwAAAACAAAAEwAAAAwAAAABAAAAVgAAADwAAACS AAAAiAEAALQAAACUAQAACAAAAA4LoQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0H JwYOC6EEJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAADwAAACSAAAAiAEA ALQAAACUAQAACAAAAA4LoQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EE FAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJQAAAAwAAAACAAAAEwAAAAwAAAAB AAAAVgAAADwAAACUAAAAegEAALYAAACGAQAACAAAABoEnwGvBRgBsAmZAnwL KwKCCo4D3AWOA8EHJQMaBJ8BJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAA ABwAAAAGAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABgAAACgAAAAMAAAA AwAAAFcAAAA8AAAAlAAAAHoBAAC2AAAAhgEAAAgAAAAaBJ8BrwUYAbAJmQJ8 CysCggqOA9wFjgPBByUDGgSfARQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYA AAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAYAAAAlAAAADAAA AAUAAIAlAAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAPAAAAJQAAAB6AQAA tgAAAIYBAAAIAAAAGgSfAa8FGAGwCZkCfAsrAoIKjgPcBY4DwQclAxoEnwEl AAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAPAAAAJQAAAB6AQAAtgAAAIYB AAAIAAAAGgSfAa8FGAGwCZkCfAsrAoIKjgPcBY4DwQclAxoEnwEUAAAADAAA AA0AAAAlAAAADAAAAAUAAIAlAAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAA PAAAALUAAACJAQAA1wAAAJUBAAAIAAAA4xJFBjoRzAZhDS0FfAu5BWwMYAQ6 EWAEIw+/BOMSRQYlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAYA AAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAGAAAAKAAAAAwAAAADAAAAVwAA ADwAAAC1AAAAiQEAANcAAACVAQAACAAAAOMSRQY6EcwGYQ0tBXwLuQVsDGAE OhFgBCMPvwTjEkUGFAAAAAwAAAANAAAAJgAAABwAAAADAAAABgAAAAUAAAAA AAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABgAAACUAAAAMAAAABQAAgCUA AAAMAAAAAgAAABMAAAAMAAAAAQAAAFYAAAA8AAAAtQAAAIkBAADXAAAAlQEA AAgAAADjEkUGOhHMBmENLQV8C7kFbAxgBDoRYAQjD78E4xJFBiUAAAAMAAAA BQAAgBQAAAAMAAAADQAAAFcAAAA8AAAAtQAAAIkBAADXAAAAlQEAAAgAAADj EkUGOhHMBmENLQV8C7kFbAxgBDoRYAQjD78E4xJFBhQAAAAMAAAADQAAACIA AAAMAAAA/////ygAAAAMAAAABQAAACUAAAAMAAAABwAAgCUAAAAMAAAAAAAA gEsAAAAQAAAAAAAAAAUAAAAoAAAADAAAAAQAAAAoAAAADAAAAAMAAAAoAAAA DAAAAAIAAABGAAAAFAAAAAgAAABHRElDAwAAACIAAAAMAAAA/////yIAAAAM AAAA/////yUAAAAMAAAADQAAgCgAAAAMAAAAAQAAACIAAAAMAAAA/////0YA AABINgAAPDYAAEVNRismQAAAEAAAAAQAAAAAAAAAJUAAABAAAAAEAAAAAAAA ACxAAAAkAAAAGAAAAMqN0z0AAACAAAAAgDEcez1zCgdEndi6QwhAAgXINQAA vDUAAAEQwNsCAAAAAwAAAKw1AAABAAAAbAAAAAEAAAABAAAA2gAAAIEAAAAA AAAAAAAAAKEWAABpDQAAIEVNRgAAAQCsNQAA+gAAAAYAAAAAAAAAAAAAAAAA AAAABQAAAAQAAFIBAAAOAQAAAAAAAAAAAAAAAAAAUCgFALAeBABGAAAATBcA AD4XAABHRElDAQAAgAADAACtn7DzAAAAACYXAAABAAkAAAOTCwAABgCGAQAA AAAFAAAADAJoDaAWBQAAAAwCaA2gFgcAAAAQBAEAAQABAAEAAwAAAB4ABAAA AAQBDQAHAAAA/AIBAAAAAAAAAAQAAAAtAQAACQAAAPoCAAAAAAAAAAAAACIA BAAAAC0BAQAJAAAA+gIGAB4AAAAAAAAAIgAEAAAALQECAAQAAAAtAQAABwAA APwCAAC/v78AAAAEAAAALQEDAAQAAAAGAQEABgEAACQDgQCPFvcDixbEA4AW kQNuFl8DVRYuAzQW/QIOFs4C4BWfAq0VcgJzFUYCMxUaAu4U8QGjFMgBUxSh Af0TewGjE1cBRBM0AeASEwF3EvMAChLWAJkRugAlEaAArBCIADAQcgCwD14A LQ9MAKgOPAAfDi8AlA0jAAYNGwB2DBQA4wsQAE8LDwC7ChAAKAoUAJgJGwAK CSMAfwgvAPYHPABxB0wA7gZeAG4GcgDyBYgAeQWgAAUFugCUBNYAJwTzAL4D EwFaAzQB+wJXAaECewFLAqEB+wHIAbAB8QFrARoCKwFGAvEAcgK+AJ8CkADO AmoA/QJJAC4DMABfAx4AkQMTAMQDDwD3AxMAKgQeAF0EMACPBEkAwARqAPEE kAAgBb4ATwXxAHwFKwGoBWsB1AWwAf0F+wEmBksCTQahAnMG+wKXBloDuga+ A9sGJwT7BpQEGAcFBTQHeQVOB/IFZgduBnwH7gaQB3EHogf2B7IHfwi/BwoJ yweYCdMHKAraB7sK3gdPC98H4wveB3YM2gcGDdMHlA3LBx8OvweoDrIHLQ+i B7APkAcwEHwHrBBmByURTgeZETQHChIYB3cS+wbgEtsGRBO6BqMTlwb9E3MG UxRNBqMUJgbuFP0FMxXUBXMVqAWtFXwF4BVPBQ4WIAU0FvEEVRbABG4WjwSA Fl0EixYqBI8W9wMEAAAALQEAAAQAAADwAQMABAAAAAQBDQAGAQAAJQOBAI8W 9wOLFsQDgBaRA24WXwNVFi4DNBb9Ag4WzgLgFZ8CrRVyAnMVRgIzFRoC7hTx AaMUyAFTFKEB/RN7AaMTVwFEEzQB4BITAXcS8wAKEtYAmRG6ACURoACsEIgA MBByALAPXgAtD0wAqA48AB8OLwCUDSMABg0bAHYMFADjCxAATwsPALsKEAAo ChQAmAkbAAoJIwB/CC8A9gc8AHEHTADuBl4AbgZyAPIFiAB5BaAABQW6AJQE 1gAnBPMAvgMTAVoDNAH7AlcBoQJ7AUsCoQH7AcgBsAHxAWsBGgIrAUYC8QBy Ar4AnwKQAM4CagD9AkkALgMwAF8DHgCRAxMAxAMPAPcDEwAqBB4AXQQwAI8E SQDABGoA8QSQACAFvgBPBfEAfAUrAagFawHUBbAB/QX7ASYGSwJNBqECcwb7 ApcGWgO6Br4D2wYnBPsGlAQYBwUFNAd5BU4H8gVmB24GfAfuBpAHcQeiB/YH sgd/CL8HCgnLB5gJ0wcoCtoHuwreB08L3wfjC94HdgzaBwYN0weUDcsHHw6/ B6gOsgctD6IHsA+QBzAQfAesEGYHJRFOB5kRNAcKEhgHdxL7BuAS2wZEE7oG oxOXBv0TcwZTFE0GoxQmBu4U/QUzFdQFcxWoBa0VfAXgFU8FDhYgBTQW8QRV FsAEbhaPBIAWXQSLFioEjxb3AwQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/ vwAAAAQAAAAtAQMABAAAAAYBAQCGAQAAJAPBAI8W9wOLFioEgBZdBG4WjwRU FsAENBbwBA0WIAXgFU4FrRV7BXMVqAUzFdMF7hT9BaMUJQZTFEwG/RNyBqMT lgZDE7kG3xLaBncS+gYKEhgHmRE0ByQRTgesEGYHLxB8B7APkActD6IHpw6y Bx8OvweTDcsHBQ3TB3UM2gfjC94HTwvfB7sK3gcpCtoHmQnTBwsJywd/CL8H 9weyB3EHogfuBpAHbwZ8B/IFZgd6BU4HBQU0B5QEGAcnBPoGvwPaBlsDuQb7 ApYGoQJyBksCTAb7ASUGsAH9BWsB0wUrAagF8QB7Bb4ATgWRACAFagDwBEoA wAQwAI8EHgBdBBMAKgQPAPcDDwD7Aw8ABwQPABoEDwAzBA8AUwQPAHgEDwCj BA8A0gQPAAUFDwA8BQ8AdgUPALIFDwDwBQ8AMAYPAHEGDwCzBg8A9QYPADYH DwB2Bw8AtAcPAPAHDwAqCA8AYQgPAJQIDwDDCA8A7ggPABMJDwAzCQ8ATAkP AF8JDwBrCQ8AbwkTAKIJHgDVCTAABwpKADgKagBoCpEAmAq+AMYK8QDzCisB IAtrAUsLsAF1C/sBnQtLAsQLoQLqC/sCDgxbAzEMvwNSDCcEcgyUBJAMBQWs DHoFxgzyBd4Mbwb0DO4GCA1xBxoN9wcqDX8INw0LCUMNmQlLDSkKUg27ClYN TwtXDeMLVg11DFINBQ1LDZMNQw0fDjcNpw4qDS0PGg2wDwgNLxD0DKwQ3gwk EcYMmRGsDAoSkAx3EnIM3xJSDEMTMQyjEw4M/RPqC1MUxAujFJ0L7hR1CzMV SwtzFSALrRXzCuAVxgoNFpgKNBZoClQWOApuFgcKgBbVCYsWogmPFm8JjxZr CY8WXwmPFkwJjxYzCY8WEwmPFu4IjxbDCI8WlAiPFmEIjxYqCI8W8AePFrQH jxZ2B48WNgePFvUGjxazBo8WcQaPFjAGjxbwBY8WsgWPFnYFjxY8BY8WBQWP FtIEjxajBI8WeASPFlMEjxYzBI8WGgSPFgcEjxb7A48W9wMEAAAALQEAAAQA AADwAQMABAAAAAQBDQCGAQAAJQPBAI8W9wOLFioEgBZdBG4WjwRUFsAENBbw BA0WIAXgFU4FrRV7BXMVqAUzFdMF7hT9BaMUJQZTFEwG/RNyBqMTlgZDE7kG 3xLaBncS+gYKEhgHmRE0ByQRTgesEGYHLxB8B7APkActD6IHpw6yBx8OvweT DcsHBQ3TB3UM2gfjC94HTwvfB7sK3gcpCtoHmQnTBwsJywd/CL8H9weyB3EH ogfuBpAHbwZ8B/IFZgd6BU4HBQU0B5QEGAcnBPoGvwPaBlsDuQb7ApYGoQJy BksCTAb7ASUGsAH9BWsB0wUrAagF8QB7Bb4ATgWRACAFagDwBEoAwAQwAI8E HgBdBBMAKgQPAPcDDwD7Aw8ABwQPABoEDwAzBA8AUwQPAHgEDwCjBA8A0gQP AAUFDwA8BQ8AdgUPALIFDwDwBQ8AMAYPAHEGDwCzBg8A9QYPADYHDwB2Bw8A tAcPAPAHDwAqCA8AYQgPAJQIDwDDCA8A7ggPABMJDwAzCQ8ATAkPAF8JDwBr CQ8AbwkTAKIJHgDVCTAABwpKADgKagBoCpEAmAq+AMYK8QDzCisBIAtrAUsL sAF1C/sBnQtLAsQLoQLqC/sCDgxbAzEMvwNSDCcEcgyUBJAMBQWsDHoFxgzy Bd4Mbwb0DO4GCA1xBxoN9wcqDX8INw0LCUMNmQlLDSkKUg27ClYNTwtXDeML Vg11DFINBQ1LDZMNQw0fDjcNpw4qDS0PGg2wDwgNLxD0DKwQ3gwkEcYMmRGs DAoSkAx3EnIM3xJSDEMTMQyjEw4M/RPqC1MUxAujFJ0L7hR1CzMVSwtzFSAL rRXzCuAVxgoNFpgKNBZoClQWOApuFgcKgBbVCYsWogmPFm8JjxZrCY8WXwmP FkwJjxYzCY8WEwmPFu4IjxbDCI8WlAiPFmEIjxYqCI8W8AePFrQHjxZ2B48W NgePFvUGjxazBo8WcQaPFjAGjxbwBY8WsgWPFnYFjxY8BY8WBQWPFtIEjxaj BI8WeASPFlMEjxYzBI8WGgSPFgcEjxb7A48W9wMEAAAABAENAAkAAAD6AgYA BQAAAAAAAAAiAAQAAAAtAQMABAAAAPABAgAEAAAALQEAAAcAAAD8AgAAAAAA AAAABAAAAC0BAgAEAAAABgEBABQAAAAkAwgAmgsHAzkNjgM6Ee8BARN7AgwS GAFhDRgBQQ+BAZoLBwMEAAAALQEAAAQAAAAEAQ0ACQAAAPoCAAAFAAAA//// ACIABAAAAC0BBAAEAAAA8AEDABQAAAAlAwgAmgsHAzkNjgM6Ee8BARN7AgwS GAFhDRgBQQ+BAZoLBwMEAAAABAENAAkAAAD6AgYABQAAAAAAAAAiAAQAAAAt AQMABAAAAPABBAAEAAAALQEAAAQAAAAtAQIABAAAAAYBAQAUAAAAJAMIAJoL BwM5DY4DOhHvAQETewIMEhgBYQ0YAUEPgQGaCwcDBAAAAC0BAAAEAAAABAEN AAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQQABAAAAPABAwAUAAAAJQMIAJoL BwM5DY4DOhHvAQETewIMEhgBYQ0YAUEPgQGaCwcDBAAAAAQBDQAJAAAA+gIG AAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQQABAAAAC0BAAAEAAAALQECAAQA AAAGAQEAFAAAACQDCADhCoMETAn8A24FmwV/Aw8FeQSLBkwJiwY6B/oF4QqD BAQAAAAtAQAABAAAAAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAALQEEAAQA AADwAQMAFAAAACUDCADhCoMETAn8A24FmwV/Aw8FeQSLBkwJiwY6B/oF4QqD BAQAAAAEAQ0ACQAAAPoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA8AEEAAQA AAAtAQAABAAAAC0BAgAEAAAABgEBABQAAAAkAwgA4QqDBEwJ/ANuBZsFfwMP BXkEiwZMCYsGOgf6BeEKgwQEAAAALQEAAAQAAAAEAQ0ACQAAAPoCAAAFAAAA ////ACIABAAAAC0BBAAEAAAA8AEDABQAAAAlAwgA4QqDBEwJ/ANuBZsFfwMP BXkEiwZMCYsGOgf6BeEKgwQEAAAABAENAAkAAAD6AgYABQAAAAAAAAAiAAQA AAAtAQMABAAAAPABBAAEAAAALQEAAAQAAAAtAQIABAAAAAYBAQAUAAAAJAMI AOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EBBAAAAC0BAAAEAAAA BAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQQABAAAAPABAwAUAAAAJQMI AOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EBBAAAAAQBDQAJAAAA +gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQQABAAAAC0BAAAEAAAALQEC AAQAAAAGAQEAFAAAACQDCADjA4EBjAX6AI0JewJZCw0CXwprA68FawOjBwcD 4wOBAQQAAAAtAQAABAAAAAQBDQAJAAAA+gIAAAUAAAD///8AIgAEAAAALQEE AAQAAADwAQMAFAAAACUDCADjA4EBjAX6AI0JewJZCw0CXwprA68FawOjBwcD 4wOBAQQAAAAEAQ0ACQAAAPoCBgAFAAAAAAAAACIABAAAAC0BAwAEAAAA8AEE AAQAAAAtAQAABAAAAC0BAgAEAAAABgEBABQAAAAkAwgAuxInBhwRrgY5DQ8F WQubBT8MOAQcETgEAA+hBLsSJwYEAAAALQEAAAQAAAAEAQ0ACQAAAPoCAAAF AAAA////ACIABAAAAC0BBAAEAAAA8AEDABQAAAAlAwgAuxInBhwRrgY5DQ8F WQubBT8MOAQcETgEAA+hBLsSJwYEAAAABAENAAkAAAD6AgYABQAAAAAAAAAi AAQAAAAtAQMABAAAAPABBAAEAAAALQEAAAQAAAAtAQIABAAAAAYBAQAUAAAA JAMIALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGBAAAAC0BAAAE AAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQQABAAAAPABAwAUAAAA JQMIALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGBAAAAAQBDQAJ AAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQQABAAAAC0BAAAHAAAA /AIAAP///wAAAAQAAAAtAQQABAAAAAYBAQAUAAAAJAMIAL0LJQNhDawDXREN Ah8TmQIvEkABfw1AAX0PnwG9CyUDBAAAAC0BAAAEAAAABAENAAkAAAD6AgAA BQAAAP///wAiAAQAAAAtAQUABAAAAPABAwAUAAAAJQMIAL0LJQNhDawDXREN Ah8TmQIvEkABfw1AAX0PnwG9CyUDBAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAA IgAEAAAALQEDAAQAAADwAQUABAAAAC0BAAAEAAAALQEEAAQAAAAGAQEAFAAA ACQDCAC9CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9D58BvQslAwQAAAAtAQAA BAAAAAQBDQAUAAAAJQMIAL0LJQNhDawDXRENAh8TmQIvEkABfw1AAX0PnwG9 CyUDBAAAAAQBDQAEAAAALQEAAAQAAAAtAQQABAAAAAYBAQAUAAAAJAMIAA4L oQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEBAAAAC0BAAAEAAAABAEN AAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQUABAAAAPABAwAUAAAAJQMIAA4L oQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEBAAAAAQBDQAJAAAA+gIG AAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQUABAAAAC0BAAAEAAAALQEEAAQA AAAGAQEAFAAAACQDCAAOC6EEbwkaBIwFuQWiAy0FnASuBm8JrgZdBycGDguh BAQAAAAtAQAABAAAAAQBDQAUAAAAJQMIAA4LoQRvCRoEjAW5BaIDLQWcBK4G bwmuBl0HJwYOC6EEBAAAAAQBDQAEAAAALQEAAAQAAAAtAQQABAAAAAYBAQAU AAAAJAMIABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EHJQMaBJ8BBAAAAC0B AAAEAAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQUABAAAAPABAwAU AAAAJQMIABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EHJQMaBJ8BBAAAAAQB DQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQUABAAAAC0BAAAE AAAALQEEAAQAAAAGAQEAFAAAACQDCAAaBJ8BrwUYAbAJmQJ8CysCggqOA9wF jgPBByUDGgSfAQQAAAAtAQAABAAAAAQBDQAUAAAAJQMIABoEnwGvBRgBsAmZ AnwLKwKCCo4D3AWOA8EHJQMaBJ8BBAAAAAQBDQAEAAAALQEAAAQAAAAtAQQA BAAAAAYBAQAUAAAAJAMIAOMSRQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMPvwTj EkUGBAAAAC0BAAAEAAAABAENAAkAAAD6AgAABQAAAP///wAiAAQAAAAtAQUA BAAAAPABAwAUAAAAJQMIAOMSRQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMPvwTj EkUGBAAAAAQBDQAJAAAA+gIGAAUAAAAAAAAAIgAEAAAALQEDAAQAAADwAQUA BAAAAC0BAAAEAAAALQEEAAQAAAAGAQEAFAAAACQDCADjEkUGOhHMBmENLQV8 C7kFbAxgBDoRYAQjD78E4xJFBgQAAAAtAQAABAAAAAQBDQAUAAAAJQMIAOMS RQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMPvwTjEkUGBAAAAAQBDQAEAAAAJwH/ /x4AAAAmBg8AMQBDb3B5cmlnaHQgqSAxOTk2IE1pY3Jvc3lzdGVtcyBFbmdp bmVlcmluZyBDb21wYW55AA8AAAAmBg8AEwBBbGwgUmlnaHRzIFJlc2VydmVk AAQAAADwAQEAAwAAAAAAAAARAAAADAAAAAgAAAALAAAAEAAAANsAAACCAAAA CQAAABAAAADbAAAAggAAAAkAAAAQAAAAoBYAAGgNAAAJAAAAEAAAAKAWAABo DQAAIAAAABgAAAABAAAAAQAAAAEAAAABAAAAIQAAAAgAAAAUAAAADAAAAA0A AAAlAAAADAAAAAUAAIAmAAAAHAAAAAEAAAAAAAAAAAAAAAAAAAAAAAAAJQAA AAwAAAABAAAAJgAAABwAAAACAAAABgAAAB4AAAAAAAAAAAAAACUAAAAMAAAA AgAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAC/v78AAAAAACUAAAAM AAAAAwAAABMAAAAMAAAAAQAAAFYAAAAgAgAAAQAAAAEAAADaAAAATAAAAIEA AACPFvcDixbEA4AWkQNuFl8DVRYuAzQW/QIOFs4C4BWfAq0VcgJzFUYCMxUa Au4U8QGjFMgBUxShAf0TewGjE1cBRBM0AeASEwF3EvMAChLWAJkRugAlEaAA rBCIADAQcgCwD14ALQ9MAKgOPAAfDi8AlA0jAAYNGwB2DBQA4wsQAE8LDwC7 ChAAKAoUAJgJGwAKCSMAfwgvAPYHPABxB0wA7gZeAG4GcgDyBYgAeQWgAAUF ugCUBNYAJwTzAL4DEwFaAzQB+wJXAaECewFLAqEB+wHIAbAB8QFrARoCKwFG AvEAcgK+AJ8CkADOAmoA/QJJAC4DMABfAx4AkQMTAMQDDwD3AxMAKgQeAF0E MACPBEkAwARqAPEEkAAgBb4ATwXxAHwFKwGoBWsB1AWwAf0F+wEmBksCTQah AnMG+wKXBloDuga+A9sGJwT7BpQEGAcFBTQHeQVOB/IFZgduBnwH7gaQB3EH ogf2B7IHfwi/BwoJyweYCdMHKAraB7sK3gdPC98H4wveB3YM2gcGDdMHlA3L Bx8OvweoDrIHLQ+iB7APkAcwEHwHrBBmByURTgeZETQHChIYB3cS+wbgEtsG RBO6BqMTlwb9E3MGUxRNBqMUJgbuFP0FMxXUBXMVqAWtFXwF4BVPBQ4WIAU0 FvEEVRbABG4WjwSAFl0EixYqBI8W9wMlAAAADAAAAAUAAIAoAAAADAAAAAMA AAAUAAAADAAAAA0AAABXAAAAIAIAAAEAAAABAAAA2gAAAEwAAACBAAAAjxb3 A4sWxAOAFpEDbhZfA1UWLgM0Fv0CDhbOAuAVnwKtFXICcxVGAjMVGgLuFPEB oxTIAVMUoQH9E3sBoxNXAUQTNAHgEhMBdxLzAAoS1gCZEboAJRGgAKwQiAAw EHIAsA9eAC0PTACoDjwAHw4vAJQNIwAGDRsAdgwUAOMLEABPCw8AuwoQACgK FACYCRsACgkjAH8ILwD2BzwAcQdMAO4GXgBuBnIA8gWIAHkFoAAFBboAlATW ACcE8wC+AxMBWgM0AfsCVwGhAnsBSwKhAfsByAGwAfEBawEaAisBRgLxAHIC vgCfApAAzgJqAP0CSQAuAzAAXwMeAJEDEwDEAw8A9wMTACoEHgBdBDAAjwRJ AMAEagDxBJAAIAW+AE8F8QB8BSsBqAVrAdQFsAH9BfsBJgZLAk0GoQJzBvsC lwZaA7oGvgPbBicE+waUBBgHBQU0B3kFTgfyBWYHbgZ8B+4GkAdxB6IH9gey B38IvwcKCcsHmAnTBygK2ge7Ct4HTwvfB+ML3gd2DNoHBg3TB5QNywcfDr8H qA6yBy0PogewD5AHMBB8B6wQZgclEU4HmRE0BwoSGAd3EvsG4BLbBkQTugaj E5cG/RNzBlMUTQajFCYG7hT9BTMV1AVzFagFrRV8BeAVTwUOFiAFNBbxBFUW wARuFo8EgBZdBIsWKgSPFvcDFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAA ABgAAAADAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAADAAAAEwAAAAwAAAABAAAA VgAAACADAAABAAAAJgAAANoAAACBAAAAwQAAAI8W9wOLFioEgBZdBG4WjwRU FsAENBbwBA0WIAXgFU4FrRV7BXMVqAUzFdMF7hT9BaMUJQZTFEwG/RNyBqMT lgZDE7kG3xLaBncS+gYKEhgHmRE0ByQRTgesEGYHLxB8B7APkActD6IHpw6y Bx8OvweTDcsHBQ3TB3UM2gfjC94HTwvfB7sK3gcpCtoHmQnTBwsJywd/CL8H 9weyB3EHogfuBpAHbwZ8B/IFZgd6BU4HBQU0B5QEGAcnBPoGvwPaBlsDuQb7 ApYGoQJyBksCTAb7ASUGsAH9BWsB0wUrAagF8QB7Bb4ATgWRACAFagDwBEoA wAQwAI8EHgBdBBMAKgQPAPcDDwD7Aw8ABwQPABoEDwAzBA8AUwQPAHgEDwCj BA8A0gQPAAUFDwA8BQ8AdgUPALIFDwDwBQ8AMAYPAHEGDwCzBg8A9QYPADYH DwB2Bw8AtAcPAPAHDwAqCA8AYQgPAJQIDwDDCA8A7ggPABMJDwAzCQ8ATAkP AF8JDwBrCQ8AbwkTAKIJHgDVCTAABwpKADgKagBoCpEAmAq+AMYK8QDzCisB IAtrAUsLsAF1C/sBnQtLAsQLoQLqC/sCDgxbAzEMvwNSDCcEcgyUBJAMBQWs DHoFxgzyBd4Mbwb0DO4GCA1xBxoN9wcqDX8INw0LCUMNmQlLDSkKUg27ClYN TwtXDeMLVg11DFINBQ1LDZMNQw0fDjcNpw4qDS0PGg2wDwgNLxD0DKwQ3gwk EcYMmRGsDAoSkAx3EnIM3xJSDEMTMQyjEw4M/RPqC1MUxAujFJ0L7hR1CzMV SwtzFSALrRXzCuAVxgoNFpgKNBZoClQWOApuFgcKgBbVCYsWogmPFm8JjxZr CY8WXwmPFkwJjxYzCY8WEwmPFu4IjxbDCI8WlAiPFmEIjxYqCI8W8AePFrQH jxZ2B48WNgePFvUGjxazBo8WcQaPFjAGjxbwBY8WsgWPFnYFjxY8BY8WBQWP FtIEjxajBI8WeASPFlMEjxYzBI8WGgSPFgcEjxb7A48W9wMlAAAADAAAAAUA AIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAIAMAAAEAAAAmAAAA2gAA AIEAAADBAAAAjxb3A4sWKgSAFl0EbhaPBFQWwAQ0FvAEDRYgBeAVTgWtFXsF cxWoBTMV0wXuFP0FoxQlBlMUTAb9E3IGoxOWBkMTuQbfEtoGdxL6BgoSGAeZ ETQHJBFOB6wQZgcvEHwHsA+QBy0PogenDrIHHw6/B5MNywcFDdMHdQzaB+ML 3gdPC98HuwreBykK2geZCdMHCwnLB38Ivwf3B7IHcQeiB+4GkAdvBnwH8gVm B3oFTgcFBTQHlAQYBycE+ga/A9oGWwO5BvsClgahAnIGSwJMBvsBJQawAf0F awHTBSsBqAXxAHsFvgBOBZEAIAVqAPAESgDABDAAjwQeAF0EEwAqBA8A9wMP APsDDwAHBA8AGgQPADMEDwBTBA8AeAQPAKMEDwDSBA8ABQUPADwFDwB2BQ8A sgUPAPAFDwAwBg8AcQYPALMGDwD1Bg8ANgcPAHYHDwC0Bw8A8AcPACoIDwBh CA8AlAgPAMMIDwDuCA8AEwkPADMJDwBMCQ8AXwkPAGsJDwBvCRMAogkeANUJ MAAHCkoAOApqAGgKkQCYCr4AxgrxAPMKKwEgC2sBSwuwAXUL+wGdC0sCxAuh AuoL+wIODFsDMQy/A1IMJwRyDJQEkAwFBawMegXGDPIF3gxvBvQM7gYIDXEH Gg33ByoNfwg3DQsJQw2ZCUsNKQpSDbsKVg1PC1cN4wtWDXUMUg0FDUsNkw1D DR8ONw2nDioNLQ8aDbAPCA0vEPQMrBDeDCQRxgyZEawMChKQDHcScgzfElIM QxMxDKMTDgz9E+oLUxTEC6MUnQvuFHULMxVLC3MVIAutFfMK4BXGCg0WmAo0 FmgKVBY4Cm4WBwqAFtUJixaiCY8WbwmPFmsJjxZfCY8WTAmPFjMJjxYTCY8W 7giPFsMIjxaUCI8WYQiPFioIjxbwB48WtAePFnYHjxY2B48W9QaPFrMGjxZx Bo8WMAaPFvAFjxayBY8WdgWPFjwFjxYFBY8W0gSPFqMEjxZ4BI8WUwSPFjME jxYaBI8WBwSPFvsDjxb3AxQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYAAAAF AAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAIAAAAlAAAADAAAAAUA AIAnAAAAGAAAAAIAAAAAAAAAAAAAAAAAAAAlAAAADAAAAAIAAAATAAAADAAA AAEAAABWAAAAPAAAAHAAAAALAAAAuAAAACMAAAAIAAAAmgsHAzkNjgM6Ee8B ARN7AgwSGAFhDRgBQQ+BAZoLBwMlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAm AAAAHAAAAAQAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAEAAAAKAAAAAwA AAADAAAAVwAAADwAAABwAAAACwAAALgAAAAiAAAACAAAAJoLBwM5DY4DOhHv AQETewIMEhgBYQ0YAUEPgQGaCwcDFAAAAAwAAAANAAAAJgAAABwAAAADAAAA BgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABAAAACUAAAAM AAAABQAAgCUAAAAMAAAAAgAAABMAAAAMAAAAAQAAAFYAAAA8AAAAcAAAAAsA AAC4AAAAIwAAAAgAAACaCwcDOQ2OAzoR7wEBE3sCDBIYAWENGAFBD4EBmgsH AyUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAABAAAAAAAAAAFAAAA AAAAAP///wAlAAAADAAAAAQAAAAoAAAADAAAAAMAAABXAAAAPAAAAHAAAAAL AAAAuAAAACIAAAAIAAAAmgsHAzkNjgM6Ee8BARN7AgwSGAFhDRgBQQ+BAZoL BwMUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAABQAAAAAAAAAAAAAAJQAA AAwAAAADAAAAKAAAAAwAAAAEAAAAJQAAAAwAAAAFAACAJQAAAAwAAAACAAAA EwAAAAwAAAABAAAAVgAAADwAAAAiAAAAJwAAAGkAAAA/AAAACAAAAOEKgwRM CfwDbgWbBX8DDwV5BIsGTAmLBjoH+gXhCoMEJQAAAAwAAAAFAACAFAAAAAwA AAANAAAAJgAAABwAAAAEAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABAAA ACgAAAAMAAAAAwAAAFcAAAA8AAAAIgAAACcAAABpAAAAPwAAAAgAAADhCoME TAn8A24FmwV/Aw8FeQSLBkwJiwY6B/oF4QqDBBQAAAAMAAAADQAAACYAAAAc AAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAQA AAAlAAAADAAAAAUAAIAlAAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAPAAA ACIAAAAnAAAAaQAAAD8AAAAIAAAA4QqDBEwJ/ANuBZsFfwMPBXkEiwZMCYsG Ogf6BeEKgwQlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAA AAAABQAAAAAAAAD///8AJQAAAAwAAAAEAAAAKAAAAAwAAAADAAAAVwAAADwA AAAiAAAAJwAAAGkAAAA/AAAACAAAAOEKgwRMCfwDbgWbBX8DDwV5BIsGTAmL BjoH+gXhCoMEFAAAAAwAAAANAAAAJgAAABwAAAADAAAABgAAAAUAAAAAAAAA AAAAACUAAAAMAAAAAwAAACgAAAAMAAAABAAAACUAAAAMAAAABQAAgCUAAAAM AAAAAgAAABMAAAAMAAAAAQAAAFYAAAA8AAAAJgAAAAoAAABuAAAAIQAAAAgA AADjA4EBjAX6AI0JewJZCw0CXwprA68FawOjBwcD4wOBASUAAAAMAAAABQAA gBQAAAAMAAAADQAAACYAAAAcAAAABAAAAAAAAAAFAAAAAAAAAP///wAlAAAA DAAAAAQAAAAoAAAADAAAAAMAAABXAAAAPAAAACYAAAAJAAAAbgAAACEAAAAI AAAA4wOBAYwF+gCNCXsCWQsNAl8KawOvBWsDowcHA+MDgQEUAAAADAAAAA0A AAAmAAAAHAAAAAMAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAA AAwAAAAEAAAAJQAAAAwAAAAFAACAJQAAAAwAAAACAAAAEwAAAAwAAAABAAAA VgAAADwAAAAmAAAACgAAAG4AAAAhAAAACAAAAOMDgQGMBfoAjQl7AlkLDQJf CmsDrwVrA6MHBwPjA4EBJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwA AAAEAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABAAAACgAAAAMAAAAAwAA AFcAAAA8AAAAJgAAAAkAAABuAAAAIQAAAAgAAADjA4EBjAX6AI0JewJZCw0C XwprA68FawOjBwcD4wOBARQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYAAAAF AAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAQAAAAlAAAADAAAAAUA AIAlAAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAPAAAAG4AAAApAAAAtQAA AEEAAAAIAAAAuxInBhwRrgY5DQ8FWQubBT8MOAQcETgEAA+hBLsSJwYlAAAA DAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAAAAAABQAAAAAAAAD/ //8AJQAAAAwAAAAEAAAAKAAAAAwAAAADAAAAVwAAADwAAABuAAAAKQAAALUA AABBAAAACAAAALsSJwYcEa4GOQ0PBVkLmwU/DDgEHBE4BAAPoQS7EicGFAAA AAwAAAANAAAAJgAAABwAAAADAAAABgAAAAUAAAAAAAAAAAAAACUAAAAMAAAA AwAAACgAAAAMAAAABAAAACUAAAAMAAAABQAAgCUAAAAMAAAAAgAAABMAAAAM AAAAAQAAAFYAAAA8AAAAbgAAACkAAAC1AAAAQQAAAAgAAAC7EicGHBGuBjkN DwVZC5sFPww4BBwROAQAD6EEuxInBiUAAAAMAAAABQAAgBQAAAAMAAAADQAA ACYAAAAcAAAABAAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAQAAAAoAAAA DAAAAAMAAABXAAAAPAAAAG4AAAApAAAAtQAAAEEAAAAIAAAAuxInBhwRrgY5 DQ8FWQubBT8MOAQcETgEAA+hBLsSJwYUAAAADAAAAA0AAAAmAAAAHAAAAAMA AAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAAEAAAAJQAA AAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAP///wAAAAAAJQAAAAwAAAAEAAAA EwAAAAwAAAABAAAAVgAAADwAAAByAAAADAAAALkAAAAkAAAACAAAAL0LJQNh DawDXRENAh8TmQIvEkABfw1AAX0PnwG9CyUDJQAAAAwAAAAFAACAFAAAAAwA AAANAAAAJgAAABwAAAAFAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABQAA ACgAAAAMAAAAAwAAAFcAAAA8AAAAcgAAAAwAAAC5AAAAJAAAAAgAAAC9CyUD YQ2sA10RDQIfE5kCLxJAAX8NQAF9D58BvQslAxQAAAAMAAAADQAAACYAAAAc AAAAAwAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAUA AAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAAAAEAAABWAAAAPAAA AHIAAAAMAAAAuQAAACQAAAAIAAAAvQslA2ENrANdEQ0CHxOZAi8SQAF/DUAB fQ+fAb0LJQMlAAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAPAAAAHIAAAAM AAAAuQAAACQAAAAIAAAAvQslA2ENrANdEQ0CHxOZAi8SQAF/DUABfQ+fAb0L JQMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAA AAEAAABWAAAAPAAAACMAAAAoAAAAawAAAEEAAAAIAAAADguhBG8JGgSMBbkF ogMtBZwErgZvCa4GXQcnBg4LoQQlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAm AAAAHAAAAAUAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAFAAAAKAAAAAwA AAADAAAAVwAAADwAAAAjAAAAKAAAAGsAAABBAAAACAAAAA4LoQRvCRoEjAW5 BaIDLQWcBK4GbwmuBl0HJwYOC6EEFAAAAAwAAAANAAAAJgAAABwAAAADAAAA BgAAAAUAAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABQAAACUAAAAM AAAABQAAgCUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAA8AAAAIwAAACgA AABrAAAAQQAAAAgAAAAOC6EEbwkaBIwFuQWiAy0FnASuBm8JrgZdBycGDguh BCUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAA8AAAAIwAAACgAAABrAAAA QQAAAAgAAAAOC6EEbwkaBIwFuQWiAy0FnASuBm8JrgZdBycGDguhBBQAAAAM AAAADQAAACUAAAAMAAAABQAAgCUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYA AAA8AAAAKAAAAAsAAABvAAAAIwAAAAgAAAAaBJ8BrwUYAbAJmQJ8CysCggqO A9wFjgPBByUDGgSfASUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAA BQAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAUAAAAoAAAADAAAAAMAAABX AAAAPAAAACgAAAALAAAAbwAAACIAAAAIAAAAGgSfAa8FGAGwCZkCfAsrAoIK jgPcBY4DwQclAxoEnwEUAAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAABQAA AAAAAAAAAAAAJQAAAAwAAAADAAAAKAAAAAwAAAAFAAAAJQAAAAwAAAAFAACA JQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADwAAAAoAAAACwAAAG8AAAAj AAAACAAAABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EHJQMaBJ8BJQAAAAwA AAAFAACAFAAAAAwAAAANAAAAVwAAADwAAAAoAAAACwAAAG8AAAAiAAAACAAA ABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EHJQMaBJ8BFAAAAAwAAAANAAAA JQAAAAwAAAAFAACAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADwAAABv AAAAKgAAALcAAABCAAAACAAAAOMSRQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMP vwTjEkUGJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAFAAAAAAAA AAUAAAAAAAAA////ACUAAAAMAAAABQAAACgAAAAMAAAAAwAAAFcAAAA8AAAA bwAAACoAAAC3AAAAQgAAAAgAAADjEkUGOhHMBmENLQV8C7kFbAxgBDoRYAQj D78E4xJFBhQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYAAAAFAAAAAAAAAAAA AAAlAAAADAAAAAMAAAAoAAAADAAAAAUAAAAlAAAADAAAAAUAAIAlAAAADAAA AAQAAAATAAAADAAAAAEAAABWAAAAPAAAAG8AAAAqAAAAtwAAAEIAAAAIAAAA 4xJFBjoRzAZhDS0FfAu5BWwMYAQ6EWAEIw+/BOMSRQYlAAAADAAAAAUAAIAU AAAADAAAAA0AAABXAAAAPAAAAG8AAAAqAAAAtwAAAEIAAAAIAAAA4xJFBjoR zAZhDS0FfAu5BWwMYAQ6EWAEIw+/BOMSRQYUAAAADAAAAA0AAAAiAAAADAAA AP////8oAAAADAAAAAEAAAAlAAAADAAAAAcAAIAlAAAADAAAAAAAAIAwAAAA DAAAAA8AAIBLAAAAEAAAAAAAAAAFAAAAKAAAAAwAAAACAAAAKAAAAAwAAAAD AAAAKAAAAAwAAAAEAAAADgAAABQAAAAAAAAAEAAAABQAAAAaQAJALAAAACAA AAD/////AgAAAAAAAAAAAAAAP2FcQx8zA0MAAAAA6APoAyEAAAAIAAAAYgAA AAwAAAABAAAAIQAAAAgAAAAhAAAACAAAAHMAAAAMAAAAAAAAABwAAAAIAAAA JQAAAAwAAAAAAACAJQAAAAwAAAAHAACAJQAAAAwAAAAOAACAMAAAAAwAAAAP AACAGQAAAAwAAAD///8AGAAAAAwAAAAAAAAAEgAAAAwAAAACAAAAEwAAAAwA AAABAAAAFAAAAAwAAAANAAAAFQAAAAwAAAABAAAAFgAAAAwAAAAAAAAADQAA ABAAAAAAAAAAAAAAADoAAAAMAAAACgAAABsAAAAQAAAAAAAAAAAAAAAjAAAA IAAAAIUN7j4AAAAAAAAAAPvx6z4AQAdEAAC7QyQAAAAkAAAAhQ3uPgAAAAAA AAAA+/HrPgBAB0QAALtDBAAAAHMAAAAMAAAAAAAAAA0AAAAQAAAAHQIAAHYB AABSAAAAcAEAAAEAAAAUAAAACgAAAAAAAAAAAAAAkAEAAAAAAIEHAgIiUwB5 AHMAdABlAG0AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACAAAAAAAAAAAAAAAEhl8Xfc ZPF3CQAAACEAigENAAAAlN8SAIU08nfQJGQDBBoh3tAkZAMmAIoBBBoh3gsA iAH//wEABAAAANzcEgCk3RIA0N0SABjukHxwBZF8/////20FkXxamYB8AAAU AAAAAABtmYB8AAAQxlBMWQMgMPV3JQAAALBGFQAAABAAsEYVAOTdEgCbKPJ3 KCAiALBGFQDQJGQDR0RJQwQaId4hAIoBCAAAAAAAEgCAc1kD2PprAxAvHAAT AAAAEC8cABAvHADY+msD2PprAyjeEgDBEvJ3EC8cABMAAACAc1kDEC8cADje EgBkdgAIAAAAACUAAAAMAAAAAQAAAEYAAAAoAAAAHAAAAEdESUMCAAAAAAAA AAAAAADbAAAAggAAAAAAAAAjAAAAIAAAAIUN7j4AAAAAAAAAAPvx6z4AQAdE AAC7QyQAAAAkAAAAhQ3uPgAAAAAAAAAA+/HrPgBAB0QAALtDBAAAACMAAAAg AAAAkaXLQgAAAAAAAAAAw6FvQgBAB0QAALtDJAAAACQAAACRpctCAAAAAAAA AADDoW9CAEAHRAAAu0MEAAAAIwAAACAAAACFDe4+AAAAAAAAAAD78es+AEAH RAAAu0MkAAAAJAAAAIUN7j4AAAAAAAAAAPvx6z4AQAdEAAC7QwQAAAAjAAAA IAAAAO8DkDwAAAAAAAAAAEP/jjwAQAdEAAC7QyQAAAAkAAAA7wOQPAAAAAAA AAAAQ/+OPABAB0QAALtDBAAAACMAAAAgAAAA7wOQPAAAAAAAAAAAQ/+OPABA B0QAALtDJAAAACQAAADvA5A8AAAAAAAAAABD/448AEAHRAAAu0MEAAAAIwAA ACAAAADvA5A8AAAAAAAAAABD/448AEAHRAAAu0MkAAAAJAAAAO8DkDwAAAAA AAAAAEP/jjwAQAdEAAC7QwQAAAAhAAAACAAAABQAAAAMAAAADQAAACUAAAAM AAAABQAAgCYAAAAcAAAAAgAAAAAAAAAAAAAAAAAAAAAAAAAlAAAADAAAAAIA AAAmAAAAHAAAAAMAAAAGAAAAHgAAAAAAAAAAAAAAJQAAAAwAAAADAAAAJQAA AAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAEAAAA EwAAAAwAAAABAAAAVgAAACACAAAdAgAAdgEAAIMCAACaAQAAgQAAAI8W9wOL FsQDgBaRA24WXwNVFi4DNBb9Ag4WzgLgFZ8CrRVyAnMVRgIzFRoC7hTxAaMU yAFTFKEB/RN7AaMTVwFEEzQB4BITAXcS8wAKEtYAmRG6ACURoACsEIgAMBBy ALAPXgAtD0wAqA48AB8OLwCUDSMABg0bAHYMFADjCxAATwsPALsKEAAoChQA mAkbAAoJIwB/CC8A9gc8AHEHTADuBl4AbgZyAPIFiAB5BaAABQW6AJQE1gAn BPMAvgMTAVoDNAH7AlcBoQJ7AUsCoQH7AcgBsAHxAWsBGgIrAUYC8QByAr4A nwKQAM4CagD9AkkALgMwAF8DHgCRAxMAxAMPAPcDEwAqBB4AXQQwAI8ESQDA BGoA8QSQACAFvgBPBfEAfAUrAagFawHUBbAB/QX7ASYGSwJNBqECcwb7ApcG WgO6Br4D2wYnBPsGlAQYBwUFNAd5BU4H8gVmB24GfAfuBpAHcQeiB/YHsgd/ CL8HCgnLB5gJ0wcoCtoHuwreB08L3wfjC94HdgzaBwYN0weUDcsHHw6/B6gO sgctD6IHsA+QBzAQfAesEGYHJRFOB5kRNAcKEhgHdxL7BuAS2wZEE7oGoxOX Bv0TcwZTFE0GoxQmBu4U/QUzFdQFcxWoBa0VfAXgFU8FDhYgBTQW8QRVFsAE bhaPBIAWXQSLFioEjxb3AyUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAM AAAADQAAAFcAAAAgAgAAHQIAAHYBAACDAgAAmgEAAIEAAACPFvcDixbEA4AW kQNuFl8DVRYuAzQW/QIOFs4C4BWfAq0VcgJzFUYCMxUaAu4U8QGjFMgBUxSh Af0TewGjE1cBRBM0AeASEwF3EvMAChLWAJkRugAlEaAArBCIADAQcgCwD14A LQ9MAKgOPAAfDi8AlA0jAAYNGwB2DBQA4wsQAE8LDwC7ChAAKAoUAJgJGwAK CSMAfwgvAPYHPABxB0wA7gZeAG4GcgDyBYgAeQWgAAUFugCUBNYAJwTzAL4D EwFaAzQB+wJXAaECewFLAqEB+wHIAbAB8QFrARoCKwFGAvEAcgK+AJ8CkADO AmoA/QJJAC4DMABfAx4AkQMTAMQDDwD3AxMAKgQeAF0EMACPBEkAwARqAPEE kAAgBb4ATwXxAHwFKwGoBWsB1AWwAf0F+wEmBksCTQahAnMG+wKXBloDuga+ A9sGJwT7BpQEGAcFBTQHeQVOB/IFZgduBnwH7gaQB3EHogf2B7IHfwi/BwoJ yweYCdMHKAraB7sK3gdPC98H4wveB3YM2gcGDdMHlA3LBx8OvweoDrIHLQ+i B7APkAcwEHwHrBBmByURTgeZETQHChIYB3cS+wbgEtsGRBO6BqMTlwb9E3MG UxRNBqMUJgbuFP0FMxXUBXMVqAWtFXwF4BVPBQ4WIAU0FvEEVRbABG4WjwSA Fl0EixYqBI8W9wMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAQA AAAAAAAAv7+/AAAAAAAlAAAADAAAAAQAAAATAAAADAAAAAEAAABWAAAAIAMA AB0CAACHAQAAgwIAALIBAADBAAAAjxb3A4sWKgSAFl0EbhaPBFQWwAQ0FvAE DRYgBeAVTgWtFXsFcxWoBTMV0wXuFP0FoxQlBlMUTAb9E3IGoxOWBkMTuQbf EtoGdxL6BgoSGAeZETQHJBFOB6wQZgcvEHwHsA+QBy0PogenDrIHHw6/B5MN ywcFDdMHdQzaB+ML3gdPC98HuwreBykK2geZCdMHCwnLB38Ivwf3B7IHcQei B+4GkAdvBnwH8gVmB3oFTgcFBTQHlAQYBycE+ga/A9oGWwO5BvsClgahAnIG SwJMBvsBJQawAf0FawHTBSsBqAXxAHsFvgBOBZEAIAVqAPAESgDABDAAjwQe AF0EEwAqBA8A9wMPAPsDDwAHBA8AGgQPADMEDwBTBA8AeAQPAKMEDwDSBA8A BQUPADwFDwB2BQ8AsgUPAPAFDwAwBg8AcQYPALMGDwD1Bg8ANgcPAHYHDwC0 Bw8A8AcPACoIDwBhCA8AlAgPAMMIDwDuCA8AEwkPADMJDwBMCQ8AXwkPAGsJ DwBvCRMAogkeANUJMAAHCkoAOApqAGgKkQCYCr4AxgrxAPMKKwEgC2sBSwuw AXUL+wGdC0sCxAuhAuoL+wIODFsDMQy/A1IMJwRyDJQEkAwFBawMegXGDPIF 3gxvBvQM7gYIDXEHGg33ByoNfwg3DQsJQw2ZCUsNKQpSDbsKVg1PC1cN4wtW DXUMUg0FDUsNkw1DDR8ONw2nDioNLQ8aDbAPCA0vEPQMrBDeDCQRxgyZEawM ChKQDHcScgzfElIMQxMxDKMTDgz9E+oLUxTEC6MUnQvuFHULMxVLC3MVIAut FfMK4BXGCg0WmAo0FmgKVBY4Cm4WBwqAFtUJixaiCY8WbwmPFmsJjxZfCY8W TAmPFjMJjxYTCY8W7giPFsMIjxaUCI8WYQiPFioIjxbwB48WtAePFnYHjxY2 B48W9QaPFrMGjxZxBo8WMAaPFvAFjxayBY8WdgWPFjwFjxYFBY8W0gSPFqME jxZ4BI8WUwSPFjMEjxYaBI8WBwSPFvsDjxb3AyUAAAAMAAAABQAAgCgAAAAM AAAABAAAABQAAAAMAAAADQAAAFcAAAAgAwAAHQIAAIcBAACDAgAAsgEAAMEA AACPFvcDixYqBIAWXQRuFo8EVBbABDQW8AQNFiAF4BVOBa0VewVzFagFMxXT Be4U/QWjFCUGUxRMBv0TcgajE5YGQxO5Bt8S2gZ3EvoGChIYB5kRNAckEU4H rBBmBy8QfAewD5AHLQ+iB6cOsgcfDr8Hkw3LBwUN0wd1DNoH4wveB08L3we7 Ct4HKQraB5kJ0wcLCcsHfwi/B/cHsgdxB6IH7gaQB28GfAfyBWYHegVOBwUF NAeUBBgHJwT6Br8D2gZbA7kG+wKWBqECcgZLAkwG+wElBrAB/QVrAdMFKwGo BfEAewW+AE4FkQAgBWoA8ARKAMAEMACPBB4AXQQTACoEDwD3Aw8A+wMPAAcE DwAaBA8AMwQPAFMEDwB4BA8AowQPANIEDwAFBQ8APAUPAHYFDwCyBQ8A8AUP ADAGDwBxBg8AswYPAPUGDwA2Bw8AdgcPALQHDwDwBw8AKggPAGEIDwCUCA8A wwgPAO4IDwATCQ8AMwkPAEwJDwBfCQ8AawkPAG8JEwCiCR4A1QkwAAcKSgA4 CmoAaAqRAJgKvgDGCvEA8worASALawFLC7ABdQv7AZ0LSwLEC6EC6gv7Ag4M WwMxDL8DUgwnBHIMlASQDAUFrAx6BcYM8gXeDG8G9AzuBggNcQcaDfcHKg1/ CDcNCwlDDZkJSw0pClINuwpWDU8LVw3jC1YNdQxSDQUNSw2TDUMNHw43DacO Kg0tDxoNsA8IDS8Q9AysEN4MJBHGDJkRrAwKEpAMdxJyDN8SUgxDEzEMoxMO DP0T6gtTFMQLoxSdC+4UdQszFUsLcxUgC60V8wrgFcYKDRaYCjQWaApUFjgK bhYHCoAW1QmLFqIJjxZvCY8WawmPFl8JjxZMCY8WMwmPFhMJjxbuCI8WwwiP FpQIjxZhCI8WKgiPFvAHjxa0B48WdgePFjYHjxb1Bo8WswaPFnEGjxYwBo8W 8AWPFrIFjxZ2BY8WPAWPFgUFjxbSBI8WowSPFngEjxZTBI8WMwSPFhoEjxYH BI8W+wOPFvcDFAAAAAwAAAANAAAAJgAAABwAAAAEAAAABgAAAAUAAAAAAAAA AAAAACUAAAAMAAAABAAAACgAAAAMAAAAAwAAACUAAAAMAAAABQAAgCcAAAAY AAAAAwAAAAAAAAAAAAAAAAAAACUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYA AAA8AAAAUQIAAHoBAABzAgAAhgEAAAgAAACaCwcDOQ2OAzoR7wEBE3sCDBIY AWENGAFBD4EBmgsHAyUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAA BQAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAUAAAAoAAAADAAAAAQAAABX AAAAPAAAAFECAAB6AQAAcwIAAIYBAAAIAAAAmgsHAzkNjgM6Ee8BARN7AgwS GAFhDRgBQQ+BAZoLBwMUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAGAAAABQAA AAAAAAAAAAAAJQAAAAwAAAAEAAAAKAAAAAwAAAAFAAAAJQAAAAwAAAAFAACA JQAAAAwAAAADAAAAEwAAAAwAAAABAAAAVgAAADwAAABRAgAAegEAAHMCAACG AQAACAAAAJoLBwM5DY4DOhHvAQETewIMEhgBYQ0YAUEPgQGaCwcDJQAAAAwA AAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAFAAAAAAAAAAUAAAAAAAAA//// ACUAAAAMAAAABQAAACgAAAAMAAAABAAAAFcAAAA8AAAAUQIAAHoBAABzAgAA hgEAAAgAAACaCwcDOQ2OAzoR7wEBE3sCDBIYAWENGAFBD4EBmgsHAxQAAAAM AAAADQAAACYAAAAcAAAABAAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAQA AAAoAAAADAAAAAUAAAAlAAAADAAAAAUAAIAlAAAADAAAAAMAAAATAAAADAAA AAEAAABWAAAAPAAAACwCAACHAQAATgIAAJQBAAAIAAAA4QqDBEwJ/ANuBZsF fwMPBXkEiwZMCYsGOgf6BeEKgwQlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAm AAAAHAAAAAUAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAFAAAAKAAAAAwA AAAEAAAAVwAAADwAAAAsAgAAhwEAAE4CAACUAQAACAAAAOEKgwRMCfwDbgWb BX8DDwV5BIsGTAmLBjoH+gXhCoMEFAAAAAwAAAANAAAAJgAAABwAAAAEAAAA BgAAAAUAAAAAAAAAAAAAACUAAAAMAAAABAAAACgAAAAMAAAABQAAACUAAAAM AAAABQAAgCUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAA8AAAALAIAAIcB AABOAgAAlAEAAAgAAADhCoMETAn8A24FmwV/Aw8FeQSLBkwJiwY6B/oF4QqD BCUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAABQAAAAAAAAAFAAAA AAAAAP///wAlAAAADAAAAAUAAAAoAAAADAAAAAQAAABXAAAAPAAAACwCAACH AQAATgIAAJQBAAAIAAAA4QqDBEwJ/ANuBZsFfwMPBXkEiwZMCYsGOgf6BeEK gwQUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAGAAAABQAAAAAAAAAAAAAAJQAA AAwAAAAEAAAAKAAAAAwAAAAFAAAAJQAAAAwAAAAFAACAJQAAAAwAAAADAAAA EwAAAAwAAAABAAAAVgAAADwAAAAuAgAAegEAAFECAACGAQAACAAAAOMDgQGM BfoAjQl7AlkLDQJfCmsDrwVrA6MHBwPjA4EBJQAAAAwAAAAFAACAFAAAAAwA AAANAAAAJgAAABwAAAAFAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABQAA ACgAAAAMAAAABAAAAFcAAAA8AAAALgIAAHoBAABRAgAAhgEAAAgAAADjA4EB jAX6AI0JewJZCw0CXwprA68FawOjBwcD4wOBARQAAAAMAAAADQAAACYAAAAc AAAABAAAAAYAAAAFAAAAAAAAAAAAAAAlAAAADAAAAAQAAAAoAAAADAAAAAUA AAAlAAAADAAAAAUAAIAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAPAAA AC4CAAB6AQAAUQIAAIYBAAAIAAAA4wOBAYwF+gCNCXsCWQsNAl8KawOvBWsD owcHA+MDgQElAAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAUAAAAA AAAABQAAAAAAAAD///8AJQAAAAwAAAAFAAAAKAAAAAwAAAAEAAAAVwAAADwA AAAuAgAAegEAAFECAACGAQAACAAAAOMDgQGMBfoAjQl7AlkLDQJfCmsDrwVr A6MHBwPjA4EBFAAAAAwAAAANAAAAJgAAABwAAAAEAAAABgAAAAUAAAAAAAAA AAAAACUAAAAMAAAABAAAACgAAAAMAAAABQAAACUAAAAMAAAABQAAgCUAAAAM AAAAAwAAABMAAAAMAAAAAQAAAFYAAAA8AAAAUAIAAIgBAAByAgAAlAEAAAgA AAC7EicGHBGuBjkNDwVZC5sFPww4BBwROAQAD6EEuxInBiUAAAAMAAAABQAA gBQAAAAMAAAADQAAACYAAAAcAAAABQAAAAAAAAAFAAAAAAAAAP///wAlAAAA DAAAAAUAAAAoAAAADAAAAAQAAABXAAAAPAAAAFACAACIAQAAcgIAAJQBAAAI AAAAuxInBhwRrgY5DQ8FWQubBT8MOAQcETgEAA+hBLsSJwYUAAAADAAAAA0A AAAmAAAAHAAAAAQAAAAGAAAABQAAAAAAAAAAAAAAJQAAAAwAAAAEAAAAKAAA AAwAAAAFAAAAJQAAAAwAAAAFAACAJQAAAAwAAAADAAAAEwAAAAwAAAABAAAA VgAAADwAAABQAgAAiAEAAHICAACUAQAACAAAALsSJwYcEa4GOQ0PBVkLmwU/ DDgEHBE4BAAPoQS7EicGJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwA AAAFAAAAAAAAAAUAAAAAAAAA////ACUAAAAMAAAABQAAACgAAAAMAAAABAAA AFcAAAA8AAAAUAIAAIgBAAByAgAAlAEAAAgAAAC7EicGHBGuBjkNDwVZC5sF Pww4BBwROAQAD6EEuxInBhQAAAAMAAAADQAAACYAAAAcAAAABAAAAAYAAAAF AAAAAAAAAAAAAAAlAAAADAAAAAQAAAAoAAAADAAAAAUAAAAlAAAADAAAAAUA AIAnAAAAGAAAAAUAAAAAAAAA////AAAAAAAlAAAADAAAAAUAAAATAAAADAAA AAEAAABWAAAAPAAAAFECAAB7AQAAdAIAAIcBAAAIAAAAvQslA2ENrANdEQ0C HxOZAi8SQAF/DUABfQ+fAb0LJQMlAAAADAAAAAUAAIAUAAAADAAAAA0AAAAm AAAAHAAAAAYAAAAAAAAABQAAAAAAAAD///8AJQAAAAwAAAAGAAAAKAAAAAwA AAAEAAAAVwAAADwAAABRAgAAewEAAHQCAACHAQAACAAAAL0LJQNhDawDXREN Ah8TmQIvEkABfw1AAX0PnwG9CyUDFAAAAAwAAAANAAAAJgAAABwAAAAEAAAA BgAAAAUAAAAAAAAAAAAAACUAAAAMAAAABAAAACgAAAAMAAAABgAAACUAAAAM AAAABQAAgCUAAAAMAAAABQAAABMAAAAMAAAAAQAAAFYAAAA8AAAAUQIAAHsB AAB0AgAAhwEAAAgAAAC9CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9D58BvQsl AyUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAA8AAAAUQIAAHsBAAB0AgAA hwEAAAgAAAC9CyUDYQ2sA10RDQIfE5kCLxJAAX8NQAF9D58BvQslAxQAAAAM AAAADQAAACUAAAAMAAAABQAAgCUAAAAMAAAABQAAABMAAAAMAAAAAQAAAFYA AAA8AAAALQIAAIgBAABPAgAAlAEAAAgAAAAOC6EEbwkaBIwFuQWiAy0FnASu Bm8JrgZdBycGDguhBCUAAAAMAAAABQAAgBQAAAAMAAAADQAAACYAAAAcAAAA BgAAAAAAAAAFAAAAAAAAAP///wAlAAAADAAAAAYAAAAoAAAADAAAAAQAAABX AAAAPAAAAC0CAACIAQAATwIAAJQBAAAIAAAADguhBG8JGgSMBbkFogMtBZwE rgZvCa4GXQcnBg4LoQQUAAAADAAAAA0AAAAmAAAAHAAAAAQAAAAGAAAABQAA AAAAAAAAAAAAJQAAAAwAAAAEAAAAKAAAAAwAAAAGAAAAJQAAAAwAAAAFAACA JQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADwAAAAtAgAAiAEAAE8CAACU AQAACAAAAA4LoQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEJQAAAAwA AAAFAACAFAAAAAwAAAANAAAAVwAAADwAAAAtAgAAiAEAAE8CAACUAQAACAAA AA4LoQRvCRoEjAW5BaIDLQWcBK4GbwmuBl0HJwYOC6EEFAAAAAwAAAANAAAA JQAAAAwAAAAFAACAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADwAAAAv AgAAegEAAFECAACGAQAACAAAABoEnwGvBRgBsAmZAnwLKwKCCo4D3AWOA8EH JQMaBJ8BJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAJgAAABwAAAAGAAAAAAAA AAUAAAAAAAAA////ACUAAAAMAAAABgAAACgAAAAMAAAABAAAAFcAAAA8AAAA LwIAAHoBAABRAgAAhgEAAAgAAAAaBJ8BrwUYAbAJmQJ8CysCggqOA9wFjgPB ByUDGgSfARQAAAAMAAAADQAAACYAAAAcAAAABAAAAAYAAAAFAAAAAAAAAAAA AAAlAAAADAAAAAQAAAAoAAAADAAAAAYAAAAlAAAADAAAAAUAAIAlAAAADAAA AAUAAAATAAAADAAAAAEAAABWAAAAPAAAAC8CAAB6AQAAUQIAAIYBAAAIAAAA GgSfAa8FGAGwCZkCfAsrAoIKjgPcBY4DwQclAxoEnwElAAAADAAAAAUAAIAU AAAADAAAAA0AAABXAAAAPAAAAC8CAAB6AQAAUQIAAIYBAAAIAAAAGgSfAa8F GAGwCZkCfAsrAoIKjgPcBY4DwQclAxoEnwEUAAAADAAAAA0AAAAlAAAADAAA AAUAAIAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAPAAAAFACAACJAQAA cgIAAJUBAAAIAAAA4xJFBjoRzAZhDS0FfAu5BWwMYAQ6EWAEIw+/BOMSRQYl AAAADAAAAAUAAIAUAAAADAAAAA0AAAAmAAAAHAAAAAYAAAAAAAAABQAAAAAA AAD///8AJQAAAAwAAAAGAAAAKAAAAAwAAAAEAAAAVwAAADwAAABQAgAAiQEA AHICAACVAQAACAAAAOMSRQY6EcwGYQ0tBXwLuQVsDGAEOhFgBCMPvwTjEkUG FAAAAAwAAAANAAAAJgAAABwAAAAEAAAABgAAAAUAAAAAAAAAAAAAACUAAAAM AAAABAAAACgAAAAMAAAABgAAACUAAAAMAAAABQAAgCUAAAAMAAAABQAAABMA AAAMAAAAAQAAAFYAAAA8AAAAUAIAAIkBAAByAgAAlQEAAAgAAADjEkUGOhHM BmENLQV8C7kFbAxgBDoRYAQjD78E4xJFBiUAAAAMAAAABQAAgBQAAAAMAAAA DQAAAFcAAAA8AAAAUAIAAIkBAAByAgAAlQEAAAgAAADjEkUGOhHMBmENLQV8 C7kFbAxgBDoRYAQjD78E4xJFBhQAAAAMAAAADQAAACIAAAAMAAAA/////ygA AAAMAAAAAgAAACUAAAAMAAAABwAAgCUAAAAMAAAAAAAAgEsAAAAQAAAAAAAA AAUAAAAoAAAADAAAAAMAAAAoAAAADAAAAAQAAAAoAAAADAAAAAUAAABGAAAA FAAAAAgAAABHRElDAwAAACIAAAAMAAAA/////yIAAAAMAAAA/////yUAAAAM AAAADQAAgCgAAAAMAAAAAQAAACIAAAAMAAAA/////0YAAAD8AAAA8AAAAEVN RismQAAAEAAAAAQAAAAAAAAACEADAjwAAAAwAAAAARDA2wAAAABOAAAAAAAA AKc+dj8CAAAAAgAAAAIAAAACAAAAARDA2wAAAAAAAAD/CEAEA5AAAACEAAAA ARDA2w0AAAAAAAAA4IK3Q547xkPggrdDrUquQ8qEm0Mw4ppD9PlxQzDimkNR 6ixDMOKaQ0Tc6UKsSq5DQdzpQp07xkM93OlCjizeQ0rqLEMLlfFD7PlxQwuV 8UPGhJtDDJXxQ96Ct0OQLN5D4IK3Q6A7xkMAAwMDAwMDAwMDAwODAAAAFUAE ABAAAAAEAAAAAwAAACQAAAAkAAAAAACAPQAAAAAAAAAAAACAPQAAAAAAAAAA AgAAAF8AAAA4AAAAAQAAADgAAAAAAAAAOAAAAAAAAAAAAAEADwAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAJQAAAAwAAAABAAAAJQAAAAwAAAAFAACAVQAAAFAA AABzAAAANAEAAHEBAADlAQAADQAAAPEWyBjxFsoVcRNdEyAPXRPPCl0TTwfK FU8HyBhPB8YbzwozHiAPMx5xEzMe8RbGG/EWyBglAAAADAAAAAcAAIAlAAAA DAAAAAAAAIAkAAAAJAAAAAAAgEEAAAAAAAAAAAAAgEEAAAAAAAAAAAIAAAAo AAAADAAAAAEAAABGAAAA7AAAAOAAAABFTUYrCEADAjwAAAAwAAAAARDA2wAA AABOAAAAAAAAAKc+dj8CAAAAAgAAAAIAAAACAAAAARDA2wAAAAAAAAD/CEAE A5AAAACEAAAAARDA2w0AAAAAAAAABLUoRJ47xkMEtShErUquQ1uiGUQw4ppD dAoHRDDimkMb5ehDMOKaQ8e/ykOsSq5Dxr/KQ507xkPFv8pDjizeQxjl6EML lfFDcgoHRAuV8UNZohlEDJXxQwO1KESQLN5DBLUoRKA7xkMAAwMDAwMDAwMD AwODAAAAFUAEABAAAAAEAAAAAwAAACQAAAAkAAAAAACAPQAAAAAAAAAAAACA PQAAAAAAAAAAAgAAAF8AAAA4AAAAAQAAADgAAAAAAAAAOAAAAAAAAAAAAAEA DwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAJQAAAAwAAAABAAAAJQAAAAwAAAAF AACAVQAAAFAAAACUAQAANAEAAKUCAADlAQAADQAAAC4qyBguKsoVaSZdE8Mh XRMdHV0TWBnKFVgZyBhYGcYbHR0zHsMhMx5pJjMeLirGGy4qyBglAAAADAAA AAcAAIAlAAAADAAAAAAAAIAkAAAAJAAAAAAAgEEAAAAAAAAAAAAAgEEAAAAA AAAAAAIAAAAoAAAADAAAAAEAAABGAAAArAAAAKAAAABFTUYrKkAAACQAAAAY AAAA8mDAQgAAAAAAAAAA6KnAQvxFpEMMlfFDCEAFBjAAAAAkAAAAARDA26uq Kj4AAAAAAAAAAAAAAAAFAAAAQQBSAEkAQQBMAAAANkAFgEgAAAA8AAAAAAAA /wEAAAABAAAAAgAAAFIAMQDe7Rc/mpmZPt69Nj+amZk+AACAPwAAAAAAAAAA AACAPwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAY AAAAUgAAAHABAAABAAAA8P///wAAAAAAAAAAAAAAAJABAAAAAAAABwAEAEEA cgBpAGEAbAAAAAAATQBvAG4AbwB0AHkAcABlADoAQQByAGkAYQBsACAAUgBl AGcAdQBsAGEAcgA6AFYAZQAAAP////8/AAAAAAAAAAA/AAAAAAAAAAA/AAAA AAAAAAAhMzMzMzMz878wAAAAAAAAAAAwAAAAAAAAAAABAP3///8AAAAA//// /wAAAAAAAGslAGDsaQ8QOGcPEOz1FQA2HQBgOGcPEOz1FQCU5RIAAABvATIH kXwIAAAACAhvAQAAbwEgb5oCbOUSAMBgTmCw5xIAGO6QfDgHkXz/////MgeR fKsGkXzrBpF8YOsSADgAAACQg5wCJzwAYFTxFQAAABIAAAAAAAEAAAD45RIA wzwAYFTxFQCs5hIAAAAAAMiAyU4EGiHelLNvAQz1FQAAAAAAvhoAYKTmEgCA 5hIAZHYACAAAAAAlAAAADAAAAAEAAABUAAAAWAAAAIIBAADxAQAAkwEAAAIC AAABAAAAAEDTQQDw0kGCAQAAAAIAAAIAAABMAAAAAAAAAAAAAAAAAAAA//// //////9QAAAAUgAxAAwAAAAAAAAAJQAAAAwAAAANAACARgAAALgAAACsAAAA RU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEJ4 yRFDn874QwhABQYwAAAAJAAAAAEQwNurqio+AAAAAAAAAAAAAAAABQAAAEEA UgBJAEEATAAAADZABYBIAAAAPAAAAAAAAP8BAAAAAQAAAAIAAABSADIA3h1G PjMzMz7vrqA+MzMzPgAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAABIAAAAMAAAA AQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAABY AAAApAAAAPMBAAC3AAAABAIAAAEAAAAAQNNBAPDSQaQAAAACAgAAAgAAAEwA AAAAAAAAAAAAAAAAAAD//////////1AAAABSADIADAAAAAAAAAAlAAAADAAA AA0AAIBGAAAAuAAAAKwAAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA 8mDAQgAAAAAAAAAA6KnAQrzfF0QYBABECEAFBjAAAAAkAAAAARDA26uqKj4A AAAAAAAAAAAAAAAFAAAAQQBSAEkAQQBMAAAANkAFgEgAAAA8AAAAAAAA/wEA AAABAAAAAgAAAFIAMwBVdYk+mpkZPlUVxz6amRk+AACAPwAAAAAAAAAAAACA PwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAA JQAAAAwAAAABAAAAVAAAAFgAAAB5AgAAAAIAAIwCAAARAgAAAQAAAABA00EA 8NJBeQIAAA8CAAACAAAATAAAAAAAAAAAAAAAAAAAAP//////////UAAAAFIA MwAMAAAAAAAAACUAAAAMAAAADQAAgEYAAACUAAAAiAAAAEVNRisrQAAADAAA AAAAAAAIQAMCPAAAADAAAAABEMDbAAAAAE4AAAAAAAAApz52PwIAAAACAAAA AgAAAAIAAAABEMDbAAAAAAAAAP8IQAQDLAAAACAAAAABEMDbAgAAAAAAAAB+ W2hDNyzKQ6G+oUM3LMpDAAEAABVABAAQAAAABAAAAAMAAAAkAAAAJAAAAAAA gD0AAAAAAAAAAAAAgD0AAAAAAAAAAAIAAABfAAAAOAAAAAUAAAA4AAAAAAAA ADgAAAAAAAAAAAABAA8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAACUAAAAMAAAA BQAAACUAAAAMAAAABQAAgFcAAAAkAAAA5gAAAJIBAABFAQAAlgEAAAIAAACG DkYZOBRGGSUAAAAMAAAABwAAgCUAAAAMAAAAAAAAgCQAAAAkAAAAAACAQQAA AAAAAAAAAACAQQAAAAAAAAAAAgAAACgAAAAMAAAABQAAAEYAAACIAAAAfAAA AEVNRisIQAMCPAAAADAAAAABEMDbAAAAAE4AAAAAAAAApz52PwIAAAACAAAA AgAAAAIAAAABEMDbAAAAAAAAAP8IQAQDLAAAACAAAAABEMDbAgAAAAAAAADA ZNVDNyzKQ3QKB0Q3LMpDAAEAABVABAAQAAAABAAAAAMAAAAkAAAAJAAAAAAA gD0AAAAAAAAAAAAAgD0AAAAAAAAAAAIAAABfAAAAOAAAAAUAAAA4AAAAAAAA ADgAAAAAAAAAAAABAA8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAACUAAAAMAAAA BQAAACUAAAAMAAAABQAAgFcAAAAkAAAAqQEAAJIBAAAeAgAAlgEAAAIAAACt GkYZwyFGGSUAAAAMAAAABwAAgCUAAAAMAAAAAAAAgCQAAAAkAAAAAACAQQAA AAAAAAAAAACAQQAAAAAAAAAAAgAAACgAAAAMAAAABQAAAEYAAAA4AQAALAEA AEVNRisqQAAAJAAAABgAAADyYMBCAAAAAAAAAADoqcBCXQYlQwCeh0MIQAUG MAAAACQAAAABEMDbq6oqPgAAAAAAAAAAAAAAAAUAAABBAFIASQBBAEwAAAA2 QAWA1AAAAMgAAAAAAAD/AQAAAAEAAAAQAAAATwBTAFAARgAgAFAAcgBvAGMA ZQBzAHMAIAAxADAAMABmBgw+MzMzPjNjiD4zMzM+3k3BPjMzMz6JOPo+MzMz PkQsFz8zMzM+7wYjPzMzMz5EfD8/MzMzPpqxTT8zMzM+RGxlPzMzMz6awXo/ MzMzPiI+iT8zMzM+zeiTPzMzMz53k54/MzMzPs2ApD8zMzM+Il6wPzMzMz53 O7w/MzMzPgAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAABIAAAAMAAAAAQAAABgA AAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAACsAAAAsgAA ABEBAAA5AQAAIgEAAAEAAAAAQNNBAPDSQbIAAAAgAQAAEAAAAEwAAAAAAAAA AAAAAAAAAAD//////////2wAAABPAFMAUABGACAAUAByAG8AYwBlAHMAcwAg ADEAMAAwAAwAAAALAAAACwAAAAoAAAAEAAAACwAAAAUAAAAJAAAACAAAAAkA AAAIAAAACAAAAAQAAAAJAAAACQAAAAAAAAAlAAAADAAAAA0AAIBGAAAARAEA ADgBAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA 6KnAQqv83UN6NYVDCEAFBjAAAAAkAAAAARDA26uqKj4AAAAAAAAAAAAAAAAF AAAAQQBSAEkAQQBMAAAANkAFgNQAAADIAAAAAAAA/wEAAAABAAAAEAAAAE8A UwBQAEYAIABQAHIAbwBjAGUAcwBzACAAMgAwADAAZgYMPmZmZj4zY4g+ZmZm Pt5NwT5mZmY+iTj6PmZmZj5ELBc/ZmZmPu8GIz9mZmY+RHw/P2ZmZj6asU0/ ZmZmPkRsZT9mZmY+msF6P2ZmZj4iPok/ZmZmPs3okz9mZmY+d5OeP2ZmZj7N gKQ/ZmZmPiJesD9mZmY+dzu8P2ZmZj4AAIA/AAAAAAAAAAAAAIA/AAAAAAAA AAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAADAAA AAEAAABUAAAArAAAAMkBAAARAQAAUAIAACIBAAABAAAAAEDTQQDw0kHJAQAA IAEAABAAAABMAAAAAAAAAAAAAAAAAAAA//////////9sAAAATwBTAFAARgAg AFAAcgBvAGMAZQBzAHMAIAAyADAAMAAMAAAACwAAAAsAAAAKAAAABAAAAAsA AAAFAAAACQAAAAgAAAAJAAAACAAAAAgAAAAEAAAACQAAAAkAAAAAAAAAJQAA AAwAAAANAACARgAAAJQAAACIAAAARU1GKytAAAAMAAAAAAAAAAhAAwI8AAAA MAAAAAEQwNsAAAAATgAAAAAAAACnPnY/AgAAAAIAAAACAAAAAgAAAAEQwNsA AAAAAAAA/whABAMsAAAAIAAAAAEQwNsCAAAAAAAAAIPdIEQ3LMpDFLs0RDcs ykMAAQAAFUAEABAAAAAEAAAAAwAAACQAAAAkAAAAAACAPQAAAAAAAAAAAACA PQAAAAAAAAAAAgAAAF8AAAA4AAAABQAAADgAAAAAAAAAOAAAAAAAAAAAAAEA DwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAJQAAAAwAAAAFAAAAJQAAAAwAAAAF AACAVwAAACQAAACCAgAAkgEAANUCAACWAQAAAgAAADgoRhkvLUYZJQAAAAwA AAAHAACAJQAAAAwAAAAAAACAJAAAACQAAAAAAIBBAAAAAAAAAAAAAIBBAAAA AAAAAAACAAAAKAAAAAwAAAAFAAAARgAAAIgAAAB8AAAARU1GKwhAAwI8AAAA MAAAAAEQwNsAAAAATgAAAAAAAACnPnY/AgAAAAIAAAACAAAAAgAAAAEQwNsA AAAAAAAA/whABAMsAAAAIAAAAAEQwNsCAAAAAAAAAEAPAUM3LMpDx6uJQjcs ykMAAQAAFUAEABAAAAAEAAAAAwAAACQAAAAkAAAAAACAPQAAAAAAAAAAAACA PQAAAAAAAAAAAgAAAF8AAAA4AAAABQAAADgAAAAAAAAAOAAAAAAAAAAAAAEA DwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAJQAAAAwAAAAFAAAAJQAAAAwAAAAF AACAVwAAACQAAABDAAAAkgEAAIMAAACWAQAAAgAAABEIRhlOBEYZJQAAAAwA AAAHAACAJQAAAAwAAAAAAACAJAAAACQAAAAAAIBBAAAAAAAAAAAAAIBBAAAA AAAAAAACAAAAKAAAAAwAAAAFAAAARgAAACBUAAAUVAAARU1GKyVAAAAQAAAA BAAAAAAAAAAsQAAAJAAAABgAAAAAsfc9AAAAgAAAAICN8bk9rdTawW33skMI QAYFsFMAAKRTAAABEMDbAgAAAAMAAACUUwAAAQAAAGwAAAD+/////v///zMB AADnAAAAAAAAAAAAAABvHwAAkBcAACBFTUYAAAEAlFMAANsBAAAGAAAAAAAA AAAAAAAAAAAAAAUAAAAEAABSAQAADgEAAAAAAAAAAAAAAAAAAFAoBQCwHgQA RgAAALAiAACiIgAAR0RJQwEAAIAAAwAAuY1p3AAAAACKIgAAAQAJAAADRREA AAYAxgAAAAAABQAAAAwCjxduHwUAAAAMAo8Xbh8HAAAAEAQBAAEAAQABAAMA AAAeAAQAAAAEAQ0ABwAAAPwCAQAAAAAAAAAEAAAALQEAAAkAAAD6AgAAAAAA AAAAAAAiAAQAAAAtAQEACQAAAPoCBgBBAAAAAAAAACIABAAAAC0BAgAEAAAA LQEAAAcAAAD8AgAA////AAAABAAAAC0BAwAEAAAABgEBAMYAAAAkA2EArQRz ESUFqRLABcYTYAbFFCMHoRXwB1UWzAjXFqgJNheYCm0XfgttF18MShdPDfoW Jg5uFu4OuhW2D94UahC6FUYRbhYTEvoW+RJKF+4TbRfFFG0XphU2F4wW1xZt F1UWNRihFfMYxRSdGcYTLhqpEroacxFVG5sRBByMEakcXxFOHfsQ5B1qEFce sQ/FHtAOAR/bDTgf1wxMH8kLOB+xCgEfqAnFHrgIVx7mB+QdKAdOHZwGqRwp BgQc9wVVG+0FuhokBi4a5ASdGb0D8xi5AjUY7AFtFy4BjBaxAKYVTQDFFCAA 7hMgAPkSPgATEp0ARhEkAWoQzgG2D6AC7g7OASYOJAFPDZ0AXww+AH4LIACY CiAAqAlNAMwIsQDwBy4BIwfsAWAGuQLABb0DJQXkBK0EJAYDBO0FWQP3BbQC KQYUApwGgwEoBwsB5gedALgITQCoCSUAsQogAMkLJQDXDE0A2w2dANAOCwGx D4MBahAUAvsQtAJfEVkDjBEDBJsRrQRzEQQAAAAtAQAABAAAAAQBDQDGAAAA JQNhAK0EcxElBakSwAXGE2AGxRQjB6EV8AdVFswI1xaoCTYXmAptF34LbRdf DEoXTw36FiYObhbuDroVtg/eFGoQuhVGEW4WExL6FvkSShfuE20XxRRtF6YV NheMFtcWbRdVFjUYoRXzGMUUnRnGEy4aqRK6GnMRVRubEQQcjBGpHF8RTh37 EOQdahBXHrEPxR7QDgEf2w04H9cMTB/JCzgfsQoBH6gJxR64CFce5gfkHSgH Th2cBqkcKQYEHPcFVRvtBboaJAYuGuQEnRm9A/MYuQI1GOwBbRcuAYwWsQCm FU0AxRQgAO4TIAD5Ej4AExKdAEYRJAFqEM4Btg+gAu4OzgEmDiQBTw2dAF8M PgB+CyAAmAogAKgJTQDMCLEA8AcuASMH7AFgBrkCwAW9AyUF5AStBCQGAwTt BVkD9wW0AikGFAKcBoMBKAcLAeYHnQC4CE0AqAklALEKIADJCyUA1wxNANsN nQDQDgsBsQ+DAWoQFAL7ELQCXxFZA4wRAwSbEa0EcxEEAAAABAENAAkAAAD6 AgYAQQAAAP8AAAAiAAQAAAAtAQQABAAAAPABAgAMAAAAJQMEAGYP7BA2CCwI NhcsCGYP7BAEAAAABAENAAkAAAD6AgYAHgAAAAAAAAAiAAQAAAAtAQIABAAA APABBAAEAAAALQEAAAcAAAD8AgAAf39/AAAABAAAAC0BBAAEAAAABgEBAA4A AAAkAwUAagY0CvEGsgnUCrIJTQo0CmoGNAoEAAAALQEAAAQAAADwAQQABAAA AAQBDQAOAAAAJQMFAGoGNArxBrIJ1AqyCU0KNApqBjQKBAAAAAQBDQAEAAAA LQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAagYv ClIKLwpSCu0KagbtCmoGLwoEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAA JQMFAGoGLwpSCi8KUgrtCmoG7QpqBi8KBAAAAAQBDQAEAAAALQEAAAcAAAD8 AgAApaWlAAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAUgryCtQKZgrUCrIJ Ugo0ClIK8goEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAFIK8grU CmYK1AqyCVIKNApSCvIKBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA2NjYAAAA BAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAbwbxBtgGjQa7Co0GUgrxBm8G8QYE AAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAG8G8QbYBo0GuwqNBlIK 8QZvBvEGBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAE AAAABgEBAA4AAAAkAwUAagYMClIKDApSCuwGagbsBmoGDAoEAAAALQEAAAQA AADwAQQABAAAAAQBDQAOAAAAJQMFAGoGDApSCgwKUgrsBmoG7AZqBgwKBAAA AAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBAAEAAAABgEBAA4A AAAkAwUAuwqIBk0K8QZNCgwKuwqjCbsKiAYEAAAALQEAAAQAAADwAQQABAAA AAQBDQAOAAAAJQMFALsKiAZNCvEGTQoMCrsKowm7CogGBAAAAAQBDQAEAAAA LQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAegXY CxoGEAtcChALvAnYC3oF2AsEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAA JQMFAHoF2AsaBhALXAoQC7wJ2At6BdgLBAAAAAQBDQAEAAAALQEAAAcAAAD8 AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAegXYCxoGEAtcChAL vAnYC3oF2AsEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAHoF2Asa BhALXAoQC7wJ2At6BdgLBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAA BAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAwQkADFcKVgtXChULwQnTC8EJAAwE AAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAMEJAAxXClYLVwoVC8EJ 0wvBCQAMBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBAAE AAAABgEBAA4AAAAkAwUAwQkADFcKVgtXChULwQnTC8EJAAwEAAAALQEAAAQA AADwAQQABAAAAAQBDQAOAAAAJQMFAMEJAAxXClYLVwoVC8EJ0wvBCQAMBAAA AAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4A AAAkAwUAfwXYC8EJ2AvBCQAMfwUADH8F2AsEAAAALQEAAAQAAADwAQQABAAA AAQBDQAOAAAAJQMFAH8F2AvBCdgLwQkADH8FAAx/BdgLBAAAAAQBDQAEAAAA LQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAfwXY C8EJ2AvBCQAMfwUADH8F2AsEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAA JQMFAH8F2AvBCdgLwQkADH8FAAx/BdgLBAAAAAQBDQAEAAAALQEAAAcAAAD8 AgAAAAD/AAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUA8wlQB8kGUAfJBqgJ 8wmoCfMJUAcEAAAALQEAAAQAAAAEAQ0ADgAAACUDBQDzCVAHyQZQB8kGqAnz CagJ8wlQBwQAAAAEAQ0ABQAAABQCmAocCQUAAAATApgK5AkEAAAABAENAAkA AAD6AgYAQQAAAAAAAAAiAAQAAAAtAQUABAAAAPABAgAEAAAALQEAAAQAAAAt AQMABAAAAAYBAQDGAAAAJANhAG4H9AhzBw0JhwcmCZYHPwmgB04JtAdiCcMH ZwnXB3EJ6wdxCf8HcQkTCHEJIghnCTYIYglFCE4JWQg/CWgITgl8CGIJiwhn CZoIcQmzCHEJxwhxCdEIcQnqCGcJ/ghiCQ0JTgkcCT8JKwkmCToJDQlECfQI Tgn0CF0J9AhsCfQIdgnqCIUJ4AiUCcwImQm9CJ4JqQijCZAIowl8CKMJaAie CU8ImQk7CJQJLAiFCRgIdgkOCGwJBAhdCQQITgkECEQJBAg6CesHKwnSBxwJ uQcNCaoH/giWB+oIkQfRCIcHxwiHB7MIhweaCIcHiwiRB3wIlgdoCKoHWQi5 B0UIqgc2CJYHIgiRBxMIhwf/B4cH6weHB9cHhwfDB5EHtAeWB6AHqgeWB7kH hwfSB3MH6wduBwQIXwcECFUHBAhGBwQIPAcOCC0HGAgeBywIGQc7CA8HTwgP B2gIDwd8CA8HkAgPB6kIGQe9CB4HzAgtB+AIPAfqCEYH9AhVB/QIXwf0CG4H 9AgEAAAALQEAAAQAAAAEAQ0AxgAAACUDYQBuB/QIcwcNCYcHJgmWBz8JoAdO CbQHYgnDB2cJ1wdxCesHcQn/B3EJEwhxCSIIZwk2CGIJRQhOCVkIPwloCE4J fAhiCYsIZwmaCHEJswhxCccIcQnRCHEJ6ghnCf4IYgkNCU4JHAk/CSsJJgk6 CQ0JRAn0CE4J9AhdCfQIbAn0CHYJ6giFCeAIlAnMCJkJvQieCakIowmQCKMJ fAijCWgInglPCJkJOwiUCSwIhQkYCHYJDghsCQQIXQkECE4JBAhECQQIOgnr BysJ0gccCbkHDQmqB/4IlgfqCJEH0QiHB8cIhwezCIcHmgiHB4sIkQd8CJYH aAiqB1kIuQdFCKoHNgiWByIIkQcTCIcH/weHB+sHhwfXB4cHwweRB7QHlgeg B6oHlge5B4cH0gdzB+sHbgcECF8HBAhVBwQIRgcECDwHDggtBxgIHgcsCBkH OwgPB08IDwdoCA8HfAgPB5AIDwepCBkHvQgeB8wILQfgCDwH6ghGB/QIVQf0 CF8H9AhuB/QIBAAAAAQBDQAJAAAA+gIGAB4AAAAAAAAAIgAEAAAALQECAAQA AADwAQUABAAAAC0BAAAHAAAA/AIAAH9/fwAAAAQAAAAtAQUABAAAAAYBAQAO AAAAJAMFAMoUNApRFbIJNBmyCa0YNArKFDQKBAAAAC0BAAAEAAAA8AEFAAQA AAAEAQ0ADgAAACUDBQDKFDQKURWyCTQZsgmtGDQKyhQ0CgQAAAAEAQ0ABAAA AC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFAMoU LwqyGC8KshjtCsoU7QrKFC8KBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAA ACUDBQDKFC8KshgvCrIY7QrKFO0KyhQvCgQAAAAEAQ0ABAAAAC0BAAAHAAAA /AIAAKWlpQAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFALIY8go0GWYKNBmy CbIYNAqyGPIKBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQCyGPIK NBlmCjQZsgmyGDQKshjyCgQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAANjY2AAA AAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFAM8U8QY4FY0GGxmNBrIY8QbPFPEG BAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQDPFPEGOBWNBhsZjQay GPEGzxTxBgQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUA BAAAAAYBAQAOAAAAJAMFAMoUDAqyGAwKshjsBsoU7AbKFAwKBAAAAC0BAAAE AAAA8AEFAAQAAAAEAQ0ADgAAACUDBQDKFAwKshgMCrIY7AbKFOwGyhQMCgQA AAAEAQ0ABAAAAC0BAAAHAAAA/AIAAKWlpQAAAAQAAAAtAQUABAAAAAYBAQAO AAAAJAMFABsZiAatGPEGrRgMChsZowkbGYgGBAAAAC0BAAAEAAAA8AEFAAQA AAAEAQ0ADgAAACUDBQAbGYgGrRjxBq0YDAobGaMJGxmIBgQAAAAEAQ0ABAAA AC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFANoT 2At6FBALvBgQCxwY2AvaE9gLBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAA ACUDBQDaE9gLehQQC7wYEAscGNgL2hPYCwQAAAAEAQ0ABAAAAC0BAAAHAAAA /AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFANoT2At6FBALvBgQ CxwY2AvaE9gLBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQDaE9gL ehQQC7wYEAscGNgL2hPYCwQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAKWlpQAA AAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFACEYAAy3GFYLtxgVCyEY0wshGAAM BAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQAhGAAMtxhWC7cYFQsh GNMLIRgADAQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAKWlpQAAAAQAAAAtAQUA BAAAAAYBAQAOAAAAJAMFACEYAAy3GFYLtxgVCyEY0wshGAAMBAAAAC0BAAAE AAAA8AEFAAQAAAAEAQ0ADgAAACUDBQAhGAAMtxhWC7cYFQshGNMLIRgADAQA AAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAO AAAAJAMFAN8T2AshGNgLIRgADN8TAAzfE9gLBAAAAC0BAAAEAAAA8AEFAAQA AAAEAQ0ADgAAACUDBQDfE9gLIRjYCyEYAAzfEwAM3xPYCwQAAAAEAQ0ABAAA AC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFAN8T 2AshGNgLIRgADN8TAAzfE9gLBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAA ACUDBQDfE9gLIRjYCyEYAAzfEwAM3xPYCwQAAAAEAQ0ABAAAAC0BAAAEAAAA LQEEAAQAAAAGAQEADgAAACQDBQBTGFAHKRVQBykVqAlTGKgJUxhQBwQAAAAt AQAABAAAAAQBDQAOAAAAJQMFAFMYUAcpFVAHKRWoCVMYqAlTGFAHBAAAAAQB DQAFAAAAFAKYCnwXBQAAABMCmApEGAQAAAAEAQ0ACQAAAPoCBgBBAAAAAAAA ACIABAAAAC0BBQAEAAAA8AECAAQAAAAtAQAABAAAAC0BAwAEAAAABgEBAMYA AAAkA2EAzhX0CNMVDQnnFSYJ9hU/CQAWTgkUFmIJIxZnCTcWcQlLFnEJXxZx CXMWcQmCFmcJlhZiCaUWTgm5Fj8JyBZOCdwWYgnrFmcJ+hZxCRMXcQknF3EJ MRdxCUoXZwleF2IJbRdOCXwXPwmLFyYJmhcNCaQX9AiuF/QIvRf0CMwX9AjW F+oI5RfgCPQXzAj5F70I/hepCAMYkAgDGHwIAxhoCP4XTwj5FzsI9BcsCOUX GAjWFw4IzBcECL0XBAiuFwQIpBcECJoX6weLF9IHfBe5B20XqgdeF5YHSheR BzEXhwcnF4cHExeHB/oWhwfrFpEH3BaWB8gWqge5FrkHpRaqB5YWlgeCFpEH cxaHB18WhwdLFocHNxaHByMWkQcUFpYHABaqB/YVuQfnFdIH0xXrB84VBAi/ FQQItRUECKYVBAicFQ4IjRUYCH4VLAh5FTsIbxVPCG8VaAhvFXwIbxWQCG8V qQh5Fb0IfhXMCI0V4AicFeoIphX0CLUV9Ai/FfQIzhX0CAQAAAAtAQAABAAA AAQBDQDGAAAAJQNhAM4V9AjTFQ0J5xUmCfYVPwkAFk4JFBZiCSMWZwk3FnEJ SxZxCV8WcQlzFnEJghZnCZYWYgmlFk4JuRY/CcgWTgncFmIJ6xZnCfoWcQkT F3EJJxdxCTEXcQlKF2cJXhdiCW0XTgl8Fz8JixcmCZoXDQmkF/QIrhf0CL0X 9AjMF/QI1hfqCOUX4Aj0F8wI+Re9CP4XqQgDGJAIAxh8CAMYaAj+F08I+Rc7 CPQXLAjlFxgI1hcOCMwXBAi9FwQIrhcECKQXBAiaF+sHixfSB3wXuQdtF6oH XheWB0oXkQcxF4cHJxeHBxMXhwf6FocH6xaRB9wWlgfIFqoHuRa5B6UWqgeW FpYHghaRB3MWhwdfFocHSxaHBzcWhwcjFpEHFBaWBwAWqgf2FbkH5xXSB9MV 6wfOFQQIvxUECLUVBAimFQQInBUOCI0VGAh+FSwIeRU7CG8VTwhvFWgIbxV8 CG8VkAhvFakIeRW9CH4VzAiNFeAInBXqCKYV9Ai1FfQIvxX0CM4V9AgEAAAA BAENAAkAAAD6AgYAHgAAAAAAAAAiAAQAAAAtAQIABAAAAPABBQAEAAAALQEA AAcAAAD8AgAAf39/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUASg20EdEN MhG0ETIRLRG0EUoNtBEEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMF AEoNtBHRDTIRtBEyES0RtBFKDbQRBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA v7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUASg2vETIRrxEyEW0SSg1t EkoNrxEEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFAEoNrxEyEa8R MhFtEkoNbRJKDa8RBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAA AC0BBQAEAAAABgEBAA4AAAAkAwUAMhFyErQR5hG0ETIRMhG0ETIRchIEAAAA LQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFADIRchK0EeYRtBEyETIRtBEy EXISBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA2NjYAAAABAAAAC0BBQAEAAAA BgEBAA4AAAAkAwUATw1xDrgNDQ6bEQ0OMhFxDk8NcQ4EAAAALQEAAAQAAADw AQUABAAAAAQBDQAOAAAAJQMFAE8NcQ64DQ0OmxENDjIRcQ5PDXEOBAAAAAQB DQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAk AwUASg2METIRjBEyEWwOSg1sDkoNjBEEAAAALQEAAAQAAADwAQUABAAAAAQB DQAOAAAAJQMFAEoNjBEyEYwRMhFsDkoNbA5KDYwRBAAAAAQBDQAEAAAALQEA AAcAAAD8AgAApaWlAAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAmxEIDi0R cQ4tEYwRmxEjEZsRCA4EAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMF AJsRCA4tEXEOLRGMEZsRIxGbEQgOBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA v7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAWgxYE/oMkBI8EZASnBBY E1oMWBMEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFAFoMWBP6DJAS PBGQEpwQWBNaDFgTBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAA AC0BBQAEAAAABgEBAA4AAAAkAwUAWgxYE/oMkBI8EZASnBBYE1oMWBMEAAAA LQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFAFoMWBP6DJASPBGQEpwQWBNa DFgTBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBQAEAAAA BgEBAA4AAAAkAwUAoRCAEzcR1hI3EZUSoRBTE6EQgBMEAAAALQEAAAQAAADw AQUABAAAAAQBDQAOAAAAJQMFAKEQgBM3EdYSNxGVEqEQUxOhEIATBAAAAAQB DQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBQAEAAAABgEBAA4AAAAk AwUAoRCAEzcR1hI3EZUSoRBTE6EQgBMEAAAALQEAAAQAAADwAQUABAAAAAQB DQAOAAAAJQMFAKEQgBM3EdYSNxGVEqEQUxOhEIATBAAAAAQBDQAEAAAALQEA AAcAAAD8AgAAv7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAXwxYE6EQ WBOhEIATXwyAE18MWBMEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMF AF8MWBOhEFgToRCAE18MgBNfDFgTBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA v7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAXwxYE6EQWBOhEIATXwyA E18MWBMEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFAF8MWBOhEFgT oRCAE18MgBNfDFgTBAAAAAQBDQAEAAAALQEAAAQAAAAtAQQABAAAAAYBAQAO AAAAJAMFANMQ0A6pDdAOqQ0oEdMQKBHTENAOBAAAAC0BAAAEAAAABAENAA4A AAAlAwUA0xDQDqkN0A6pDSgR0xAoEdMQ0A4EAAAABAENAAUAAAAUAhgS/A8F AAAAEwIYEsQQBAAAAAQBDQAJAAAA+gIGAEEAAAAAAAAAIgAEAAAALQEFAAQA AADwAQIABAAAAC0BAAAEAAAALQEDAAQAAAAGAQEAxgAAACQDYQBODnQQUw6N EGcOphB2Dr8QgA7OEJQO4hCjDucQtw7xEMsO8RDfDvEQ8w7xEAIP5xAWD+IQ JQ/OEDkPvxBID84QXA/iEGsP5xB6D/EQkw/xEKcP8RCxD/EQyg/nEN4P4hDt D84Q/A+/EAsQphAaEI0QJBB0EC4QdBA9EHQQTBB0EFYQahBlEGAQdBBMEHkQ PRB+ECkQgxAQEIMQ/A+DEOgPfhDPD3kQuw90EKwPZRCYD1YQjg9MEIQPPRCE Dy4QhA8kEIQPGhBrDwsQUg/8DzkP7Q8qD94PFg/KDxEPsQ8HD6cPBw+TDwcP eg8HD2sPEQ9cDxYPSA8qDzkPOQ8lDyoPFg8WDwIPEQ/zDgcP3w4HD8sOBw+3 DgcPow4RD5QOFg+ADioPdg45D2cOUg9TDmsPTg6EDz8OhA81DoQPJg6EDxwO jg8NDpgP/g2sD/kNuw/vDc8P7w3oD+8N/A/vDRAQ7w0pEPkNPRD+DUwQDQ5g EBwOahAmDnQQNQ50ED8OdBBODnQQBAAAAC0BAAAEAAAABAENAMYAAAAlA2EA Tg50EFMOjRBnDqYQdg6/EIAOzhCUDuIQow7nELcO8RDLDvEQ3w7xEPMO8RAC D+cQFg/iECUPzhA5D78QSA/OEFwP4hBrD+cQeg/xEJMP8RCnD/EQsQ/xEMoP 5xDeD+IQ7Q/OEPwPvxALEKYQGhCNECQQdBAuEHQQPRB0EEwQdBBWEGoQZRBg EHQQTBB5ED0QfhApEIMQEBCDEPwPgxDoD34Qzw95ELsPdBCsD2UQmA9WEI4P TBCEDz0QhA8uEIQPJBCEDxoQaw8LEFIP/A85D+0PKg/eDxYPyg8RD7EPBw+n DwcPkw8HD3oPBw9rDxEPXA8WD0gPKg85DzkPJQ8qDxYPFg8CDxEP8w4HD98O Bw/LDgcPtw4HD6MOEQ+UDhYPgA4qD3YOOQ9nDlIPUw5rD04OhA8/DoQPNQ6E DyYOhA8cDo4PDQ6YD/4NrA/5DbsP7w3PD+8N6A/vDfwP7w0QEO8NKRD5DT0Q /g1MEA0OYBAcDmoQJg50EDUOdBA/DnQQTg50EAQAAAAEAQ0ABAAAACcB//8e AAAAJgYPADEAQ29weXJpZ2h0IKkgMTk5NiBNaWNyb3N5c3RlbXMgRW5naW5l ZXJpbmcgQ29tcGFueQAPAAAAJgYPABMAQWxsIFJpZ2h0cyBSZXNlcnZlZAAE AAAA8AEBAAMAAAAAAAAAEQAAAAwAAAAIAAAACwAAABAAAAAxAQAA5QAAAAkA AAAQAAAAMQEAAOUAAAAJAAAAEAAAAG4fAACPFwAACQAAABAAAABuHwAAjxcA ACAAAAAYAAAAAQAAAAEAAAABAAAAAQAAACEAAAAIAAAAFAAAAAwAAAANAAAA JQAAAAwAAAAFAACAJgAAABwAAAABAAAAAAAAAAAAAAAAAAAAAAAAACUAAAAM AAAAAQAAACYAAAAcAAAAAgAAAAYAAABBAAAAAAAAAAAAAAAlAAAADAAAAAIA AAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAA////AAAAAAAlAAAADAAA AAMAAAATAAAADAAAAAEAAABWAAAAoAEAAP7////+////MwEAAOcAAABhAAAA rQRzESUFqRLABcYTYAbFFCMHoRXwB1UWzAjXFqgJNheYCm0XfgttF18MShdP DfoWJg5uFu4OuhW2D94UahC6FUYRbhYTEvoW+RJKF+4TbRfFFG0XphU2F4wW 1xZtF1UWNRihFfMYxRSdGcYTLhqpEroacxFVG5sRBByMEakcXxFOHfsQ5B1q EFcesQ/FHtAOAR/bDTgf1wxMH8kLOB+xCgEfqAnFHrgIVx7mB+QdKAdOHZwG qRwpBgQc9wVVG+0FuhokBi4a5ASdGb0D8xi5AjUY7AFtFy4BjBaxAKYVTQDF FCAA7hMgAPkSPgATEp0ARhEkAWoQzgG2D6AC7g7OASYOJAFPDZ0AXww+AH4L IACYCiAAqAlNAMwIsQDwBy4BIwfsAWAGuQLABb0DJQXkBK0EJAYDBO0FWQP3 BbQCKQYUApwGgwEoBwsB5gedALgITQCoCSUAsQogAMkLJQDXDE0A2w2dANAO CwGxD4MBahAUAvsQtAJfEVkDjBEDBJsRrQRzESUAAAAMAAAABQAAgBQAAAAM AAAADQAAAFcAAACgAQAA/v////7///8zAQAA5wAAAGEAAACtBHMRJQWpEsAF xhNgBsUUIwehFfAHVRbMCNcWqAk2F5gKbRd+C20XXwxKF08N+hYmDm4W7g66 FbYP3hRqELoVRhFuFhMS+hb5EkoX7hNtF8UUbRemFTYXjBbXFm0XVRY1GKEV 8xjFFJ0ZxhMuGqkSuhpzEVUbmxEEHIwRqRxfEU4d+xDkHWoQVx6xD8Ue0A4B H9sNOB/XDEwfyQs4H7EKAR+oCcUeuAhXHuYH5B0oB04dnAapHCkGBBz3BVUb 7QW6GiQGLhrkBJ0ZvQPzGLkCNRjsAW0XLgGMFrEAphVNAMUUIADuEyAA+RI+ ABMSnQBGESQBahDOAbYPoALuDs4BJg4kAU8NnQBfDD4AfgsgAJgKIACoCU0A zAixAPAHLgEjB+wBYAa5AsAFvQMlBeQErQQkBgME7QVZA/cFtAIpBhQCnAaD ASgHCwHmB50AuAhNAKgJJQCxCiAAyQslANcMTQDbDZ0A0A4LAbEPgwFqEBQC +xC0Al8RWQOMEQMEmxGtBHMRFAAAAAwAAAANAAAAJgAAABwAAAAEAAAABgAA AEEAAAAAAAAA/wAAACUAAAAMAAAABAAAACgAAAAMAAAAAgAAAFcAAAAsAAAA TQAAAEwAAADkAAAAqAAAAAQAAABmD+wQNggsCDYXLAhmD+wQFAAAAAwAAAAN AAAAJgAAABwAAAACAAAABgAAAB4AAAAAAAAAAAAAACUAAAAMAAAAAgAAACgA AAAMAAAABAAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAB/f38AAAAA ACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAPgAAAF4AAABpAAAA YwAAAAUAAABqBjQK8QayCdQKsglNCjQKagY0CiUAAAAMAAAABQAAgCgAAAAM AAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAPgAAAF4AAABpAAAAYwAAAAUA AABqBjQK8QayCdQKsglNCjQKagY0ChQAAAAMAAAADQAAACUAAAAMAAAABQAA gCcAAAAYAAAABAAAAAAAAAC/v78AAAAAACUAAAAMAAAABAAAABMAAAAMAAAA AQAAAFYAAAAwAAAAPgAAAGMAAABkAAAAagAAAAUAAABqBi8KUgovClIK7Qpq Bu0KagYvCiUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcA AAAwAAAAPgAAAGMAAABkAAAAagAAAAUAAABqBi8KUgovClIK7QpqBu0KagYv ChQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAClpaUA AAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAZAAAAF4AAABp AAAAagAAAAUAAABSCvIK1ApmCtQKsglSCjQKUgryCiUAAAAMAAAABQAAgCgA AAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAZAAAAF4AAABpAAAAagAA AAUAAABSCvIK1ApmCtQKsglSCjQKUgryChQAAAAMAAAADQAAACUAAAAMAAAA BQAAgCcAAAAYAAAABAAAAAAAAADY2NgAAAAAACUAAAAMAAAABAAAABMAAAAM AAAAAQAAAFYAAAAwAAAAPgAAAEAAAABoAAAARAAAAAUAAABvBvEG2AaNBrsK jQZSCvEGbwbxBiUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAA AFcAAAAwAAAAPgAAAEAAAABoAAAAQwAAAAUAAABvBvEG2AaNBrsKjQZSCvEG bwbxBhQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAC/ v78AAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAPgAAAEMA AABkAAAAYgAAAAUAAABqBgwKUgoMClIK7AZqBuwGagYMCiUAAAAMAAAABQAA gCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAPgAAAEMAAABkAAAA YgAAAAUAAABqBgwKUgoMClIK7AZqBuwGagYMChQAAAAMAAAADQAAACUAAAAM AAAABQAAgCcAAAAYAAAABAAAAAAAAAClpaUAAAAAACUAAAAMAAAABAAAABMA AAAMAAAAAQAAAFYAAAAwAAAAZAAAAEAAAABoAAAAYgAAAAUAAAC7CogGTQrx Bk0KDAq7CqMJuwqIBiUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAA DQAAAFcAAAAwAAAAZAAAAD8AAABoAAAAYgAAAAUAAAC7CogGTQrxBk0KDAq7 CqMJuwqIBhQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAA AAC/v78AAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAANQAA AGwAAABlAAAAcwAAAAUAAAB6BdgLGgYQC1wKEAu8CdgLegXYCyUAAAAMAAAA BQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAANQAAAGwAAABl AAAAcwAAAAUAAAB6BdgLGgYQC1wKEAu8CdgLegXYCxQAAAAMAAAADQAAACUA AAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAC/v78AAAAAACUAAAAMAAAABAAA ABMAAAAMAAAAAQAAAFYAAAAwAAAANQAAAGwAAABlAAAAcwAAAAUAAAB6BdgL GgYQC1wKEAu8CdgLegXYCyUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAM AAAADQAAAFcAAAAwAAAANQAAAGwAAABlAAAAcwAAAAUAAAB6BdgLGgYQC1wK EAu8CdgLegXYCxQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAA AAAAAAClpaUAAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAA XwAAAGwAAABkAAAAdQAAAAUAAADBCQAMVwpWC1cKFQvBCdMLwQkADCUAAAAM AAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAXwAAAGwA AABkAAAAdQAAAAUAAADBCQAMVwpWC1cKFQvBCdMLwQkADBQAAAAMAAAADQAA ACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAClpaUAAAAAACUAAAAMAAAA BAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAXwAAAGwAAABkAAAAdQAAAAUAAADB CQAMVwpWC1cKFQvBCdMLwQkADCUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQA AAAMAAAADQAAAFcAAAAwAAAAXwAAAGwAAABkAAAAdQAAAAUAAADBCQAMVwpW C1cKFQvBCdMLwQkADBQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAA BAAAAAAAAAC/v78AAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAw AAAANQAAAHMAAABfAAAAdQAAAAUAAAB/BdgLwQnYC8EJAAx/BQAMfwXYCyUA AAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAANQAA AHMAAABfAAAAdQAAAAUAAAB/BdgLwQnYC8EJAAx/BQAMfwXYCxQAAAAMAAAA DQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAC/v78AAAAAACUAAAAM AAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAANQAAAHMAAABfAAAAdQAAAAUA AAB/BdgLwQnYC8EJAAx/BQAMfwXYCyUAAAAMAAAABQAAgCgAAAAMAAAABAAA ABQAAAAMAAAADQAAAFcAAAAwAAAANQAAAHMAAABfAAAAdQAAAAUAAAB/BdgL wQnYC8EJAAx/BQAMfwXYCxQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAY AAAABAAAAAAAAAAAAP8AAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYA AAAwAAAAQgAAAEcAAABhAAAAXgAAAAUAAADzCVAHyQZQB8kGqAnzCagJ8wlQ ByUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAAwAAAAQgAAAEcAAABhAAAA XgAAAAUAAADzCVAHyQZQB8kGqAnzCagJ8wlQBxQAAAAMAAAADQAAABsAAAAQ AAAAHAkAAJgKAAA2AAAAEAAAAOQJAACYCgAAFAAAAAwAAAANAAAAJgAAABwA AAAFAAAABgAAAEEAAAAAAAAAAAAAACUAAAAMAAAABQAAACgAAAAMAAAAAgAA ACUAAAAMAAAABQAAgCUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAACgAQAA QgAAAEYAAABhAAAAXwAAAGEAAABuB/QIcwcNCYcHJgmWBz8JoAdOCbQHYgnD B2cJ1wdxCesHcQn/B3EJEwhxCSIIZwk2CGIJRQhOCVkIPwloCE4JfAhiCYsI ZwmaCHEJswhxCccIcQnRCHEJ6ghnCf4IYgkNCU4JHAk/CSsJJgk6CQ0JRAn0 CE4J9AhdCfQIbAn0CHYJ6giFCeAIlAnMCJkJvQieCakIowmQCKMJfAijCWgI nglPCJkJOwiUCSwIhQkYCHYJDghsCQQIXQkECE4JBAhECQQIOgnrBysJ0gcc CbkHDQmqB/4IlgfqCJEH0QiHB8cIhwezCIcHmgiHB4sIkQd8CJYHaAiqB1kI uQdFCKoHNgiWByIIkQcTCIcH/weHB+sHhwfXB4cHwweRB7QHlgegB6oHlge5 B4cH0gdzB+sHbgcECF8HBAhVBwQIRgcECDwHDggtBxgIHgcsCBkHOwgPB08I DwdoCA8HfAgPB5AIDwepCBkHvQgeB8wILQfgCDwH6ghGB/QIVQf0CF8H9Ahu B/QIJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAAKABAABCAAAARgAAAGEA AABfAAAAYQAAAG4H9AhzBw0JhwcmCZYHPwmgB04JtAdiCcMHZwnXB3EJ6wdx Cf8HcQkTCHEJIghnCTYIYglFCE4JWQg/CWgITgl8CGIJiwhnCZoIcQmzCHEJ xwhxCdEIcQnqCGcJ/ghiCQ0JTgkcCT8JKwkmCToJDQlECfQITgn0CF0J9Ahs CfQIdgnqCIUJ4AiUCcwImQm9CJ4JqQijCZAIowl8CKMJaAieCU8ImQk7CJQJ LAiFCRgIdgkOCGwJBAhdCQQITgkECEQJBAg6CesHKwnSBxwJuQcNCaoH/giW B+oIkQfRCIcHxwiHB7MIhweaCIcHiwiRB3wIlgdoCKoHWQi5B0UIqgc2CJYH IgiRBxMIhwf/B4cH6weHB9cHhwfDB5EHtAeWB6AHqgeWB7kHhwfSB3MH6wdu BwQIXwcECFUHBAhGBwQIPAcOCC0HGAgeBywIGQc7CA8HTwgPB2gIDwd8CA8H kAgPB6kIGQe9CB4HzAgtB+AIPAfqCEYH9AhVB/QIXwf0CG4H9AgUAAAADAAA AA0AAAAmAAAAHAAAAAIAAAAGAAAAHgAAAAAAAAAAAAAAJQAAAAwAAAACAAAA KAAAAAwAAAAFAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAH9/fwAA AAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADKAAAAXgAAAPUA AABjAAAABQAAAMoUNApRFbIJNBmyCa0YNArKFDQKJQAAAAwAAAAFAACAKAAA AAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADKAAAAXgAAAPUAAABjAAAA BQAAAMoUNApRFbIJNBmyCa0YNArKFDQKFAAAAAwAAAANAAAAJQAAAAwAAAAF AACAJwAAABgAAAAFAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwA AAABAAAAVgAAADAAAADKAAAAYwAAAPAAAABqAAAABQAAAMoULwqyGC8Kshjt CsoU7QrKFC8KJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAA VwAAADAAAADKAAAAYwAAAPAAAABqAAAABQAAAMoULwqyGC8KshjtCsoU7QrK FC8KFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAKWl pQAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADwAAAAXgAA APUAAABqAAAABQAAALIY8go0GWYKNBmyCbIYNAqyGPIKJQAAAAwAAAAFAACA KAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADwAAAAXgAAAPUAAABq AAAABQAAALIY8go0GWYKNBmyCbIYNAqyGPIKFAAAAAwAAAANAAAAJQAAAAwA AAAFAACAJwAAABgAAAAFAAAAAAAAANjY2AAAAAAAJQAAAAwAAAAFAAAAEwAA AAwAAAABAAAAVgAAADAAAADKAAAAQAAAAPQAAABEAAAABQAAAM8U8QY4FY0G GxmNBrIY8QbPFPEGJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAAN AAAAVwAAADAAAADKAAAAQAAAAPQAAABDAAAABQAAAM8U8QY4FY0GGxmNBrIY 8QbPFPEGFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAA AL+/vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADKAAAA QwAAAPAAAABiAAAABQAAAMoUDAqyGAwKshjsBsoU7AbKFAwKJQAAAAwAAAAF AACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADKAAAAQwAAAPAA AABiAAAABQAAAMoUDAqyGAwKshjsBsoU7AbKFAwKFAAAAAwAAAANAAAAJQAA AAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAKWlpQAAAAAAJQAAAAwAAAAFAAAA EwAAAAwAAAABAAAAVgAAADAAAADvAAAAQAAAAPQAAABiAAAABQAAABsZiAat GPEGrRgMChsZowkbGYgGJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwA AAANAAAAVwAAADAAAADvAAAAPwAAAPQAAABiAAAABQAAABsZiAatGPEGrRgM ChsZowkbGYgGFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAA AAAAAL+/vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADB AAAAbAAAAPAAAABzAAAABQAAANoT2At6FBALvBgQCxwY2AvaE9gLJQAAAAwA AAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADBAAAAbAAA APAAAABzAAAABQAAANoT2At6FBALvBgQCxwY2AvaE9gLFAAAAAwAAAANAAAA JQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAF AAAAEwAAAAwAAAABAAAAVgAAADAAAADBAAAAbAAAAPAAAABzAAAABQAAANoT 2At6FBALvBgQCxwY2AvaE9gLJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAA AAwAAAANAAAAVwAAADAAAADBAAAAbAAAAPAAAABzAAAABQAAANoT2At6FBAL vBgQCxwY2AvaE9gLFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAF AAAAAAAAAKWlpQAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAA AADqAAAAbAAAAPAAAAB1AAAABQAAACEYAAy3GFYLtxgVCyEY0wshGAAMJQAA AAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADqAAAA bAAAAPAAAAB1AAAABQAAACEYAAy3GFYLtxgVCyEY0wshGAAMFAAAAAwAAAAN AAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAKWlpQAAAAAAJQAAAAwA AAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADqAAAAbAAAAPAAAAB1AAAABQAA ACEYAAy3GFYLtxgVCyEY0wshGAAMJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAA FAAAAAwAAAANAAAAVwAAADAAAADqAAAAbAAAAPAAAAB1AAAABQAAACEYAAy3 GFYLtxgVCyEY0wshGAAMFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgA AAAFAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAA ADAAAADBAAAAcwAAAOoAAAB1AAAABQAAAN8T2AshGNgLIRgADN8TAAzfE9gL JQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADB AAAAcwAAAOoAAAB1AAAABQAAAN8T2AshGNgLIRgADN8TAAzfE9gLFAAAAAwA AAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAL+/vwAAAAAAJQAA AAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADBAAAAcwAAAOoAAAB1AAAA BQAAAN8T2AshGNgLIRgADN8TAAzfE9gLJQAAAAwAAAAFAACAKAAAAAwAAAAF AAAAFAAAAAwAAAANAAAAVwAAADAAAADBAAAAcwAAAOoAAAB1AAAABQAAAN8T 2AshGNgLIRgADN8TAAzfE9gLFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJQAA AAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADNAAAARwAAAOwAAABeAAAA BQAAAFMYUAcpFVAHKRWoCVMYqAlTGFAHJQAAAAwAAAAFAACAFAAAAAwAAAAN AAAAVwAAADAAAADNAAAARwAAAOwAAABeAAAABQAAAFMYUAcpFVAHKRWoCVMY qAlTGFAHFAAAAAwAAAANAAAAGwAAABAAAAB8FwAAmAoAADYAAAAQAAAARBgA AJgKAAAUAAAADAAAAA0AAAAmAAAAHAAAAAUAAAAGAAAAQQAAAAAAAAAAAAAA JQAAAAwAAAAFAAAAKAAAAAwAAAACAAAAJQAAAAwAAAAFAACAJQAAAAwAAAAD AAAAEwAAAAwAAAABAAAAVgAAAKABAADNAAAARgAAAOwAAABfAAAAYQAAAM4V 9AjTFQ0J5xUmCfYVPwkAFk4JFBZiCSMWZwk3FnEJSxZxCV8WcQlzFnEJghZn CZYWYgmlFk4JuRY/CcgWTgncFmIJ6xZnCfoWcQkTF3EJJxdxCTEXcQlKF2cJ XhdiCW0XTgl8Fz8JixcmCZoXDQmkF/QIrhf0CL0X9AjMF/QI1hfqCOUX4Aj0 F8wI+Re9CP4XqQgDGJAIAxh8CAMYaAj+F08I+Rc7CPQXLAjlFxgI1hcOCMwX BAi9FwQIrhcECKQXBAiaF+sHixfSB3wXuQdtF6oHXheWB0oXkQcxF4cHJxeH BxMXhwf6FocH6xaRB9wWlgfIFqoHuRa5B6UWqgeWFpYHghaRB3MWhwdfFocH SxaHBzcWhwcjFpEHFBaWBwAWqgf2FbkH5xXSB9MV6wfOFQQIvxUECLUVBAim FQQInBUOCI0VGAh+FSwIeRU7CG8VTwhvFWgIbxV8CG8VkAhvFakIeRW9CH4V zAiNFeAInBXqCKYV9Ai1FfQIvxX0CM4V9AglAAAADAAAAAUAAIAUAAAADAAA AA0AAABXAAAAoAEAAM0AAABGAAAA7AAAAF8AAABhAAAAzhX0CNMVDQnnFSYJ 9hU/CQAWTgkUFmIJIxZnCTcWcQlLFnEJXxZxCXMWcQmCFmcJlhZiCaUWTgm5 Fj8JyBZOCdwWYgnrFmcJ+hZxCRMXcQknF3EJMRdxCUoXZwleF2IJbRdOCXwX PwmLFyYJmhcNCaQX9AiuF/QIvRf0CMwX9AjWF+oI5RfgCPQXzAj5F70I/hep CAMYkAgDGHwIAxhoCP4XTwj5FzsI9BcsCOUXGAjWFw4IzBcECL0XBAiuFwQI pBcECJoX6weLF9IHfBe5B20XqgdeF5YHSheRBzEXhwcnF4cHExeHB/oWhwfr FpEH3BaWB8gWqge5FrkHpRaqB5YWlgeCFpEHcxaHB18WhwdLFocHNxaHByMW kQcUFpYHABaqB/YVuQfnFdIH0xXrB84VBAi/FQQItRUECKYVBAicFQ4IjRUY CH4VLAh5FTsIbxVPCG8VaAhvFXwIbxWQCG8VqQh5Fb0IfhXMCI0V4AicFeoI phX0CLUV9Ai/FfQIzhX0CBQAAAAMAAAADQAAACYAAAAcAAAAAgAAAAYAAAAe AAAAAAAAAAAAAAAlAAAADAAAAAIAAAAoAAAADAAAAAUAAAAlAAAADAAAAAUA AIAnAAAAGAAAAAUAAAAAAAAAf39/AAAAAAAlAAAADAAAAAUAAAATAAAADAAA AAEAAABWAAAAMAAAAIEAAACnAAAArAAAAKwAAAAFAAAASg20EdENMhG0ETIR LRG0EUoNtBElAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABX AAAAMAAAAIEAAACnAAAArAAAAKwAAAAFAAAASg20EdENMhG0ETIRLRG0EUoN tBEUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/ AAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAIEAAACsAAAA pwAAALMAAAAFAAAASg2vETIRrxEyEW0SSg1tEkoNrxElAAAADAAAAAUAAIAo AAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAIEAAACsAAAApwAAALMA AAAFAAAASg2vETIRrxEyEW0SSg1tEkoNrxEUAAAADAAAAA0AAAAlAAAADAAA AAUAAIAnAAAAGAAAAAUAAAAAAAAApaWlAAAAAAAlAAAADAAAAAUAAAATAAAA DAAAAAEAAABWAAAAMAAAAKcAAACnAAAArAAAALMAAAAFAAAAMhFyErQR5hG0 ETIRMhG0ETIRchIlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0A AABXAAAAMAAAAKcAAACnAAAArAAAALMAAAAFAAAAMhFyErQR5hG0ETIRMhG0 ETIRchIUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAA 2NjYAAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAIEAAACJ AAAAqwAAAIwAAAAFAAAATw1xDrgNDQ6bEQ0OMhFxDk8NcQ4lAAAADAAAAAUA AIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAIEAAACJAAAAqwAA AIwAAAAFAAAATw1xDrgNDQ6bEQ0OMhFxDk8NcQ4UAAAADAAAAA0AAAAlAAAA DAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAUAAAAT AAAADAAAAAEAAABWAAAAMAAAAIEAAACMAAAApwAAAKsAAAAFAAAASg2METIR jBEyEWwOSg1sDkoNjBElAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAA AA0AAABXAAAAMAAAAIEAAACMAAAApwAAAKsAAAAFAAAASg2METIRjBEyEWwO Sg1sDkoNjBEUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAA AAAApaWlAAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAKcA AACIAAAAqwAAAKsAAAAFAAAAmxEIDi0RcQ4tEYwRmxEjEZsRCA4lAAAADAAA AAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAKcAAACIAAAA qwAAAKsAAAAFAAAAmxEIDi0RcQ4tEYwRmxEjEZsRCA4UAAAADAAAAA0AAAAl AAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAUA AAATAAAADAAAAAEAAABWAAAAMAAAAHgAAAC0AAAApwAAALwAAAAFAAAAWgxY E/oMkBI8EZASnBBYE1oMWBMlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAA DAAAAA0AAABXAAAAMAAAAHgAAAC0AAAApwAAALwAAAAFAAAAWgxYE/oMkBI8 EZASnBBYE1oMWBMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUA AAAAAAAAv7+/AAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAA AHgAAAC0AAAApwAAALwAAAAFAAAAWgxYE/oMkBI8EZASnBBYE1oMWBMlAAAA DAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAHgAAAC0 AAAApwAAALwAAAAFAAAAWgxYE/oMkBI8EZASnBBYE1oMWBMUAAAADAAAAA0A AAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAApaWlAAAAAAAlAAAADAAA AAUAAAATAAAADAAAAAEAAABWAAAAMAAAAKEAAAC1AAAApwAAAL4AAAAFAAAA oRCAEzcR1hI3EZUSoRBTE6EQgBMlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAU AAAADAAAAA0AAABXAAAAMAAAAKEAAAC1AAAApwAAAL4AAAAFAAAAoRCAEzcR 1hI3EZUSoRBTE6EQgBMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAA AAUAAAAAAAAApaWlAAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAA MAAAAKEAAAC1AAAApwAAAL4AAAAFAAAAoRCAEzcR1hI3EZUSoRBTE6EQgBMl AAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAKEA AAC1AAAApwAAAL4AAAAFAAAAoRCAEzcR1hI3EZUSoRBTE6EQgBMUAAAADAAA AA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/AAAAAAAlAAAA DAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAHgAAAC8AAAAoQAAAL4AAAAF AAAAXwxYE6EQWBOhEIATXwyAE18MWBMlAAAADAAAAAUAAIAoAAAADAAAAAUA AAAUAAAADAAAAA0AAABXAAAAMAAAAHgAAAC8AAAAoQAAAL4AAAAFAAAAXwxY E6EQWBOhEIATXwyAE18MWBMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAA GAAAAAUAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABW AAAAMAAAAHgAAAC8AAAAoQAAAL4AAAAFAAAAXwxYE6EQWBOhEIATXwyAE18M WBMlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAA AHgAAAC8AAAAoQAAAL4AAAAFAAAAXwxYE6EQWBOhEIATXwyAE18MWBMUAAAA DAAAAA0AAAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAAAAEAAABW AAAAMAAAAIUAAACQAAAAowAAAKcAAAAFAAAA0xDQDqkN0A6pDSgR0xAoEdMQ 0A4lAAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAMAAAAIUAAACQAAAAowAA AKcAAAAFAAAA0xDQDqkN0A6pDSgR0xAoEdMQ0A4UAAAADAAAAA0AAAAbAAAA EAAAAPwPAAAYEgAANgAAABAAAADEEAAAGBIAABQAAAAMAAAADQAAACYAAAAc AAAABQAAAAYAAABBAAAAAAAAAAAAAAAlAAAADAAAAAUAAAAoAAAADAAAAAIA AAAlAAAADAAAAAUAAIAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAoAEA AIQAAACPAAAAowAAAKgAAABhAAAATg50EFMOjRBnDqYQdg6/EIAOzhCUDuIQ ow7nELcO8RDLDvEQ3w7xEPMO8RACD+cQFg/iECUPzhA5D78QSA/OEFwP4hBr D+cQeg/xEJMP8RCnD/EQsQ/xEMoP5xDeD+IQ7Q/OEPwPvxALEKYQGhCNECQQ dBAuEHQQPRB0EEwQdBBWEGoQZRBgEHQQTBB5ED0QfhApEIMQEBCDEPwPgxDo D34Qzw95ELsPdBCsD2UQmA9WEI4PTBCEDz0QhA8uEIQPJBCEDxoQaw8LEFIP /A85D+0PKg/eDxYPyg8RD7EPBw+nDwcPkw8HD3oPBw9rDxEPXA8WD0gPKg85 DzkPJQ8qDxYPFg8CDxEP8w4HD98OBw/LDgcPtw4HD6MOEQ+UDhYPgA4qD3YO OQ9nDlIPUw5rD04OhA8/DoQPNQ6EDyYOhA8cDo4PDQ6YD/4NrA/5DbsP7w3P D+8N6A/vDfwP7w0QEO8NKRD5DT0Q/g1MEA0OYBAcDmoQJg50EDUOdBA/DnQQ Tg50ECUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAACgAQAAhAAAAI8AAACj AAAAqAAAAGEAAABODnQQUw6NEGcOphB2Dr8QgA7OEJQO4hCjDucQtw7xEMsO 8RDfDvEQ8w7xEAIP5xAWD+IQJQ/OEDkPvxBID84QXA/iEGsP5xB6D/EQkw/x EKcP8RCxD/EQyg/nEN4P4hDtD84Q/A+/EAsQphAaEI0QJBB0EC4QdBA9EHQQ TBB0EFYQahBlEGAQdBBMEHkQPRB+ECkQgxAQEIMQ/A+DEOgPfhDPD3kQuw90 EKwPZRCYD1YQjg9MEIQPPRCEDy4QhA8kEIQPGhBrDwsQUg/8DzkP7Q8qD94P Fg/KDxEPsQ8HD6cPBw+TDwcPeg8HD2sPEQ9cDxYPSA8qDzkPOQ8lDyoPFg8W DwIPEQ/zDgcP3w4HD8sOBw+3DgcPow4RD5QOFg+ADioPdg45D2cOUg9TDmsP Tg6EDz8OhA81DoQPJg6EDxwOjg8NDpgP/g2sD/kNuw/vDc8P7w3oD+8N/A/v DRAQ7w0pEPkNPRD+DUwQDQ5gEBwOahAmDnQQNQ50ED8OdBBODnQQFAAAAAwA AAANAAAAIgAAAAwAAAD/////KAAAAAwAAAABAAAAJQAAAAwAAAAHAACAJQAA AAwAAAAAAACAMAAAAAwAAAAPAACASwAAABAAAAAAAAAABQAAACgAAAAMAAAA AwAAACgAAAAMAAAABAAAACgAAAAMAAAABQAAAA4AAAAUAAAAAAAAABAAAAAU AAAAGkAGQCwAAAAgAAAA/////wIAAAAAAAAAAAAAAIXemEPoxGVDAAAAAOgD 6AMhAAAACAAAAGIAAAAMAAAAAQAAACEAAAAIAAAAIQAAAAgAAABzAAAADAAA AAAAAAAcAAAACAAAACUAAAAMAAAAAAAAgCUAAAAMAAAABwAAgCUAAAAMAAAA DgAAgDAAAAAMAAAADwAAgBkAAAAMAAAA////ABgAAAAMAAAAAAAAABIAAAAM AAAAAgAAABMAAAAMAAAAAQAAABQAAAAMAAAADQAAABUAAAAMAAAAAQAAABYA AAAMAAAAAAAAAA0AAAAQAAAAAAAAAAAAAAA6AAAADAAAAAoAAAAbAAAAEAAA AAAAAAAAAAAAIwAAACAAAACRnck+AAAAAAAAAAD8bMk+AADYwQAAs0MkAAAA JAAAAJGdyT4AAAAAAAAAAPxsyT4AANjBAACzQwQAAABzAAAADAAAAAAAAAAN AAAAEAAAAOX///9mAQAAUgAAAHABAAAFAAAAFAAAAAoAAAAAAAAAAAAAAJAB AAAAAACBBwICIlMAeQBzAHQAZQBtAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAgAAAA AAAAAAAAAABIZfF33GTxdwkAAAAhAIoBDQAAAJTfEgCFNPJ30CRkAwQaId7Q JGQDJgCKAQQaId4LAIgB//8BAAQAAADc3BIApN0SANDdEgAY7pB8cAWRfMjd EgAAADcAMgeRfAUAAAB4BzcAAAA3AAAACwGg3RIAIDD1d+TfEgAY7pB8OAeR fP////8yB5F8qwaRfOsGkXwHxMJ3HAAAAP////84Zw8Q7PUVAAgeAGAAABUA AAAAAAAAAAAWQwBg7PUVADhnDxBU8RUAwGBOYDDeEgA4Zw8QEEIAYAAAAAAB AAAAayUAYOxpDxA4Zw8QZHYACAAAAAAlAAAADAAAAAUAAABGAAAAKAAAABwA AABHRElDAgAAAAAAAAAAAAAAMQEAAOUAAAAAAAAAIwAAACAAAACRnck+AAAA AAAAAAD8bMk+AADYwQAAs0MkAAAAJAAAAJGdyT4AAAAAAAAAAPxsyT4AANjB AACzQwQAAAAjAAAAIAAAALo08EIAAAAAAAAAAH0utEIAANjBAACzQyQAAAAk AAAAujTwQgAAAAAAAAAAfS60QgAA2MEAALNDBAAAACMAAAAgAAAAkZ3JPgAA AAAAAAAA/GzJPgAA2MEAALNDJAAAACQAAACRnck+AAAAAAAAAAD8bMk+AADY wQAAs0MEAAAAIwAAACAAAACNkHQ8AAAAAAAAAABTvnQ8AADYwQAAs0MkAAAA JAAAAI2QdDwAAAAAAAAAAFO+dDwAANjBAACzQwQAAAAjAAAAIAAAAI2QdDwA AAAAAAAAAFO+dDwAANjBAACzQyQAAAAkAAAAjZB0PAAAAAAAAAAAU750PAAA 2MEAALNDBAAAACMAAAAgAAAAjZB0PAAAAAAAAAAAU750PAAA2MEAALNDJAAA ACQAAACNkHQ8AAAAAAAAAABTvnQ8AADYwQAAs0MEAAAAIQAAAAgAAAAUAAAA DAAAAA0AAAAlAAAADAAAAAUAAIAmAAAAHAAAAAQAAAAAAAAAAAAAAAAAAAAA AAAAJQAAAAwAAAAEAAAAJgAAABwAAAADAAAABgAAAEEAAAAAAAAAAAAAACUA AAAMAAAAAwAAACUAAAAMAAAABQAAgCcAAAAYAAAAAgAAAAAAAAD///8AAAAA ACUAAAAMAAAAAgAAABMAAAAMAAAAAQAAAFYAAACgAQAA5f///2YBAABdAAAA wAEAAGEAAACtBHMRJQWpEsAFxhNgBsUUIwehFfAHVRbMCNcWqAk2F5gKbRd+ C20XXwxKF08N+hYmDm4W7g66FbYP3hRqELoVRhFuFhMS+hb5EkoX7hNtF8UU bRemFTYXjBbXFm0XVRY1GKEV8xjFFJ0ZxhMuGqkSuhpzEVUbmxEEHIwRqRxf EU4d+xDkHWoQVx6xD8Ue0A4BH9sNOB/XDEwfyQs4H7EKAR+oCcUeuAhXHuYH 5B0oB04dnAapHCkGBBz3BVUb7QW6GiQGLhrkBJ0ZvQPzGLkCNRjsAW0XLgGM FrEAphVNAMUUIADuEyAA+RI+ABMSnQBGESQBahDOAbYPoALuDs4BJg4kAU8N nQBfDD4AfgsgAJgKIACoCU0AzAixAPAHLgEjB+wBYAa5AsAFvQMlBeQErQQk BgME7QVZA/cFtAIpBhQCnAaDASgHCwHmB50AuAhNAKgJJQCxCiAAyQslANcM TQDbDZ0A0A4LAbEPgwFqEBQC+xC0Al8RWQOMEQMEmxGtBHMRJQAAAAwAAAAF AACAFAAAAAwAAAANAAAAVwAAAKABAADl////ZgEAAF0AAADAAQAAYQAAAK0E cxElBakSwAXGE2AGxRQjB6EV8AdVFswI1xaoCTYXmAptF34LbRdfDEoXTw36 FiYObhbuDroVtg/eFGoQuhVGEW4WExL6FvkSShfuE20XxRRtF6YVNheMFtcW bRdVFjUYoRXzGMUUnRnGEy4aqRK6GnMRVRubEQQcjBGpHF8RTh37EOQdahBX HrEPxR7QDgEf2w04H9cMTB/JCzgfsQoBH6gJxR64CFce5gfkHSgHTh2cBqkc KQYEHPcFVRvtBboaJAYuGuQEnRm9A/MYuQI1GOwBbRcuAYwWsQCmFU0AxRQg AO4TIAD5Ej4AExKdAEYRJAFqEM4Btg+gAu4OzgEmDiQBTw2dAF8MPgB+CyAA mAogAKgJTQDMCLEA8AcuASMH7AFgBrkCwAW9AyUF5AStBCQGAwTtBVkD9wW0 AikGFAKcBoMBKAcLAeYHnQC4CE0AqAklALEKIADJCyUA1wxNANsNnQDQDgsB sQ+DAWoQFAL7ELQCXxFZA4wRAwSbEa0EcxEUAAAADAAAAA0AAAAmAAAAHAAA AAYAAAAGAAAAQQAAAAAAAAD/AAAAJQAAAAwAAAAGAAAAKAAAAAwAAAADAAAA VwAAACwAAAAEAAAAhQEAAD4AAACnAQAABAAAAGYP7BA2CCwINhcsCGYP7BAU AAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAAHgAAAAAAAAAAAAAAJQAAAAwA AAADAAAAKAAAAAwAAAAGAAAAJQAAAAwAAAAFAACAJwAAABgAAAAGAAAAAAAA AH9/fwAAAAAAJQAAAAwAAAAGAAAAEwAAAAwAAAABAAAAVgAAADAAAAD9//// iwEAAA8AAACNAQAABQAAAGoGNArxBrIJ1AqyCU0KNApqBjQKJQAAAAwAAAAF AACAKAAAAAwAAAAGAAAAFAAAAAwAAAANAAAAVwAAADAAAAD9////iwEAAA8A AACNAQAABQAAAGoGNArxBrIJ1AqyCU0KNApqBjQKFAAAAAwAAAANAAAAJQAA AAwAAAAFAACAJwAAABgAAAAGAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAGAAAA EwAAAAwAAAABAAAAVgAAADAAAAD9////jAEAAA0AAACQAQAABQAAAGoGLwpS Ci8KUgrtCmoG7QpqBi8KJQAAAAwAAAAFAACAKAAAAAwAAAAGAAAAFAAAAAwA AAANAAAAVwAAADAAAAD9////jAEAAA0AAACQAQAABQAAAGoGLwpSCi8KUgrt CmoG7QpqBi8KFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAGAAAA AAAAAKWlpQAAAAAAJQAAAAwAAAAGAAAAEwAAAAwAAAABAAAAVgAAADAAAAAM AAAAiwEAAA8AAACQAQAABQAAAFIK8grUCmYK1AqyCVIKNApSCvIKJQAAAAwA AAAFAACAKAAAAAwAAAAGAAAAFAAAAAwAAAANAAAAVwAAADAAAAAMAAAAiwEA AA8AAACQAQAABQAAAFIK8grUCmYK1AqyCVIKNApSCvIKFAAAAAwAAAANAAAA JQAAAAwAAAAFAACAJwAAABgAAAAGAAAAAAAAANjY2AAAAAAAJQAAAAwAAAAG AAAAEwAAAAwAAAABAAAAVgAAADAAAAD9////fwEAAA4AAACBAQAABQAAAG8G 8QbYBo0GuwqNBlIK8QZvBvEGJQAAAAwAAAAFAACAKAAAAAwAAAAGAAAAFAAA AAwAAAANAAAAVwAAADAAAAD9////fwEAAA4AAACBAQAABQAAAG8G8QbYBo0G uwqNBlIK8QZvBvEGFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAG AAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAGAAAAEwAAAAwAAAABAAAAVgAAADAA AAD9////gAEAAA0AAACNAQAABQAAAGoGDApSCgwKUgrsBmoG7AZqBgwKJQAA AAwAAAAFAACAKAAAAAwAAAAGAAAAFAAAAAwAAAANAAAAVwAAADAAAAD9//// gAEAAA0AAACNAQAABQAAAGoGDApSCgwKUgrsBmoG7AZqBgwKFAAAAAwAAAAN AAAAJQAAAAwAAAAFAACAJwAAABgAAAAGAAAAAAAAAKWlpQAAAAAAJQAAAAwA AAAGAAAAEwAAAAwAAAABAAAAVgAAADAAAAAMAAAAfwEAAA4AAACNAQAABQAA ALsKiAZNCvEGTQoMCrsKowm7CogGJQAAAAwAAAAFAACAKAAAAAwAAAAGAAAA FAAAAAwAAAANAAAAVwAAADAAAAAMAAAAfwEAAA4AAACNAQAABQAAALsKiAZN CvEGTQoMCrsKowm7CogGFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgA AAAGAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAGAAAAEwAAAAwAAAABAAAAVgAA ADAAAAD5////kAEAAA0AAACUAQAABQAAAHoF2AsaBhALXAoQC7wJ2At6BdgL JQAAAAwAAAAFAACAKAAAAAwAAAAGAAAAFAAAAAwAAAANAAAAVwAAADAAAAD5 ////kAEAAA0AAACUAQAABQAAAHoF2AsaBhALXAoQC7wJ2At6BdgLFAAAAAwA AAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAGAAAAAAAAAL+/vwAAAAAAJQAA AAwAAAAGAAAAEwAAAAwAAAABAAAAVgAAADAAAAD5////kAEAAA0AAACUAQAA BQAAAHoF2AsaBhALXAoQC7wJ2At6BdgLJQAAAAwAAAAFAACAKAAAAAwAAAAG AAAAFAAAAAwAAAANAAAAVwAAADAAAAD5////kAEAAA0AAACUAQAABQAAAHoF 2AsaBhALXAoQC7wJ2At6BdgLFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAA ABgAAAAGAAAAAAAAAKWlpQAAAAAAJQAAAAwAAAAGAAAAEwAAAAwAAAABAAAA VgAAADAAAAAKAAAAkAEAAA0AAACUAQAABQAAAMEJAAxXClYLVwoVC8EJ0wvB CQAMJQAAAAwAAAAFAACAKAAAAAwAAAAGAAAAFAAAAAwAAAANAAAAVwAAADAA AAAKAAAAkAEAAA0AAACUAQAABQAAAMEJAAxXClYLVwoVC8EJ0wvBCQAMFAAA AAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAGAAAAAAAAAKWlpQAAAAAA JQAAAAwAAAAGAAAAEwAAAAwAAAABAAAAVgAAADAAAAAKAAAAkAEAAA0AAACU AQAABQAAAMEJAAxXClYLVwoVC8EJ0wvBCQAMJQAAAAwAAAAFAACAKAAAAAwA AAAGAAAAFAAAAAwAAAANAAAAVwAAADAAAAAKAAAAkAEAAA0AAACUAQAABQAA AMEJAAxXClYLVwoVC8EJ0wvBCQAMFAAAAAwAAAANAAAAJQAAAAwAAAAFAACA JwAAABgAAAAGAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAGAAAAEwAAAAwAAAAB AAAAVgAAADAAAAD6////kwEAAAsAAACUAQAABQAAAH8F2AvBCdgLwQkADH8F AAx/BdgLJQAAAAwAAAAFAACAKAAAAAwAAAAGAAAAFAAAAAwAAAANAAAAVwAA ADAAAAD6////kwEAAAsAAACUAQAABQAAAH8F2AvBCdgLwQkADH8FAAx/BdgL FAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAGAAAAAAAAAL+/vwAA AAAAJQAAAAwAAAAGAAAAEwAAAAwAAAABAAAAVgAAADAAAAD6////kwEAAAsA AACUAQAABQAAAH8F2AvBCdgLwQkADH8FAAx/BdgLJQAAAAwAAAAFAACAKAAA AAwAAAAGAAAAFAAAAAwAAAANAAAAVwAAADAAAAD6////kwEAAAsAAACUAQAA BQAAAH8F2AvBCdgLwQkADH8FAAx/BdgLFAAAAAwAAAANAAAAJQAAAAwAAAAF AACAJwAAABgAAAAGAAAAAAAAAAAA/wAAAAAAJQAAAAwAAAAGAAAAEwAAAAwA AAABAAAAVgAAADAAAAD+////gQEAAAsAAACLAQAABQAAAPMJUAfJBlAHyQao CfMJqAnzCVAHJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAADAAAAD+//// gQEAAAsAAACLAQAABQAAAPMJUAfJBlAHyQaoCfMJqAnzCVAHFAAAAAwAAAAN AAAAGwAAABAAAAAcCQAAmAoAADYAAAAQAAAA5AkAAJgKAAAUAAAADAAAAA0A AAAmAAAAHAAAAAcAAAAGAAAAQQAAAAAAAAAAAAAAJQAAAAwAAAAHAAAAKAAA AAwAAAADAAAAJQAAAAwAAAAFAACAJQAAAAwAAAACAAAAEwAAAAwAAAABAAAA VgAAAKABAAAAAAAAggEAAAoAAACLAQAAYQAAAG4H9AhzBw0JhwcmCZYHPwmg B04JtAdiCcMHZwnXB3EJ6wdxCf8HcQkTCHEJIghnCTYIYglFCE4JWQg/CWgI Tgl8CGIJiwhnCZoIcQmzCHEJxwhxCdEIcQnqCGcJ/ghiCQ0JTgkcCT8JKwkm CToJDQlECfQITgn0CF0J9AhsCfQIdgnqCIUJ4AiUCcwImQm9CJ4JqQijCZAI owl8CKMJaAieCU8ImQk7CJQJLAiFCRgIdgkOCGwJBAhdCQQITgkECEQJBAg6 CesHKwnSBxwJuQcNCaoH/giWB+oIkQfRCIcHxwiHB7MIhweaCIcHiwiRB3wI lgdoCKoHWQi5B0UIqgc2CJYHIgiRBxMIhwf/B4cH6weHB9cHhwfDB5EHtAeW B6AHqgeWB7kHhwfSB3MH6wduBwQIXwcECFUHBAhGBwQIPAcOCC0HGAgeBywI GQc7CA8HTwgPB2gIDwd8CA8HkAgPB6kIGQe9CB4HzAgtB+AIPAfqCEYH9AhV B/QIXwf0CG4H9AglAAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAoAEAAAAA AACCAQAACgAAAIsBAABhAAAAbgf0CHMHDQmHByYJlgc/CaAHTgm0B2IJwwdn CdcHcQnrB3EJ/wdxCRMIcQkiCGcJNghiCUUITglZCD8JaAhOCXwIYgmLCGcJ mghxCbMIcQnHCHEJ0QhxCeoIZwn+CGIJDQlOCRwJPwkrCSYJOgkNCUQJ9AhO CfQIXQn0CGwJ9Ah2CeoIhQngCJQJzAiZCb0IngmpCKMJkAijCXwIowloCJ4J TwiZCTsIlAksCIUJGAh2CQ4IbAkECF0JBAhOCQQIRAkECDoJ6wcrCdIHHAm5 Bw0Jqgf+CJYH6giRB9EIhwfHCIcHswiHB5oIhweLCJEHfAiWB2gIqgdZCLkH RQiqBzYIlgciCJEHEwiHB/8HhwfrB4cH1weHB8MHkQe0B5YHoAeqB5YHuQeH B9IHcwfrB24HBAhfBwQIVQcECEYHBAg8Bw4ILQcYCB4HLAgZBzsIDwdPCA8H aAgPB3wIDweQCA8HqQgZB70IHgfMCC0H4Ag8B+oIRgf0CFUH9AhfB/QIbgf0 CBQAAAAMAAAADQAAACYAAAAcAAAAAwAAAAYAAAAeAAAAAAAAAAAAAAAlAAAA DAAAAAMAAAAoAAAADAAAAAcAAAAlAAAADAAAAAUAAIAnAAAAGAAAAAcAAAAA AAAAf39/AAAAAAAlAAAADAAAAAcAAAATAAAADAAAAAEAAABWAAAAMAAAADQA AACLAQAARgAAAI0BAAAFAAAAyhQ0ClEVsgk0GbIJrRg0CsoUNAolAAAADAAA AAUAAIAoAAAADAAAAAcAAAAUAAAADAAAAA0AAABXAAAAMAAAADQAAACLAQAA RgAAAI0BAAAFAAAAyhQ0ClEVsgk0GbIJrRg0CsoUNAoUAAAADAAAAA0AAAAl AAAADAAAAAUAAIAnAAAAGAAAAAcAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAcA AAATAAAADAAAAAEAAABWAAAAMAAAADQAAACMAQAARAAAAJABAAAFAAAAyhQv CrIYLwqyGO0KyhTtCsoULwolAAAADAAAAAUAAIAoAAAADAAAAAcAAAAUAAAA DAAAAA0AAABXAAAAMAAAADQAAACMAQAARAAAAJABAAAFAAAAyhQvCrIYLwqy GO0KyhTtCsoULwoUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAcA AAAAAAAApaWlAAAAAAAlAAAADAAAAAcAAAATAAAADAAAAAEAAABWAAAAMAAA AEMAAACLAQAARgAAAJABAAAFAAAAshjyCjQZZgo0GbIJshg0CrIY8golAAAA DAAAAAUAAIAoAAAADAAAAAcAAAAUAAAADAAAAA0AAABXAAAAMAAAAEMAAACL AQAARgAAAJABAAAFAAAAshjyCjQZZgo0GbIJshg0CrIY8goUAAAADAAAAA0A AAAlAAAADAAAAAUAAIAnAAAAGAAAAAcAAAAAAAAA2NjYAAAAAAAlAAAADAAA AAcAAAATAAAADAAAAAEAAABWAAAAMAAAADQAAAB/AQAARQAAAIEBAAAFAAAA zxTxBjgVjQYbGY0GshjxBs8U8QYlAAAADAAAAAUAAIAoAAAADAAAAAcAAAAU AAAADAAAAA0AAABXAAAAMAAAADQAAAB/AQAARQAAAIEBAAAFAAAAzxTxBjgV jQYbGY0GshjxBs8U8QYUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAA AAcAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAcAAAATAAAADAAAAAEAAABWAAAA MAAAADQAAACAAQAARAAAAI0BAAAFAAAAyhQMCrIYDAqyGOwGyhTsBsoUDAol AAAADAAAAAUAAIAoAAAADAAAAAcAAAAUAAAADAAAAA0AAABXAAAAMAAAADQA AACAAQAARAAAAI0BAAAFAAAAyhQMCrIYDAqyGOwGyhTsBsoUDAoUAAAADAAA AA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAcAAAAAAAAApaWlAAAAAAAlAAAA DAAAAAcAAAATAAAADAAAAAEAAABWAAAAMAAAAEMAAAB/AQAARQAAAI0BAAAF AAAAGxmIBq0Y8QatGAwKGxmjCRsZiAYlAAAADAAAAAUAAIAoAAAADAAAAAcA AAAUAAAADAAAAA0AAABXAAAAMAAAAEMAAAB/AQAARQAAAI0BAAAFAAAAGxmI Bq0Y8QatGAwKGxmjCRsZiAYUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAA GAAAAAcAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAcAAAATAAAADAAAAAEAAABW AAAAMAAAADAAAACQAQAARAAAAJQBAAAFAAAA2hPYC3oUEAu8GBALHBjYC9oT 2AslAAAADAAAAAUAAIAoAAAADAAAAAcAAAAUAAAADAAAAA0AAABXAAAAMAAA ADAAAACQAQAARAAAAJQBAAAFAAAA2hPYC3oUEAu8GBALHBjYC9oT2AsUAAAA DAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAcAAAAAAAAAv7+/AAAAAAAl AAAADAAAAAcAAAATAAAADAAAAAEAAABWAAAAMAAAADAAAACQAQAARAAAAJQB AAAFAAAA2hPYC3oUEAu8GBALHBjYC9oT2AslAAAADAAAAAUAAIAoAAAADAAA AAcAAAAUAAAADAAAAA0AAABXAAAAMAAAADAAAACQAQAARAAAAJQBAAAFAAAA 2hPYC3oUEAu8GBALHBjYC9oT2AsUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAn AAAAGAAAAAcAAAAAAAAApaWlAAAAAAAlAAAADAAAAAcAAAATAAAADAAAAAEA AABWAAAAMAAAAEEAAACQAQAARAAAAJQBAAAFAAAAIRgADLcYVgu3GBULIRjT CyEYAAwlAAAADAAAAAUAAIAoAAAADAAAAAcAAAAUAAAADAAAAA0AAABXAAAA MAAAAEEAAACQAQAARAAAAJQBAAAFAAAAIRgADLcYVgu3GBULIRjTCyEYAAwU AAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAcAAAAAAAAApaWlAAAA AAAlAAAADAAAAAcAAAATAAAADAAAAAEAAABWAAAAMAAAAEEAAACQAQAARAAA AJQBAAAFAAAAIRgADLcYVgu3GBULIRjTCyEYAAwlAAAADAAAAAUAAIAoAAAA DAAAAAcAAAAUAAAADAAAAA0AAABXAAAAMAAAAEEAAACQAQAARAAAAJQBAAAF AAAAIRgADLcYVgu3GBULIRjTCyEYAAwUAAAADAAAAA0AAAAlAAAADAAAAAUA AIAnAAAAGAAAAAcAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAcAAAATAAAADAAA AAEAAABWAAAAMAAAADAAAACTAQAAQgAAAJQBAAAFAAAA3xPYCyEY2AshGAAM 3xMADN8T2AslAAAADAAAAAUAAIAoAAAADAAAAAcAAAAUAAAADAAAAA0AAABX AAAAMAAAADAAAACTAQAAQgAAAJQBAAAFAAAA3xPYCyEY2AshGAAM3xMADN8T 2AsUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAcAAAAAAAAAv7+/ AAAAAAAlAAAADAAAAAcAAAATAAAADAAAAAEAAABWAAAAMAAAADAAAACTAQAA QgAAAJQBAAAFAAAA3xPYCyEY2AshGAAM3xMADN8T2AslAAAADAAAAAUAAIAo AAAADAAAAAcAAAAUAAAADAAAAA0AAABXAAAAMAAAADAAAACTAQAAQgAAAJQB AAAFAAAA3xPYCyEY2AshGAAM3xMADN8T2AsUAAAADAAAAA0AAAAlAAAADAAA AAUAAIAlAAAADAAAAAYAAAATAAAADAAAAAEAAABWAAAAMAAAADUAAACBAQAA QgAAAIsBAAAFAAAAUxhQBykVUAcpFagJUxioCVMYUAclAAAADAAAAAUAAIAU AAAADAAAAA0AAABXAAAAMAAAADUAAACBAQAAQgAAAIsBAAAFAAAAUxhQBykV UAcpFagJUxioCVMYUAcUAAAADAAAAA0AAAAbAAAAEAAAAHwXAACYCgAANgAA ABAAAABEGAAAmAoAABQAAAAMAAAADQAAACYAAAAcAAAABwAAAAYAAABBAAAA AAAAAAAAAAAlAAAADAAAAAcAAAAoAAAADAAAAAMAAAAlAAAADAAAAAUAAIAl AAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAoAEAADYAAACCAQAAQQAAAIsB AABhAAAAzhX0CNMVDQnnFSYJ9hU/CQAWTgkUFmIJIxZnCTcWcQlLFnEJXxZx CXMWcQmCFmcJlhZiCaUWTgm5Fj8JyBZOCdwWYgnrFmcJ+hZxCRMXcQknF3EJ MRdxCUoXZwleF2IJbRdOCXwXPwmLFyYJmhcNCaQX9AiuF/QIvRf0CMwX9AjW F+oI5RfgCPQXzAj5F70I/hepCAMYkAgDGHwIAxhoCP4XTwj5FzsI9BcsCOUX GAjWFw4IzBcECL0XBAiuFwQIpBcECJoX6weLF9IHfBe5B20XqgdeF5YHSheR BzEXhwcnF4cHExeHB/oWhwfrFpEH3BaWB8gWqge5FrkHpRaqB5YWlgeCFpEH cxaHB18WhwdLFocHNxaHByMWkQcUFpYHABaqB/YVuQfnFdIH0xXrB84VBAi/ FQQItRUECKYVBAicFQ4IjRUYCH4VLAh5FTsIbxVPCG8VaAhvFXwIbxWQCG8V qQh5Fb0IfhXMCI0V4AicFeoIphX0CLUV9Ai/FfQIzhX0CCUAAAAMAAAABQAA gBQAAAAMAAAADQAAAFcAAACgAQAANgAAAIIBAABBAAAAiwEAAGEAAADOFfQI 0xUNCecVJgn2FT8JABZOCRQWYgkjFmcJNxZxCUsWcQlfFnEJcxZxCYIWZwmW FmIJpRZOCbkWPwnIFk4J3BZiCesWZwn6FnEJExdxCScXcQkxF3EJShdnCV4X YgltF04JfBc/CYsXJgmaFw0JpBf0CK4X9Ai9F/QIzBf0CNYX6gjlF+AI9BfM CPkXvQj+F6kIAxiQCAMYfAgDGGgI/hdPCPkXOwj0FywI5RcYCNYXDgjMFwQI vRcECK4XBAikFwQImhfrB4sX0gd8F7kHbReqB14XlgdKF5EHMReHBycXhwcT F4cH+haHB+sWkQfcFpYHyBaqB7kWuQelFqoHlhaWB4IWkQdzFocHXxaHB0sW hwc3FocHIxaRBxQWlgcAFqoH9hW5B+cV0gfTFesHzhUECL8VBAi1FQQIphUE CJwVDgiNFRgIfhUsCHkVOwhvFU8IbxVoCG8VfAhvFZAIbxWpCHkVvQh+FcwI jRXgCJwV6gimFfQItRX0CL8V9AjOFfQIFAAAAAwAAAANAAAAJgAAABwAAAAD AAAABgAAAB4AAAAAAAAAAAAAACUAAAAMAAAAAwAAACgAAAAMAAAABwAAACUA AAAMAAAABQAAgCcAAAAYAAAABwAAAAAAAAB/f38AAAAAACUAAAAMAAAABwAA ABMAAAAMAAAAAQAAAFYAAAAwAAAAFwAAAKcBAAApAAAAqgEAAAUAAABKDbQR 0Q0yEbQRMhEtEbQRSg20ESUAAAAMAAAABQAAgCgAAAAMAAAABwAAABQAAAAM AAAADQAAAFcAAAAwAAAAFwAAAKcBAAApAAAAqgEAAAUAAABKDbQR0Q0yEbQR MhEtEbQRSg20ERQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABwAA AAAAAAC/v78AAAAAACUAAAAMAAAABwAAABMAAAAMAAAAAQAAAFYAAAAwAAAA FwAAAKkBAAAnAAAArQEAAAUAAABKDa8RMhGvETIRbRJKDW0SSg2vESUAAAAM AAAABQAAgCgAAAAMAAAABwAAABQAAAAMAAAADQAAAFcAAAAwAAAAFwAAAKkB AAAnAAAArQEAAAUAAABKDa8RMhGvETIRbRJKDW0SSg2vERQAAAAMAAAADQAA ACUAAAAMAAAABQAAgCcAAAAYAAAABwAAAAAAAAClpaUAAAAAACUAAAAMAAAA BwAAABMAAAAMAAAAAQAAAFYAAAAwAAAAJgAAAKcBAAApAAAArQEAAAUAAAAy EXIStBHmEbQRMhEyEbQRMhFyEiUAAAAMAAAABQAAgCgAAAAMAAAABwAAABQA AAAMAAAADQAAAFcAAAAwAAAAJgAAAKcBAAApAAAArQEAAAUAAAAyEXIStBHm EbQRMhEyEbQRMhFyEhQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAA BwAAAAAAAADY2NgAAAAAACUAAAAMAAAABwAAABMAAAAMAAAAAQAAAFYAAAAw AAAAFwAAAJsBAAApAAAAngEAAAUAAABPDXEOuA0NDpsRDQ4yEXEOTw1xDiUA AAAMAAAABQAAgCgAAAAMAAAABwAAABQAAAAMAAAADQAAAFcAAAAwAAAAFwAA AJsBAAApAAAAngEAAAUAAABPDXEOuA0NDpsRDQ4yEXEOTw1xDhQAAAAMAAAA DQAAACUAAAAMAAAABQAAgCcAAAAYAAAABwAAAAAAAAC/v78AAAAAACUAAAAM AAAABwAAABMAAAAMAAAAAQAAAFYAAAAwAAAAFwAAAJ0BAAAnAAAAqgEAAAUA AABKDYwRMhGMETIRbA5KDWwOSg2MESUAAAAMAAAABQAAgCgAAAAMAAAABwAA ABQAAAAMAAAADQAAAFcAAAAwAAAAFwAAAJ0BAAAnAAAAqgEAAAUAAABKDYwR MhGMETIRbA5KDWwOSg2MERQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAY AAAABwAAAAAAAAClpaUAAAAAACUAAAAMAAAABwAAABMAAAAMAAAAAQAAAFYA AAAwAAAAJgAAAJsBAAApAAAAqgEAAAUAAACbEQgOLRFxDi0RjBGbESMRmxEI DiUAAAAMAAAABQAAgCgAAAAMAAAABwAAABQAAAAMAAAADQAAAFcAAAAwAAAA JgAAAJsBAAApAAAAqgEAAAUAAACbEQgOLRFxDi0RjBGbESMRmxEIDhQAAAAM AAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABwAAAAAAAAC/v78AAAAAACUA AAAMAAAABwAAABMAAAAMAAAAAQAAAFYAAAAwAAAAFAAAAK0BAAAnAAAAsAEA AAUAAABaDFgT+gyQEjwRkBKcEFgTWgxYEyUAAAAMAAAABQAAgCgAAAAMAAAA BwAAABQAAAAMAAAADQAAAFcAAAAwAAAAFAAAAK0BAAAnAAAAsAEAAAUAAABa DFgT+gyQEjwRkBKcEFgTWgxYExQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcA AAAYAAAABwAAAAAAAAC/v78AAAAAACUAAAAMAAAABwAAABMAAAAMAAAAAQAA AFYAAAAwAAAAFAAAAK0BAAAnAAAAsAEAAAUAAABaDFgT+gyQEjwRkBKcEFgT WgxYEyUAAAAMAAAABQAAgCgAAAAMAAAABwAAABQAAAAMAAAADQAAAFcAAAAw AAAAFAAAAK0BAAAnAAAAsAEAAAUAAABaDFgT+gyQEjwRkBKcEFgTWgxYExQA AAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABwAAAAAAAAClpaUAAAAA ACUAAAAMAAAABwAAABMAAAAMAAAAAQAAAFYAAAAwAAAAJAAAAK0BAAAnAAAA sQEAAAUAAAChEIATNxHWEjcRlRKhEFMToRCAEyUAAAAMAAAABQAAgCgAAAAM AAAABwAAABQAAAAMAAAADQAAAFcAAAAwAAAAJAAAAK0BAAAnAAAAsQEAAAUA AAChEIATNxHWEjcRlRKhEFMToRCAExQAAAAMAAAADQAAACUAAAAMAAAABQAA gCcAAAAYAAAABwAAAAAAAAClpaUAAAAAACUAAAAMAAAABwAAABMAAAAMAAAA AQAAAFYAAAAwAAAAJAAAAK0BAAAnAAAAsQEAAAUAAAChEIATNxHWEjcRlRKh EFMToRCAEyUAAAAMAAAABQAAgCgAAAAMAAAABwAAABQAAAAMAAAADQAAAFcA AAAwAAAAJAAAAK0BAAAnAAAAsQEAAAUAAAChEIATNxHWEjcRlRKhEFMToRCA ExQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABwAAAAAAAAC/v78A AAAAACUAAAAMAAAABwAAABMAAAAMAAAAAQAAAFYAAAAwAAAAFAAAALABAAAl AAAAsQEAAAUAAABfDFgToRBYE6EQgBNfDIATXwxYEyUAAAAMAAAABQAAgCgA AAAMAAAABwAAABQAAAAMAAAADQAAAFcAAAAwAAAAFAAAALABAAAlAAAAsQEA AAUAAABfDFgToRBYE6EQgBNfDIATXwxYExQAAAAMAAAADQAAACUAAAAMAAAA BQAAgCcAAAAYAAAABwAAAAAAAAC/v78AAAAAACUAAAAMAAAABwAAABMAAAAM AAAAAQAAAFYAAAAwAAAAFAAAALABAAAlAAAAsQEAAAUAAABfDFgToRBYE6EQ gBNfDIATXwxYEyUAAAAMAAAABQAAgCgAAAAMAAAABwAAABQAAAAMAAAADQAA AFcAAAAwAAAAFAAAALABAAAlAAAAsQEAAAUAAABfDFgToRBYE6EQgBNfDIAT XwxYExQAAAAMAAAADQAAACUAAAAMAAAABQAAgCUAAAAMAAAABgAAABMAAAAM AAAAAQAAAFYAAAAwAAAAGQAAAJ4BAAAmAAAAqAEAAAUAAADTENAOqQ3QDqkN KBHTECgR0xDQDiUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAAwAAAAGQAA AJ4BAAAmAAAAqAEAAAUAAADTENAOqQ3QDqkNKBHTECgR0xDQDhQAAAAMAAAA DQAAABsAAAAQAAAA/A8AABgSAAA2AAAAEAAAAMQQAAAYEgAAFAAAAAwAAAAN AAAAJgAAABwAAAAHAAAABgAAAEEAAAAAAAAAAAAAACUAAAAMAAAABwAAACgA AAAMAAAAAwAAACUAAAAMAAAABQAAgCUAAAAMAAAAAgAAABMAAAAMAAAAAQAA AFYAAACgAQAAGgAAAJ8BAAAlAAAApwEAAGEAAABODnQQUw6NEGcOphB2Dr8Q gA7OEJQO4hCjDucQtw7xEMsO8RDfDvEQ8w7xEAIP5xAWD+IQJQ/OEDkPvxBI D84QXA/iEGsP5xB6D/EQkw/xEKcP8RCxD/EQyg/nEN4P4hDtD84Q/A+/EAsQ phAaEI0QJBB0EC4QdBA9EHQQTBB0EFYQahBlEGAQdBBMEHkQPRB+ECkQgxAQ EIMQ/A+DEOgPfhDPD3kQuw90EKwPZRCYD1YQjg9MEIQPPRCEDy4QhA8kEIQP GhBrDwsQUg/8DzkP7Q8qD94PFg/KDxEPsQ8HD6cPBw+TDwcPeg8HD2sPEQ9c DxYPSA8qDzkPOQ8lDyoPFg8WDwIPEQ/zDgcP3w4HD8sOBw+3DgcPow4RD5QO Fg+ADioPdg45D2cOUg9TDmsPTg6EDz8OhA81DoQPJg6EDxwOjg8NDpgP/g2s D/kNuw/vDc8P7w3oD+8N/A/vDRAQ7w0pEPkNPRD+DUwQDQ5gEBwOahAmDnQQ NQ50ED8OdBBODnQQJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAAKABAAAa AAAAnwEAACUAAACnAQAAYQAAAE4OdBBTDo0QZw6mEHYOvxCADs4QlA7iEKMO 5xC3DvEQyw7xEN8O8RDzDvEQAg/nEBYP4hAlD84QOQ+/EEgPzhBcD+IQaw/n EHoP8RCTD/EQpw/xELEP8RDKD+cQ3g/iEO0PzhD8D78QCxCmEBoQjRAkEHQQ LhB0ED0QdBBMEHQQVhBqEGUQYBB0EEwQeRA9EH4QKRCDEBAQgxD8D4MQ6A9+ EM8PeRC7D3QQrA9lEJgPVhCOD0wQhA89EIQPLhCEDyQQhA8aEGsPCxBSD/wP OQ/tDyoP3g8WD8oPEQ+xDwcPpw8HD5MPBw96DwcPaw8RD1wPFg9IDyoPOQ85 DyUPKg8WDxYPAg8RD/MOBw/fDgcPyw4HD7cOBw+jDhEPlA4WD4AOKg92DjkP Zw5SD1MOaw9ODoQPPw6EDzUOhA8mDoQPHA6ODw0OmA/+DawP+Q27D+8Nzw/v DegP7w38D+8NEBDvDSkQ+Q09EP4NTBANDmAQHA5qECYOdBA1DnQQPw50EE4O dBAUAAAADAAAAA0AAAAiAAAADAAAAP////8oAAAADAAAAAQAAAAlAAAADAAA AAcAAIAlAAAADAAAAAAAAIBLAAAAEAAAAAAAAAAFAAAAKAAAAAwAAAACAAAA KAAAAAwAAAAGAAAAKAAAAAwAAAAHAAAARgAAABQAAAAIAAAAR0RJQwMAAAAi AAAADAAAAP////8iAAAADAAAAP////8lAAAADAAAAA0AAIAoAAAADAAAAAUA AAAiAAAADAAAAP////9GAAAAMFQAACRUAABFTUYrJkAAABAAAAAEAAAAAAAA ACVAAAAQAAAABAAAAAAAAAAsQAAAJAAAABgAAAD9sPc9AAAAgAAAAICN8bk9 cvIwRG33skMIQAcFsFMAAKRTAAABEMDbAgAAAAMAAACUUwAAAQAAAGwAAAD+ /////v///zMBAADnAAAAAAAAAAAAAABvHwAAkBcAACBFTUYAAAEAlFMAANsB AAAGAAAAAAAAAAAAAAAAAAAAAAUAAAAEAABSAQAADgEAAAAAAAAAAAAAAAAA AFAoBQCwHgQARgAAALAiAACiIgAAR0RJQwEAAIAAAwAAuY1p3AAAAACKIgAA AQAJAAADRREAAAYAxgAAAAAABQAAAAwCjxduHwUAAAAMAo8Xbh8HAAAAEAQB AAEAAQABAAMAAAAeAAQAAAAEAQ0ABwAAAPwCAQAAAAAAAAAEAAAALQEAAAkA AAD6AgAAAAAAAAAAAAAiAAQAAAAtAQEACQAAAPoCBgBBAAAAAAAAACIABAAA AC0BAgAEAAAALQEAAAcAAAD8AgAA////AAAABAAAAC0BAwAEAAAABgEBAMYA AAAkA2EArQRzESUFqRLABcYTYAbFFCMHoRXwB1UWzAjXFqgJNheYCm0Xfgtt F18MShdPDfoWJg5uFu4OuhW2D94UahC6FUYRbhYTEvoW+RJKF+4TbRfFFG0X phU2F4wW1xZtF1UWNRihFfMYxRSdGcYTLhqpEroacxFVG5sRBByMEakcXxFO HfsQ5B1qEFcesQ/FHtAOAR/bDTgf1wxMH8kLOB+xCgEfqAnFHrgIVx7mB+Qd KAdOHZwGqRwpBgQc9wVVG+0FuhokBi4a5ASdGb0D8xi5AjUY7AFtFy4BjBax AKYVTQDFFCAA7hMgAPkSPgATEp0ARhEkAWoQzgG2D6AC7g7OASYOJAFPDZ0A Xww+AH4LIACYCiAAqAlNAMwIsQDwBy4BIwfsAWAGuQLABb0DJQXkBK0EJAYD BO0FWQP3BbQCKQYUApwGgwEoBwsB5gedALgITQCoCSUAsQogAMkLJQDXDE0A 2w2dANAOCwGxD4MBahAUAvsQtAJfEVkDjBEDBJsRrQRzEQQAAAAtAQAABAAA AAQBDQDGAAAAJQNhAK0EcxElBakSwAXGE2AGxRQjB6EV8AdVFswI1xaoCTYX mAptF34LbRdfDEoXTw36FiYObhbuDroVtg/eFGoQuhVGEW4WExL6FvkSShfu E20XxRRtF6YVNheMFtcWbRdVFjUYoRXzGMUUnRnGEy4aqRK6GnMRVRubEQQc jBGpHF8RTh37EOQdahBXHrEPxR7QDgEf2w04H9cMTB/JCzgfsQoBH6gJxR64 CFce5gfkHSgHTh2cBqkcKQYEHPcFVRvtBboaJAYuGuQEnRm9A/MYuQI1GOwB bRcuAYwWsQCmFU0AxRQgAO4TIAD5Ej4AExKdAEYRJAFqEM4Btg+gAu4OzgEm DiQBTw2dAF8MPgB+CyAAmAogAKgJTQDMCLEA8AcuASMH7AFgBrkCwAW9AyUF 5AStBCQGAwTtBVkD9wW0AikGFAKcBoMBKAcLAeYHnQC4CE0AqAklALEKIADJ CyUA1wxNANsNnQDQDgsBsQ+DAWoQFAL7ELQCXxFZA4wRAwSbEa0EcxEEAAAA BAENAAkAAAD6AgYAQQAAAP8AAAAiAAQAAAAtAQQABAAAAPABAgAMAAAAJQME AGYP7BA2CCwINhcsCGYP7BAEAAAABAENAAkAAAD6AgYAHgAAAAAAAAAiAAQA AAAtAQIABAAAAPABBAAEAAAALQEAAAcAAAD8AgAAf39/AAAABAAAAC0BBAAE AAAABgEBAA4AAAAkAwUAagY0CvEGsgnUCrIJTQo0CmoGNAoEAAAALQEAAAQA AADwAQQABAAAAAQBDQAOAAAAJQMFAGoGNArxBrIJ1AqyCU0KNApqBjQKBAAA AAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4A AAAkAwUAagYvClIKLwpSCu0KagbtCmoGLwoEAAAALQEAAAQAAADwAQQABAAA AAQBDQAOAAAAJQMFAGoGLwpSCi8KUgrtCmoG7QpqBi8KBAAAAAQBDQAEAAAA LQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAUgry CtQKZgrUCrIJUgo0ClIK8goEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAA JQMFAFIK8grUCmYK1AqyCVIKNApSCvIKBAAAAAQBDQAEAAAALQEAAAcAAAD8 AgAA2NjYAAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAbwbxBtgGjQa7Co0G UgrxBm8G8QYEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAG8G8QbY Bo0GuwqNBlIK8QZvBvEGBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAA BAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAagYMClIKDApSCuwGagbsBmoGDAoE AAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAGoGDApSCgwKUgrsBmoG 7AZqBgwKBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBAAE AAAABgEBAA4AAAAkAwUAuwqIBk0K8QZNCgwKuwqjCbsKiAYEAAAALQEAAAQA AADwAQQABAAAAAQBDQAOAAAAJQMFALsKiAZNCvEGTQoMCrsKowm7CogGBAAA AAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4A AAAkAwUAegXYCxoGEAtcChALvAnYC3oF2AsEAAAALQEAAAQAAADwAQQABAAA AAQBDQAOAAAAJQMFAHoF2AsaBhALXAoQC7wJ2At6BdgLBAAAAAQBDQAEAAAA LQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAegXY CxoGEAtcChALvAnYC3oF2AsEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAA JQMFAHoF2AsaBhALXAoQC7wJ2At6BdgLBAAAAAQBDQAEAAAALQEAAAcAAAD8 AgAApaWlAAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAwQkADFcKVgtXChUL wQnTC8EJAAwEAAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAMEJAAxX ClYLVwoVC8EJ0wvBCQAMBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAA BAAAAC0BBAAEAAAABgEBAA4AAAAkAwUAwQkADFcKVgtXChULwQnTC8EJAAwE AAAALQEAAAQAAADwAQQABAAAAAQBDQAOAAAAJQMFAMEJAAxXClYLVwoVC8EJ 0wvBCQAMBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAE AAAABgEBAA4AAAAkAwUAfwXYC8EJ2AvBCQAMfwUADH8F2AsEAAAALQEAAAQA AADwAQQABAAAAAQBDQAOAAAAJQMFAH8F2AvBCdgLwQkADH8FAAx/BdgLBAAA AAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBAAEAAAABgEBAA4A AAAkAwUAfwXYC8EJ2AvBCQAMfwUADH8F2AsEAAAALQEAAAQAAADwAQQABAAA AAQBDQAOAAAAJQMFAH8F2AvBCdgLwQkADH8FAAx/BdgLBAAAAAQBDQAEAAAA LQEAAAcAAAD8AgAAAAD/AAAABAAAAC0BBAAEAAAABgEBAA4AAAAkAwUA8wlQ B8kGUAfJBqgJ8wmoCfMJUAcEAAAALQEAAAQAAAAEAQ0ADgAAACUDBQDzCVAH yQZQB8kGqAnzCagJ8wlQBwQAAAAEAQ0ABQAAABQCmAocCQUAAAATApgK5AkE AAAABAENAAkAAAD6AgYAQQAAAAAAAAAiAAQAAAAtAQUABAAAAPABAgAEAAAA LQEAAAQAAAAtAQMABAAAAAYBAQDGAAAAJANhAG4H9AhzBw0JhwcmCZYHPwmg B04JtAdiCcMHZwnXB3EJ6wdxCf8HcQkTCHEJIghnCTYIYglFCE4JWQg/CWgI Tgl8CGIJiwhnCZoIcQmzCHEJxwhxCdEIcQnqCGcJ/ghiCQ0JTgkcCT8JKwkm CToJDQlECfQITgn0CF0J9AhsCfQIdgnqCIUJ4AiUCcwImQm9CJ4JqQijCZAI owl8CKMJaAieCU8ImQk7CJQJLAiFCRgIdgkOCGwJBAhdCQQITgkECEQJBAg6 CesHKwnSBxwJuQcNCaoH/giWB+oIkQfRCIcHxwiHB7MIhweaCIcHiwiRB3wI lgdoCKoHWQi5B0UIqgc2CJYHIgiRBxMIhwf/B4cH6weHB9cHhwfDB5EHtAeW B6AHqgeWB7kHhwfSB3MH6wduBwQIXwcECFUHBAhGBwQIPAcOCC0HGAgeBywI GQc7CA8HTwgPB2gIDwd8CA8HkAgPB6kIGQe9CB4HzAgtB+AIPAfqCEYH9AhV B/QIXwf0CG4H9AgEAAAALQEAAAQAAAAEAQ0AxgAAACUDYQBuB/QIcwcNCYcH JgmWBz8JoAdOCbQHYgnDB2cJ1wdxCesHcQn/B3EJEwhxCSIIZwk2CGIJRQhO CVkIPwloCE4JfAhiCYsIZwmaCHEJswhxCccIcQnRCHEJ6ghnCf4IYgkNCU4J HAk/CSsJJgk6CQ0JRAn0CE4J9AhdCfQIbAn0CHYJ6giFCeAIlAnMCJkJvQie CakIowmQCKMJfAijCWgInglPCJkJOwiUCSwIhQkYCHYJDghsCQQIXQkECE4J BAhECQQIOgnrBysJ0gccCbkHDQmqB/4IlgfqCJEH0QiHB8cIhwezCIcHmgiH B4sIkQd8CJYHaAiqB1kIuQdFCKoHNgiWByIIkQcTCIcH/weHB+sHhwfXB4cH wweRB7QHlgegB6oHlge5B4cH0gdzB+sHbgcECF8HBAhVBwQIRgcECDwHDggt BxgIHgcsCBkHOwgPB08IDwdoCA8HfAgPB5AIDwepCBkHvQgeB8wILQfgCDwH 6ghGB/QIVQf0CF8H9AhuB/QIBAAAAAQBDQAJAAAA+gIGAB4AAAAAAAAAIgAE AAAALQECAAQAAADwAQUABAAAAC0BAAAHAAAA/AIAAH9/fwAAAAQAAAAtAQUA BAAAAAYBAQAOAAAAJAMFAMoUNApRFbIJNBmyCa0YNArKFDQKBAAAAC0BAAAE AAAA8AEFAAQAAAAEAQ0ADgAAACUDBQDKFDQKURWyCTQZsgmtGDQKyhQ0CgQA AAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAO AAAAJAMFAMoULwqyGC8KshjtCsoU7QrKFC8KBAAAAC0BAAAEAAAA8AEFAAQA AAAEAQ0ADgAAACUDBQDKFC8KshgvCrIY7QrKFO0KyhQvCgQAAAAEAQ0ABAAA AC0BAAAHAAAA/AIAAKWlpQAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFALIY 8go0GWYKNBmyCbIYNAqyGPIKBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAA ACUDBQCyGPIKNBlmCjQZsgmyGDQKshjyCgQAAAAEAQ0ABAAAAC0BAAAHAAAA /AIAANjY2AAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFAM8U8QY4FY0GGxmN BrIY8QbPFPEGBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQDPFPEG OBWNBhsZjQayGPEGzxTxBgQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/vwAA AAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFAMoUDAqyGAwKshjsBsoU7AbKFAwK BAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQDKFAwKshgMCrIY7AbK FOwGyhQMCgQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAKWlpQAAAAQAAAAtAQUA BAAAAAYBAQAOAAAAJAMFABsZiAatGPEGrRgMChsZowkbGYgGBAAAAC0BAAAE AAAA8AEFAAQAAAAEAQ0ADgAAACUDBQAbGYgGrRjxBq0YDAobGaMJGxmIBgQA AAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAO AAAAJAMFANoT2At6FBALvBgQCxwY2AvaE9gLBAAAAC0BAAAEAAAA8AEFAAQA AAAEAQ0ADgAAACUDBQDaE9gLehQQC7wYEAscGNgL2hPYCwQAAAAEAQ0ABAAA AC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFANoT 2At6FBALvBgQCxwY2AvaE9gLBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAA ACUDBQDaE9gLehQQC7wYEAscGNgL2hPYCwQAAAAEAQ0ABAAAAC0BAAAHAAAA /AIAAKWlpQAAAAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFACEYAAy3GFYLtxgV CyEY0wshGAAMBAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQAhGAAM txhWC7cYFQshGNMLIRgADAQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAKWlpQAA AAQAAAAtAQUABAAAAAYBAQAOAAAAJAMFACEYAAy3GFYLtxgVCyEY0wshGAAM BAAAAC0BAAAEAAAA8AEFAAQAAAAEAQ0ADgAAACUDBQAhGAAMtxhWC7cYFQsh GNMLIRgADAQAAAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUA BAAAAAYBAQAOAAAAJAMFAN8T2AshGNgLIRgADN8TAAzfE9gLBAAAAC0BAAAE AAAA8AEFAAQAAAAEAQ0ADgAAACUDBQDfE9gLIRjYCyEYAAzfEwAM3xPYCwQA AAAEAQ0ABAAAAC0BAAAHAAAA/AIAAL+/vwAAAAQAAAAtAQUABAAAAAYBAQAO AAAAJAMFAN8T2AshGNgLIRgADN8TAAzfE9gLBAAAAC0BAAAEAAAA8AEFAAQA AAAEAQ0ADgAAACUDBQDfE9gLIRjYCyEYAAzfEwAM3xPYCwQAAAAEAQ0ABAAA AC0BAAAEAAAALQEEAAQAAAAGAQEADgAAACQDBQBTGFAHKRVQBykVqAlTGKgJ UxhQBwQAAAAtAQAABAAAAAQBDQAOAAAAJQMFAFMYUAcpFVAHKRWoCVMYqAlT GFAHBAAAAAQBDQAFAAAAFAKYCnwXBQAAABMCmApEGAQAAAAEAQ0ACQAAAPoC BgBBAAAAAAAAACIABAAAAC0BBQAEAAAA8AECAAQAAAAtAQAABAAAAC0BAwAE AAAABgEBAMYAAAAkA2EAzhX0CNMVDQnnFSYJ9hU/CQAWTgkUFmIJIxZnCTcW cQlLFnEJXxZxCXMWcQmCFmcJlhZiCaUWTgm5Fj8JyBZOCdwWYgnrFmcJ+hZx CRMXcQknF3EJMRdxCUoXZwleF2IJbRdOCXwXPwmLFyYJmhcNCaQX9AiuF/QI vRf0CMwX9AjWF+oI5RfgCPQXzAj5F70I/hepCAMYkAgDGHwIAxhoCP4XTwj5 FzsI9BcsCOUXGAjWFw4IzBcECL0XBAiuFwQIpBcECJoX6weLF9IHfBe5B20X qgdeF5YHSheRBzEXhwcnF4cHExeHB/oWhwfrFpEH3BaWB8gWqge5FrkHpRaq B5YWlgeCFpEHcxaHB18WhwdLFocHNxaHByMWkQcUFpYHABaqB/YVuQfnFdIH 0xXrB84VBAi/FQQItRUECKYVBAicFQ4IjRUYCH4VLAh5FTsIbxVPCG8VaAhv FXwIbxWQCG8VqQh5Fb0IfhXMCI0V4AicFeoIphX0CLUV9Ai/FfQIzhX0CAQA AAAtAQAABAAAAAQBDQDGAAAAJQNhAM4V9AjTFQ0J5xUmCfYVPwkAFk4JFBZi CSMWZwk3FnEJSxZxCV8WcQlzFnEJghZnCZYWYgmlFk4JuRY/CcgWTgncFmIJ 6xZnCfoWcQkTF3EJJxdxCTEXcQlKF2cJXhdiCW0XTgl8Fz8JixcmCZoXDQmk F/QIrhf0CL0X9AjMF/QI1hfqCOUX4Aj0F8wI+Re9CP4XqQgDGJAIAxh8CAMY aAj+F08I+Rc7CPQXLAjlFxgI1hcOCMwXBAi9FwQIrhcECKQXBAiaF+sHixfS B3wXuQdtF6oHXheWB0oXkQcxF4cHJxeHBxMXhwf6FocH6xaRB9wWlgfIFqoH uRa5B6UWqgeWFpYHghaRB3MWhwdfFocHSxaHBzcWhwcjFpEHFBaWBwAWqgf2 FbkH5xXSB9MV6wfOFQQIvxUECLUVBAimFQQInBUOCI0VGAh+FSwIeRU7CG8V TwhvFWgIbxV8CG8VkAhvFakIeRW9CH4VzAiNFeAInBXqCKYV9Ai1FfQIvxX0 CM4V9AgEAAAABAENAAkAAAD6AgYAHgAAAAAAAAAiAAQAAAAtAQIABAAAAPAB BQAEAAAALQEAAAcAAAD8AgAAf39/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAk AwUASg20EdENMhG0ETIRLRG0EUoNtBEEAAAALQEAAAQAAADwAQUABAAAAAQB DQAOAAAAJQMFAEoNtBHRDTIRtBEyES0RtBFKDbQRBAAAAAQBDQAEAAAALQEA AAcAAAD8AgAAv7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUASg2vETIR rxEyEW0SSg1tEkoNrxEEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMF AEoNrxEyEa8RMhFtEkoNbRJKDa8RBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA paWlAAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAMhFyErQR5hG0ETIRMhG0 ETIRchIEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFADIRchK0EeYR tBEyETIRtBEyEXISBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA2NjYAAAABAAA AC0BBQAEAAAABgEBAA4AAAAkAwUATw1xDrgNDQ6bEQ0OMhFxDk8NcQ4EAAAA LQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFAE8NcQ64DQ0OmxENDjIRcQ5P DXEOBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBQAEAAAA BgEBAA4AAAAkAwUASg2METIRjBEyEWwOSg1sDkoNjBEEAAAALQEAAAQAAADw AQUABAAAAAQBDQAOAAAAJQMFAEoNjBEyEYwRMhFsDkoNbA5KDYwRBAAAAAQB DQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBQAEAAAABgEBAA4AAAAk AwUAmxEIDi0RcQ4tEYwRmxEjEZsRCA4EAAAALQEAAAQAAADwAQUABAAAAAQB DQAOAAAAJQMFAJsRCA4tEXEOLRGMEZsRIxGbEQgOBAAAAAQBDQAEAAAALQEA AAcAAAD8AgAAv7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAWgxYE/oM kBI8EZASnBBYE1oMWBMEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMF AFoMWBP6DJASPBGQEpwQWBNaDFgTBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAA v7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAWgxYE/oMkBI8EZASnBBY E1oMWBMEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFAFoMWBP6DJAS PBGQEpwQWBNaDFgTBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAA AC0BBQAEAAAABgEBAA4AAAAkAwUAoRCAEzcR1hI3EZUSoRBTE6EQgBMEAAAA LQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMFAKEQgBM3EdYSNxGVEqEQUxOh EIATBAAAAAQBDQAEAAAALQEAAAcAAAD8AgAApaWlAAAABAAAAC0BBQAEAAAA BgEBAA4AAAAkAwUAoRCAEzcR1hI3EZUSoRBTE6EQgBMEAAAALQEAAAQAAADw AQUABAAAAAQBDQAOAAAAJQMFAKEQgBM3EdYSNxGVEqEQUxOhEIATBAAAAAQB DQAEAAAALQEAAAcAAAD8AgAAv7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAk AwUAXwxYE6EQWBOhEIATXwyAE18MWBMEAAAALQEAAAQAAADwAQUABAAAAAQB DQAOAAAAJQMFAF8MWBOhEFgToRCAE18MgBNfDFgTBAAAAAQBDQAEAAAALQEA AAcAAAD8AgAAv7+/AAAABAAAAC0BBQAEAAAABgEBAA4AAAAkAwUAXwxYE6EQ WBOhEIATXwyAE18MWBMEAAAALQEAAAQAAADwAQUABAAAAAQBDQAOAAAAJQMF AF8MWBOhEFgToRCAE18MgBNfDFgTBAAAAAQBDQAEAAAALQEAAAQAAAAtAQQA BAAAAAYBAQAOAAAAJAMFANMQ0A6pDdAOqQ0oEdMQKBHTENAOBAAAAC0BAAAE AAAABAENAA4AAAAlAwUA0xDQDqkN0A6pDSgR0xAoEdMQ0A4EAAAABAENAAUA AAAUAhgS/A8FAAAAEwIYEsQQBAAAAAQBDQAJAAAA+gIGAEEAAAAAAAAAIgAE AAAALQEFAAQAAADwAQIABAAAAC0BAAAEAAAALQEDAAQAAAAGAQEAxgAAACQD YQBODnQQUw6NEGcOphB2Dr8QgA7OEJQO4hCjDucQtw7xEMsO8RDfDvEQ8w7x EAIP5xAWD+IQJQ/OEDkPvxBID84QXA/iEGsP5xB6D/EQkw/xEKcP8RCxD/EQ yg/nEN4P4hDtD84Q/A+/EAsQphAaEI0QJBB0EC4QdBA9EHQQTBB0EFYQahBl EGAQdBBMEHkQPRB+ECkQgxAQEIMQ/A+DEOgPfhDPD3kQuw90EKwPZRCYD1YQ jg9MEIQPPRCEDy4QhA8kEIQPGhBrDwsQUg/8DzkP7Q8qD94PFg/KDxEPsQ8H D6cPBw+TDwcPeg8HD2sPEQ9cDxYPSA8qDzkPOQ8lDyoPFg8WDwIPEQ/zDgcP 3w4HD8sOBw+3DgcPow4RD5QOFg+ADioPdg45D2cOUg9TDmsPTg6EDz8OhA81 DoQPJg6EDxwOjg8NDpgP/g2sD/kNuw/vDc8P7w3oD+8N/A/vDRAQ7w0pEPkN PRD+DUwQDQ5gEBwOahAmDnQQNQ50ED8OdBBODnQQBAAAAC0BAAAEAAAABAEN AMYAAAAlA2EATg50EFMOjRBnDqYQdg6/EIAOzhCUDuIQow7nELcO8RDLDvEQ 3w7xEPMO8RACD+cQFg/iECUPzhA5D78QSA/OEFwP4hBrD+cQeg/xEJMP8RCn D/EQsQ/xEMoP5xDeD+IQ7Q/OEPwPvxALEKYQGhCNECQQdBAuEHQQPRB0EEwQ dBBWEGoQZRBgEHQQTBB5ED0QfhApEIMQEBCDEPwPgxDoD34Qzw95ELsPdBCs D2UQmA9WEI4PTBCEDz0QhA8uEIQPJBCEDxoQaw8LEFIP/A85D+0PKg/eDxYP yg8RD7EPBw+nDwcPkw8HD3oPBw9rDxEPXA8WD0gPKg85DzkPJQ8qDxYPFg8C DxEP8w4HD98OBw/LDgcPtw4HD6MOEQ+UDhYPgA4qD3YOOQ9nDlIPUw5rD04O hA8/DoQPNQ6EDyYOhA8cDo4PDQ6YD/4NrA/5DbsP7w3PD+8N6A/vDfwP7w0Q EO8NKRD5DT0Q/g1MEA0OYBAcDmoQJg50EDUOdBA/DnQQTg50EAQAAAAEAQ0A BAAAACcB//8eAAAAJgYPADEAQ29weXJpZ2h0IKkgMTk5NiBNaWNyb3N5c3Rl bXMgRW5naW5lZXJpbmcgQ29tcGFueQAPAAAAJgYPABMAQWxsIFJpZ2h0cyBS ZXNlcnZlZAAEAAAA8AEBAAMAAAAAAAAAEQAAAAwAAAAIAAAACwAAABAAAAAx AQAA5QAAAAkAAAAQAAAAMQEAAOUAAAAJAAAAEAAAAG4fAACPFwAACQAAABAA AABuHwAAjxcAACAAAAAYAAAAAQAAAAEAAAABAAAAAQAAACEAAAAIAAAAFAAA AAwAAAANAAAAJQAAAAwAAAAFAACAJgAAABwAAAABAAAAAAAAAAAAAAAAAAAA AAAAACUAAAAMAAAAAQAAACYAAAAcAAAAAgAAAAYAAABBAAAAAAAAAAAAAAAl AAAADAAAAAIAAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAA////AAAA AAAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAoAEAAP7////+////MwEA AOcAAABhAAAArQRzESUFqRLABcYTYAbFFCMHoRXwB1UWzAjXFqgJNheYCm0X fgttF18MShdPDfoWJg5uFu4OuhW2D94UahC6FUYRbhYTEvoW+RJKF+4TbRfF FG0XphU2F4wW1xZtF1UWNRihFfMYxRSdGcYTLhqpEroacxFVG5sRBByMEakc XxFOHfsQ5B1qEFcesQ/FHtAOAR/bDTgf1wxMH8kLOB+xCgEfqAnFHrgIVx7m B+QdKAdOHZwGqRwpBgQc9wVVG+0FuhokBi4a5ASdGb0D8xi5AjUY7AFtFy4B jBaxAKYVTQDFFCAA7hMgAPkSPgATEp0ARhEkAWoQzgG2D6AC7g7OASYOJAFP DZ0AXww+AH4LIACYCiAAqAlNAMwIsQDwBy4BIwfsAWAGuQLABb0DJQXkBK0E JAYDBO0FWQP3BbQCKQYUApwGgwEoBwsB5gedALgITQCoCSUAsQogAMkLJQDX DE0A2w2dANAOCwGxD4MBahAUAvsQtAJfEVkDjBEDBJsRrQRzESUAAAAMAAAA BQAAgBQAAAAMAAAADQAAAFcAAACgAQAA/v////7///8zAQAA5wAAAGEAAACt BHMRJQWpEsAFxhNgBsUUIwehFfAHVRbMCNcWqAk2F5gKbRd+C20XXwxKF08N +hYmDm4W7g66FbYP3hRqELoVRhFuFhMS+hb5EkoX7hNtF8UUbRemFTYXjBbX Fm0XVRY1GKEV8xjFFJ0ZxhMuGqkSuhpzEVUbmxEEHIwRqRxfEU4d+xDkHWoQ Vx6xD8Ue0A4BH9sNOB/XDEwfyQs4H7EKAR+oCcUeuAhXHuYH5B0oB04dnAap HCkGBBz3BVUb7QW6GiQGLhrkBJ0ZvQPzGLkCNRjsAW0XLgGMFrEAphVNAMUU IADuEyAA+RI+ABMSnQBGESQBahDOAbYPoALuDs4BJg4kAU8NnQBfDD4Afgsg AJgKIACoCU0AzAixAPAHLgEjB+wBYAa5AsAFvQMlBeQErQQkBgME7QVZA/cF tAIpBhQCnAaDASgHCwHmB50AuAhNAKgJJQCxCiAAyQslANcMTQDbDZ0A0A4L AbEPgwFqEBQC+xC0Al8RWQOMEQMEmxGtBHMRFAAAAAwAAAANAAAAJgAAABwA AAAEAAAABgAAAEEAAAAAAAAA/wAAACUAAAAMAAAABAAAACgAAAAMAAAAAgAA AFcAAAAsAAAATQAAAEwAAADkAAAAqAAAAAQAAABmD+wQNggsCDYXLAhmD+wQ FAAAAAwAAAANAAAAJgAAABwAAAACAAAABgAAAB4AAAAAAAAAAAAAACUAAAAM AAAAAgAAACgAAAAMAAAABAAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAA AAB/f38AAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAPgAA AF4AAABpAAAAYwAAAAUAAABqBjQK8QayCdQKsglNCjQKagY0CiUAAAAMAAAA BQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAPgAAAF4AAABp AAAAYwAAAAUAAABqBjQK8QayCdQKsglNCjQKagY0ChQAAAAMAAAADQAAACUA AAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAC/v78AAAAAACUAAAAMAAAABAAA ABMAAAAMAAAAAQAAAFYAAAAwAAAAPgAAAGMAAABkAAAAagAAAAUAAABqBi8K UgovClIK7QpqBu0KagYvCiUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAM AAAADQAAAFcAAAAwAAAAPgAAAGMAAABkAAAAagAAAAUAAABqBi8KUgovClIK 7QpqBu0KagYvChQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAA AAAAAAClpaUAAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAA ZAAAAF4AAABpAAAAagAAAAUAAABSCvIK1ApmCtQKsglSCjQKUgryCiUAAAAM AAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAZAAAAF4A AABpAAAAagAAAAUAAABSCvIK1ApmCtQKsglSCjQKUgryChQAAAAMAAAADQAA ACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAADY2NgAAAAAACUAAAAMAAAA BAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAPgAAAEAAAABoAAAARAAAAAUAAABv BvEG2AaNBrsKjQZSCvEGbwbxBiUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQA AAAMAAAADQAAAFcAAAAwAAAAPgAAAEAAAABoAAAAQwAAAAUAAABvBvEG2AaN BrsKjQZSCvEGbwbxBhQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAA BAAAAAAAAAC/v78AAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAw AAAAPgAAAEMAAABkAAAAYgAAAAUAAABqBgwKUgoMClIK7AZqBuwGagYMCiUA AAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAPgAA AEMAAABkAAAAYgAAAAUAAABqBgwKUgoMClIK7AZqBuwGagYMChQAAAAMAAAA DQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAClpaUAAAAAACUAAAAM AAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAZAAAAEAAAABoAAAAYgAAAAUA AAC7CogGTQrxBk0KDAq7CqMJuwqIBiUAAAAMAAAABQAAgCgAAAAMAAAABAAA ABQAAAAMAAAADQAAAFcAAAAwAAAAZAAAAD8AAABoAAAAYgAAAAUAAAC7CogG TQrxBk0KDAq7CqMJuwqIBhQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAY AAAABAAAAAAAAAC/v78AAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYA AAAwAAAANQAAAGwAAABlAAAAcwAAAAUAAAB6BdgLGgYQC1wKEAu8CdgLegXY CyUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAA NQAAAGwAAABlAAAAcwAAAAUAAAB6BdgLGgYQC1wKEAu8CdgLegXYCxQAAAAM AAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAC/v78AAAAAACUA AAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAANQAAAGwAAABlAAAAcwAA AAUAAAB6BdgLGgYQC1wKEAu8CdgLegXYCyUAAAAMAAAABQAAgCgAAAAMAAAA BAAAABQAAAAMAAAADQAAAFcAAAAwAAAANQAAAGwAAABlAAAAcwAAAAUAAAB6 BdgLGgYQC1wKEAu8CdgLegXYCxQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcA AAAYAAAABAAAAAAAAAClpaUAAAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAA AFYAAAAwAAAAXwAAAGwAAABkAAAAdQAAAAUAAADBCQAMVwpWC1cKFQvBCdML wQkADCUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAw AAAAXwAAAGwAAABkAAAAdQAAAAUAAADBCQAMVwpWC1cKFQvBCdMLwQkADBQA AAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAClpaUAAAAA ACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAAXwAAAGwAAABkAAAA dQAAAAUAAADBCQAMVwpWC1cKFQvBCdMLwQkADCUAAAAMAAAABQAAgCgAAAAM AAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAAXwAAAGwAAABkAAAAdQAAAAUA AADBCQAMVwpWC1cKFQvBCdMLwQkADBQAAAAMAAAADQAAACUAAAAMAAAABQAA gCcAAAAYAAAABAAAAAAAAAC/v78AAAAAACUAAAAMAAAABAAAABMAAAAMAAAA AQAAAFYAAAAwAAAANQAAAHMAAABfAAAAdQAAAAUAAAB/BdgLwQnYC8EJAAx/ BQAMfwXYCyUAAAAMAAAABQAAgCgAAAAMAAAABAAAABQAAAAMAAAADQAAAFcA AAAwAAAANQAAAHMAAABfAAAAdQAAAAUAAAB/BdgLwQnYC8EJAAx/BQAMfwXY CxQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAABAAAAAAAAAC/v78A AAAAACUAAAAMAAAABAAAABMAAAAMAAAAAQAAAFYAAAAwAAAANQAAAHMAAABf AAAAdQAAAAUAAAB/BdgLwQnYC8EJAAx/BQAMfwXYCyUAAAAMAAAABQAAgCgA AAAMAAAABAAAABQAAAAMAAAADQAAAFcAAAAwAAAANQAAAHMAAABfAAAAdQAA AAUAAAB/BdgLwQnYC8EJAAx/BQAMfwXYCxQAAAAMAAAADQAAACUAAAAMAAAA BQAAgCcAAAAYAAAABAAAAAAAAAAAAP8AAAAAACUAAAAMAAAABAAAABMAAAAM AAAAAQAAAFYAAAAwAAAAQgAAAEcAAABhAAAAXgAAAAUAAADzCVAHyQZQB8kG qAnzCagJ8wlQByUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAAAwAAAAQgAA AEcAAABhAAAAXgAAAAUAAADzCVAHyQZQB8kGqAnzCagJ8wlQBxQAAAAMAAAA DQAAABsAAAAQAAAAHAkAAJgKAAA2AAAAEAAAAOQJAACYCgAAFAAAAAwAAAAN AAAAJgAAABwAAAAFAAAABgAAAEEAAAAAAAAAAAAAACUAAAAMAAAABQAAACgA AAAMAAAAAgAAACUAAAAMAAAABQAAgCUAAAAMAAAAAwAAABMAAAAMAAAAAQAA AFYAAACgAQAAQgAAAEYAAABhAAAAXwAAAGEAAABuB/QIcwcNCYcHJgmWBz8J oAdOCbQHYgnDB2cJ1wdxCesHcQn/B3EJEwhxCSIIZwk2CGIJRQhOCVkIPwlo CE4JfAhiCYsIZwmaCHEJswhxCccIcQnRCHEJ6ghnCf4IYgkNCU4JHAk/CSsJ Jgk6CQ0JRAn0CE4J9AhdCfQIbAn0CHYJ6giFCeAIlAnMCJkJvQieCakIowmQ CKMJfAijCWgInglPCJkJOwiUCSwIhQkYCHYJDghsCQQIXQkECE4JBAhECQQI OgnrBysJ0gccCbkHDQmqB/4IlgfqCJEH0QiHB8cIhwezCIcHmgiHB4sIkQd8 CJYHaAiqB1kIuQdFCKoHNgiWByIIkQcTCIcH/weHB+sHhwfXB4cHwweRB7QH lgegB6oHlge5B4cH0gdzB+sHbgcECF8HBAhVBwQIRgcECDwHDggtBxgIHgcs CBkHOwgPB08IDwdoCA8HfAgPB5AIDwepCBkHvQgeB8wILQfgCDwH6ghGB/QI VQf0CF8H9AhuB/QIJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAAKABAABC AAAARgAAAGEAAABfAAAAYQAAAG4H9AhzBw0JhwcmCZYHPwmgB04JtAdiCcMH ZwnXB3EJ6wdxCf8HcQkTCHEJIghnCTYIYglFCE4JWQg/CWgITgl8CGIJiwhn CZoIcQmzCHEJxwhxCdEIcQnqCGcJ/ghiCQ0JTgkcCT8JKwkmCToJDQlECfQI Tgn0CF0J9AhsCfQIdgnqCIUJ4AiUCcwImQm9CJ4JqQijCZAIowl8CKMJaAie CU8ImQk7CJQJLAiFCRgIdgkOCGwJBAhdCQQITgkECEQJBAg6CesHKwnSBxwJ uQcNCaoH/giWB+oIkQfRCIcHxwiHB7MIhweaCIcHiwiRB3wIlgdoCKoHWQi5 B0UIqgc2CJYHIgiRBxMIhwf/B4cH6weHB9cHhwfDB5EHtAeWB6AHqgeWB7kH hwfSB3MH6wduBwQIXwcECFUHBAhGBwQIPAcOCC0HGAgeBywIGQc7CA8HTwgP B2gIDwd8CA8HkAgPB6kIGQe9CB4HzAgtB+AIPAfqCEYH9AhVB/QIXwf0CG4H 9AgUAAAADAAAAA0AAAAmAAAAHAAAAAIAAAAGAAAAHgAAAAAAAAAAAAAAJQAA AAwAAAACAAAAKAAAAAwAAAAFAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAA AAAAAH9/fwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADK AAAAXgAAAPUAAABjAAAABQAAAMoUNApRFbIJNBmyCa0YNArKFDQKJQAAAAwA AAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADKAAAAXgAA APUAAABjAAAABQAAAMoUNApRFbIJNBmyCa0YNArKFDQKFAAAAAwAAAANAAAA JQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAF AAAAEwAAAAwAAAABAAAAVgAAADAAAADKAAAAYwAAAPAAAABqAAAABQAAAMoU LwqyGC8KshjtCsoU7QrKFC8KJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAA AAwAAAANAAAAVwAAADAAAADKAAAAYwAAAPAAAABqAAAABQAAAMoULwqyGC8K shjtCsoU7QrKFC8KFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAF AAAAAAAAAKWlpQAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAA AADwAAAAXgAAAPUAAABqAAAABQAAALIY8go0GWYKNBmyCbIYNAqyGPIKJQAA AAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADwAAAA XgAAAPUAAABqAAAABQAAALIY8go0GWYKNBmyCbIYNAqyGPIKFAAAAAwAAAAN AAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAANjY2AAAAAAAJQAAAAwA AAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADKAAAAQAAAAPQAAABEAAAABQAA AM8U8QY4FY0GGxmNBrIY8QbPFPEGJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAA FAAAAAwAAAANAAAAVwAAADAAAADKAAAAQAAAAPQAAABDAAAABQAAAM8U8QY4 FY0GGxmNBrIY8QbPFPEGFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgA AAAFAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAA ADAAAADKAAAAQwAAAPAAAABiAAAABQAAAMoUDAqyGAwKshjsBsoU7AbKFAwK JQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADK AAAAQwAAAPAAAABiAAAABQAAAMoUDAqyGAwKshjsBsoU7AbKFAwKFAAAAAwA AAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAKWlpQAAAAAAJQAA AAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADvAAAAQAAAAPQAAABiAAAA BQAAABsZiAatGPEGrRgMChsZowkbGYgGJQAAAAwAAAAFAACAKAAAAAwAAAAF AAAAFAAAAAwAAAANAAAAVwAAADAAAADvAAAAPwAAAPQAAABiAAAABQAAABsZ iAatGPEGrRgMChsZowkbGYgGFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAA ABgAAAAFAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAA VgAAADAAAADBAAAAbAAAAPAAAABzAAAABQAAANoT2At6FBALvBgQCxwY2Ava E9gLJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAA AADBAAAAbAAAAPAAAABzAAAABQAAANoT2At6FBALvBgQCxwY2AvaE9gLFAAA AAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAL+/vwAAAAAA JQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADBAAAAbAAAAPAAAABz AAAABQAAANoT2At6FBALvBgQCxwY2AvaE9gLJQAAAAwAAAAFAACAKAAAAAwA AAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADBAAAAbAAAAPAAAABzAAAABQAA ANoT2At6FBALvBgQCxwY2AvaE9gLFAAAAAwAAAANAAAAJQAAAAwAAAAFAACA JwAAABgAAAAFAAAAAAAAAKWlpQAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAAB AAAAVgAAADAAAADqAAAAbAAAAPAAAAB1AAAABQAAACEYAAy3GFYLtxgVCyEY 0wshGAAMJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAA ADAAAADqAAAAbAAAAPAAAAB1AAAABQAAACEYAAy3GFYLtxgVCyEY0wshGAAM FAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAKWlpQAA AAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADqAAAAbAAAAPAA AAB1AAAABQAAACEYAAy3GFYLtxgVCyEY0wshGAAMJQAAAAwAAAAFAACAKAAA AAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADqAAAAbAAAAPAAAAB1AAAA BQAAACEYAAy3GFYLtxgVCyEY0wshGAAMFAAAAAwAAAANAAAAJQAAAAwAAAAF AACAJwAAABgAAAAFAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwA AAABAAAAVgAAADAAAADBAAAAcwAAAOoAAAB1AAAABQAAAN8T2AshGNgLIRgA DN8TAAzfE9gLJQAAAAwAAAAFAACAKAAAAAwAAAAFAAAAFAAAAAwAAAANAAAA VwAAADAAAADBAAAAcwAAAOoAAAB1AAAABQAAAN8T2AshGNgLIRgADN8TAAzf E9gLFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAFAAAAAAAAAL+/ vwAAAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAAVgAAADAAAADBAAAAcwAA AOoAAAB1AAAABQAAAN8T2AshGNgLIRgADN8TAAzfE9gLJQAAAAwAAAAFAACA KAAAAAwAAAAFAAAAFAAAAAwAAAANAAAAVwAAADAAAADBAAAAcwAAAOoAAAB1 AAAABQAAAN8T2AshGNgLIRgADN8TAAzfE9gLFAAAAAwAAAANAAAAJQAAAAwA AAAFAACAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADNAAAARwAA AOwAAABeAAAABQAAAFMYUAcpFVAHKRWoCVMYqAlTGFAHJQAAAAwAAAAFAACA FAAAAAwAAAANAAAAVwAAADAAAADNAAAARwAAAOwAAABeAAAABQAAAFMYUAcp FVAHKRWoCVMYqAlTGFAHFAAAAAwAAAANAAAAGwAAABAAAAB8FwAAmAoAADYA AAAQAAAARBgAAJgKAAAUAAAADAAAAA0AAAAmAAAAHAAAAAUAAAAGAAAAQQAA AAAAAAAAAAAAJQAAAAwAAAAFAAAAKAAAAAwAAAACAAAAJQAAAAwAAAAFAACA JQAAAAwAAAADAAAAEwAAAAwAAAABAAAAVgAAAKABAADNAAAARgAAAOwAAABf AAAAYQAAAM4V9AjTFQ0J5xUmCfYVPwkAFk4JFBZiCSMWZwk3FnEJSxZxCV8W cQlzFnEJghZnCZYWYgmlFk4JuRY/CcgWTgncFmIJ6xZnCfoWcQkTF3EJJxdx CTEXcQlKF2cJXhdiCW0XTgl8Fz8JixcmCZoXDQmkF/QIrhf0CL0X9AjMF/QI 1hfqCOUX4Aj0F8wI+Re9CP4XqQgDGJAIAxh8CAMYaAj+F08I+Rc7CPQXLAjl FxgI1hcOCMwXBAi9FwQIrhcECKQXBAiaF+sHixfSB3wXuQdtF6oHXheWB0oX kQcxF4cHJxeHBxMXhwf6FocH6xaRB9wWlgfIFqoHuRa5B6UWqgeWFpYHghaR B3MWhwdfFocHSxaHBzcWhwcjFpEHFBaWBwAWqgf2FbkH5xXSB9MV6wfOFQQI vxUECLUVBAimFQQInBUOCI0VGAh+FSwIeRU7CG8VTwhvFWgIbxV8CG8VkAhv FakIeRW9CH4VzAiNFeAInBXqCKYV9Ai1FfQIvxX0CM4V9AglAAAADAAAAAUA AIAUAAAADAAAAA0AAABXAAAAoAEAAM0AAABGAAAA7AAAAF8AAABhAAAAzhX0 CNMVDQnnFSYJ9hU/CQAWTgkUFmIJIxZnCTcWcQlLFnEJXxZxCXMWcQmCFmcJ lhZiCaUWTgm5Fj8JyBZOCdwWYgnrFmcJ+hZxCRMXcQknF3EJMRdxCUoXZwle F2IJbRdOCXwXPwmLFyYJmhcNCaQX9AiuF/QIvRf0CMwX9AjWF+oI5RfgCPQX zAj5F70I/hepCAMYkAgDGHwIAxhoCP4XTwj5FzsI9BcsCOUXGAjWFw4IzBcE CL0XBAiuFwQIpBcECJoX6weLF9IHfBe5B20XqgdeF5YHSheRBzEXhwcnF4cH ExeHB/oWhwfrFpEH3BaWB8gWqge5FrkHpRaqB5YWlgeCFpEHcxaHB18WhwdL FocHNxaHByMWkQcUFpYHABaqB/YVuQfnFdIH0xXrB84VBAi/FQQItRUECKYV BAicFQ4IjRUYCH4VLAh5FTsIbxVPCG8VaAhvFXwIbxWQCG8VqQh5Fb0IfhXM CI0V4AicFeoIphX0CLUV9Ai/FfQIzhX0CBQAAAAMAAAADQAAACYAAAAcAAAA AgAAAAYAAAAeAAAAAAAAAAAAAAAlAAAADAAAAAIAAAAoAAAADAAAAAUAAAAl AAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAf39/AAAAAAAlAAAADAAAAAUA AAATAAAADAAAAAEAAABWAAAAMAAAAIEAAACnAAAArAAAAKwAAAAFAAAASg20 EdENMhG0ETIRLRG0EUoNtBElAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAA DAAAAA0AAABXAAAAMAAAAIEAAACnAAAArAAAAKwAAAAFAAAASg20EdENMhG0 ETIRLRG0EUoNtBEUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUA AAAAAAAAv7+/AAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAA AIEAAACsAAAApwAAALMAAAAFAAAASg2vETIRrxEyEW0SSg1tEkoNrxElAAAA DAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAIEAAACs AAAApwAAALMAAAAFAAAASg2vETIRrxEyEW0SSg1tEkoNrxEUAAAADAAAAA0A AAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAApaWlAAAAAAAlAAAADAAA AAUAAAATAAAADAAAAAEAAABWAAAAMAAAAKcAAACnAAAArAAAALMAAAAFAAAA MhFyErQR5hG0ETIRMhG0ETIRchIlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAU AAAADAAAAA0AAABXAAAAMAAAAKcAAACnAAAArAAAALMAAAAFAAAAMhFyErQR 5hG0ETIRMhG0ETIRchIUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAA AAUAAAAAAAAA2NjYAAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAA MAAAAIEAAACJAAAAqwAAAIwAAAAFAAAATw1xDrgNDQ6bEQ0OMhFxDk8NcQ4l AAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAIEA AACJAAAAqwAAAIwAAAAFAAAATw1xDrgNDQ6bEQ0OMhFxDk8NcQ4UAAAADAAA AA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/AAAAAAAlAAAA DAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAIEAAACMAAAApwAAAKsAAAAF AAAASg2METIRjBEyEWwOSg1sDkoNjBElAAAADAAAAAUAAIAoAAAADAAAAAUA AAAUAAAADAAAAA0AAABXAAAAMAAAAIEAAACMAAAApwAAAKsAAAAFAAAASg2M ETIRjBEyEWwOSg1sDkoNjBEUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAA GAAAAAUAAAAAAAAApaWlAAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABW AAAAMAAAAKcAAACIAAAAqwAAAKsAAAAFAAAAmxEIDi0RcQ4tEYwRmxEjEZsR CA4lAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAA AKcAAACIAAAAqwAAAKsAAAAFAAAAmxEIDi0RcQ4tEYwRmxEjEZsRCA4UAAAA DAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/AAAAAAAl AAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAHgAAAC0AAAApwAAALwA AAAFAAAAWgxYE/oMkBI8EZASnBBYE1oMWBMlAAAADAAAAAUAAIAoAAAADAAA AAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAHgAAAC0AAAApwAAALwAAAAFAAAA WgxYE/oMkBI8EZASnBBYE1oMWBMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAn AAAAGAAAAAUAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEA AABWAAAAMAAAAHgAAAC0AAAApwAAALwAAAAFAAAAWgxYE/oMkBI8EZASnBBY E1oMWBMlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAA MAAAAHgAAAC0AAAApwAAALwAAAAFAAAAWgxYE/oMkBI8EZASnBBYE1oMWBMU AAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAApaWlAAAA AAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAKEAAAC1AAAApwAA AL4AAAAFAAAAoRCAEzcR1hI3EZUSoRBTE6EQgBMlAAAADAAAAAUAAIAoAAAA DAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAKEAAAC1AAAApwAAAL4AAAAF AAAAoRCAEzcR1hI3EZUSoRBTE6EQgBMUAAAADAAAAA0AAAAlAAAADAAAAAUA AIAnAAAAGAAAAAUAAAAAAAAApaWlAAAAAAAlAAAADAAAAAUAAAATAAAADAAA AAEAAABWAAAAMAAAAKEAAAC1AAAApwAAAL4AAAAFAAAAoRCAEzcR1hI3EZUS oRBTE6EQgBMlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0AAABX AAAAMAAAAKEAAAC1AAAApwAAAL4AAAAFAAAAoRCAEzcR1hI3EZUSoRBTE6EQ gBMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/ AAAAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAABWAAAAMAAAAHgAAAC8AAAA oQAAAL4AAAAFAAAAXwxYE6EQWBOhEIATXwyAE18MWBMlAAAADAAAAAUAAIAo AAAADAAAAAUAAAAUAAAADAAAAA0AAABXAAAAMAAAAHgAAAC8AAAAoQAAAL4A AAAFAAAAXwxYE6EQWBOhEIATXwyAE18MWBMUAAAADAAAAA0AAAAlAAAADAAA AAUAAIAnAAAAGAAAAAUAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAUAAAATAAAA DAAAAAEAAABWAAAAMAAAAHgAAAC8AAAAoQAAAL4AAAAFAAAAXwxYE6EQWBOh EIATXwyAE18MWBMlAAAADAAAAAUAAIAoAAAADAAAAAUAAAAUAAAADAAAAA0A AABXAAAAMAAAAHgAAAC8AAAAoQAAAL4AAAAFAAAAXwxYE6EQWBOhEIATXwyA E18MWBMUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAA DAAAAAEAAABWAAAAMAAAAIUAAACQAAAAowAAAKcAAAAFAAAA0xDQDqkN0A6p DSgR0xAoEdMQ0A4lAAAADAAAAAUAAIAUAAAADAAAAA0AAABXAAAAMAAAAIUA AACQAAAAowAAAKcAAAAFAAAA0xDQDqkN0A6pDSgR0xAoEdMQ0A4UAAAADAAA AA0AAAAbAAAAEAAAAPwPAAAYEgAANgAAABAAAADEEAAAGBIAABQAAAAMAAAA DQAAACYAAAAcAAAABQAAAAYAAABBAAAAAAAAAAAAAAAlAAAADAAAAAUAAAAo AAAADAAAAAIAAAAlAAAADAAAAAUAAIAlAAAADAAAAAMAAAATAAAADAAAAAEA AABWAAAAoAEAAIQAAACPAAAAowAAAKgAAABhAAAATg50EFMOjRBnDqYQdg6/ EIAOzhCUDuIQow7nELcO8RDLDvEQ3w7xEPMO8RACD+cQFg/iECUPzhA5D78Q SA/OEFwP4hBrD+cQeg/xEJMP8RCnD/EQsQ/xEMoP5xDeD+IQ7Q/OEPwPvxAL EKYQGhCNECQQdBAuEHQQPRB0EEwQdBBWEGoQZRBgEHQQTBB5ED0QfhApEIMQ EBCDEPwPgxDoD34Qzw95ELsPdBCsD2UQmA9WEI4PTBCEDz0QhA8uEIQPJBCE DxoQaw8LEFIP/A85D+0PKg/eDxYPyg8RD7EPBw+nDwcPkw8HD3oPBw9rDxEP XA8WD0gPKg85DzkPJQ8qDxYPFg8CDxEP8w4HD98OBw/LDgcPtw4HD6MOEQ+U DhYPgA4qD3YOOQ9nDlIPUw5rD04OhA8/DoQPNQ6EDyYOhA8cDo4PDQ6YD/4N rA/5DbsP7w3PD+8N6A/vDfwP7w0QEO8NKRD5DT0Q/g1MEA0OYBAcDmoQJg50 EDUOdBA/DnQQTg50ECUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAACgAQAA hAAAAI8AAACjAAAAqAAAAGEAAABODnQQUw6NEGcOphB2Dr8QgA7OEJQO4hCj DucQtw7xEMsO8RDfDvEQ8w7xEAIP5xAWD+IQJQ/OEDkPvxBID84QXA/iEGsP 5xB6D/EQkw/xEKcP8RCxD/EQyg/nEN4P4hDtD84Q/A+/EAsQphAaEI0QJBB0 EC4QdBA9EHQQTBB0EFYQahBlEGAQdBBMEHkQPRB+ECkQgxAQEIMQ/A+DEOgP fhDPD3kQuw90EKwPZRCYD1YQjg9MEIQPPRCEDy4QhA8kEIQPGhBrDwsQUg/8 DzkP7Q8qD94PFg/KDxEPsQ8HD6cPBw+TDwcPeg8HD2sPEQ9cDxYPSA8qDzkP OQ8lDyoPFg8WDwIPEQ/zDgcP3w4HD8sOBw+3DgcPow4RD5QOFg+ADioPdg45 D2cOUg9TDmsPTg6EDz8OhA81DoQPJg6EDxwOjg8NDpgP/g2sD/kNuw/vDc8P 7w3oD+8N/A/vDRAQ7w0pEPkNPRD+DUwQDQ5gEBwOahAmDnQQNQ50ED8OdBBO DnQQFAAAAAwAAAANAAAAIgAAAAwAAAD/////KAAAAAwAAAABAAAAJQAAAAwA AAAHAACAJQAAAAwAAAAAAACAMAAAAAwAAAAPAACASwAAABAAAAAAAAAABQAA ACgAAAAMAAAAAwAAACgAAAAMAAAABAAAACgAAAAMAAAABQAAAA4AAAAUAAAA AAAAABAAAAAUAAAAGkAHQCwAAAAgAAAA/////wIAAAAAAAAAAAAAAIXemEPo xGVDAAAAAOgD6AMhAAAACAAAAGIAAAAMAAAAAQAAACEAAAAIAAAAIQAAAAgA AABzAAAADAAAAAAAAAAcAAAACAAAACUAAAAMAAAAAAAAgCUAAAAMAAAABwAA gCUAAAAMAAAADgAAgDAAAAAMAAAADwAAgBkAAAAMAAAA////ABgAAAAMAAAA AAAAABIAAAAMAAAAAgAAABMAAAAMAAAAAQAAABQAAAAMAAAADQAAABUAAAAM AAAAAQAAABYAAAAMAAAAAAAAAA0AAAAQAAAAAAAAAAAAAAA6AAAADAAAAAoA AAAbAAAAEAAAAAAAAAAAAAAAIwAAACAAAACRnck+AAAAAAAAAAD8bMk+AAAx RAAAs0MkAAAAJAAAAJGdyT4AAAAAAAAAAPxsyT4AADFEAACzQwQAAABzAAAA DAAAAAAAAAANAAAAEAAAAMQCAABmAQAAUgAAAHABAAAFAAAAFAAAAAoAAAAA AAAAAAAAAJABAAAAAACBBwICIlMAeQBzAHQAZQBtAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA/g0A AEwQAAAgAAAAYBAAABwOAABIZfF33GTxdwkAAAAhAIoBDQAAAJTfEgCFNPJ3 0CRkAwQaId7QJGQDJgCKAQQaId4LAIgBv78BAAQAAADc3BIApN0SANDdEgAY 7pB8cAWRfP////9tBZF8WpmAfAAAFAAAAAAAbZmAfAAAMFiQT1kDIDD1dyUA AACwRhUAAAAQALBGFQDk3RIAmyjyd+gUIwCwRhUA0CRkA0dESUMEGiHeIQCK AQgAAAAAABIAaG1ZA2ivYQNANiAAEwAAAEA2IABANiAAaK9hA2ivYQMo3hIA wRLyd0A2IAATAAAAaG1ZA0A2IAA43hIAZHYACAAAAAAlAAAADAAAAAUAAABG AAAAKAAAABwAAABHRElDAgAAAAAAAAAAAAAAMQEAAOUAAAAAAAAAIwAAACAA AACRnck+AAAAAAAAAAD8bMk+AAAxRAAAs0MkAAAAJAAAAJGdyT4AAAAAAAAA APxsyT4AADFEAACzQwQAAAAjAAAAIAAAALo08EIAAAAAAAAAAH0utEIAADFE AACzQyQAAAAkAAAAujTwQgAAAAAAAAAAfS60QgAAMUQAALNDBAAAACMAAAAg AAAAkZ3JPgAAAAAAAAAA/GzJPgAAMUQAALNDJAAAACQAAACRnck+AAAAAAAA AAD8bMk+AAAxRAAAs0MEAAAAIwAAACAAAACNkHQ8AAAAAAAAAABTvnQ8AAAx RAAAs0MkAAAAJAAAAI2QdDwAAAAAAAAAAFO+dDwAADFEAACzQwQAAAAjAAAA IAAAAI2QdDwAAAAAAAAAAFO+dDwAADFEAACzQyQAAAAkAAAAjZB0PAAAAAAA AAAAU750PAAAMUQAALNDBAAAACMAAAAgAAAAjZB0PAAAAAAAAAAAU750PAAA MUQAALNDJAAAACQAAACNkHQ8AAAAAAAAAABTvnQ8AAAxRAAAs0MEAAAAIQAA AAgAAAAUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAmAAAAHAAAAAcAAAAAAAAA AAAAAAAAAAAAAAAAJQAAAAwAAAAHAAAAJgAAABwAAAAGAAAABgAAAEEAAAAA AAAAAAAAACUAAAAMAAAABgAAACUAAAAMAAAABQAAgCcAAAAYAAAAAgAAAAAA AAD///8AAAAAACUAAAAMAAAAAgAAABMAAAAMAAAAAQAAAFYAAACgAQAAxAIA AGYBAAA8AwAAwAEAAGEAAACtBHMRJQWpEsAFxhNgBsUUIwehFfAHVRbMCNcW qAk2F5gKbRd+C20XXwxKF08N+hYmDm4W7g66FbYP3hRqELoVRhFuFhMS+hb5 EkoX7hNtF8UUbRemFTYXjBbXFm0XVRY1GKEV8xjFFJ0ZxhMuGqkSuhpzEVUb mxEEHIwRqRxfEU4d+xDkHWoQVx6xD8Ue0A4BH9sNOB/XDEwfyQs4H7EKAR+o CcUeuAhXHuYH5B0oB04dnAapHCkGBBz3BVUb7QW6GiQGLhrkBJ0ZvQPzGLkC NRjsAW0XLgGMFrEAphVNAMUUIADuEyAA+RI+ABMSnQBGESQBahDOAbYPoALu Ds4BJg4kAU8NnQBfDD4AfgsgAJgKIACoCU0AzAixAPAHLgEjB+wBYAa5AsAF vQMlBeQErQQkBgME7QVZA/cFtAIpBhQCnAaDASgHCwHmB50AuAhNAKgJJQCx CiAAyQslANcMTQDbDZ0A0A4LAbEPgwFqEBQC+xC0Al8RWQOMEQMEmxGtBHMR JQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAAAKABAADEAgAAZgEAADwDAADA AQAAYQAAAK0EcxElBakSwAXGE2AGxRQjB6EV8AdVFswI1xaoCTYXmAptF34L bRdfDEoXTw36FiYObhbuDroVtg/eFGoQuhVGEW4WExL6FvkSShfuE20XxRRt F6YVNheMFtcWbRdVFjUYoRXzGMUUnRnGEy4aqRK6GnMRVRubEQQcjBGpHF8R Th37EOQdahBXHrEPxR7QDgEf2w04H9cMTB/JCzgfsQoBH6gJxR64CFce5gfk HSgHTh2cBqkcKQYEHPcFVRvtBboaJAYuGuQEnRm9A/MYuQI1GOwBbRcuAYwW sQCmFU0AxRQgAO4TIAD5Ej4AExKdAEYRJAFqEM4Btg+gAu4OzgEmDiQBTw2d AF8MPgB+CyAAmAogAKgJTQDMCLEA8AcuASMH7AFgBrkCwAW9AyUF5AStBCQG AwTtBVkD9wW0AikGFAKcBoMBKAcLAeYHnQC4CE0AqAklALEKIADJCyUA1wxN ANsNnQDQDgsBsQ+DAWoQFAL7ELQCXxFZA4wRAwSbEa0EcxEUAAAADAAAAA0A AAAmAAAAHAAAAAQAAAAGAAAAQQAAAAAAAAD/AAAAJQAAAAwAAAAEAAAAKAAA AAwAAAAGAAAAVwAAACwAAADjAgAAhQEAAB0DAACnAQAABAAAAGYP7BA2CCwI NhcsCGYP7BAUAAAADAAAAA0AAAAmAAAAHAAAAAYAAAAGAAAAHgAAAAAAAAAA AAAAJQAAAAwAAAAGAAAAKAAAAAwAAAAEAAAAJQAAAAwAAAAFAACAJwAAABgA AAAEAAAAAAAAAH9/fwAAAAAAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAA ADAAAADcAgAAiwEAAO4CAACNAQAABQAAAGoGNArxBrIJ1AqyCU0KNApqBjQK JQAAAAwAAAAFAACAKAAAAAwAAAAEAAAAFAAAAAwAAAANAAAAVwAAADAAAADc AgAAiwEAAO4CAACNAQAABQAAAGoGNArxBrIJ1AqyCU0KNApqBjQKFAAAAAwA AAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAL+/vwAAAAAAJQAA AAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADcAgAAjAEAAOwCAACQAQAA BQAAAGoGLwpSCi8KUgrtCmoG7QpqBi8KJQAAAAwAAAAFAACAKAAAAAwAAAAE AAAAFAAAAAwAAAANAAAAVwAAADAAAADcAgAAjAEAAOwCAACQAQAABQAAAGoG LwpSCi8KUgrtCmoG7QpqBi8KFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAA ABgAAAAEAAAAAAAAAKWlpQAAAAAAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAA VgAAADAAAADrAgAAiwEAAO4CAACQAQAABQAAAFIK8grUCmYK1AqyCVIKNApS CvIKJQAAAAwAAAAFAACAKAAAAAwAAAAEAAAAFAAAAAwAAAANAAAAVwAAADAA AADrAgAAiwEAAO4CAACQAQAABQAAAFIK8grUCmYK1AqyCVIKNApSCvIKFAAA AAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAAAAAAANjY2AAAAAAA JQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADcAgAAfwEAAO0CAACB AQAABQAAAG8G8QbYBo0GuwqNBlIK8QZvBvEGJQAAAAwAAAAFAACAKAAAAAwA AAAEAAAAFAAAAAwAAAANAAAAVwAAADAAAADcAgAAfwEAAO0CAACBAQAABQAA AG8G8QbYBo0GuwqNBlIK8QZvBvEGFAAAAAwAAAANAAAAJQAAAAwAAAAFAACA JwAAABgAAAAEAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAEAAAAEwAAAAwAAAAB AAAAVgAAADAAAADcAgAAgAEAAOwCAACNAQAABQAAAGoGDApSCgwKUgrsBmoG 7AZqBgwKJQAAAAwAAAAFAACAKAAAAAwAAAAEAAAAFAAAAAwAAAANAAAAVwAA ADAAAADcAgAAgAEAAOwCAACNAQAABQAAAGoGDApSCgwKUgrsBmoG7AZqBgwK FAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAKWlpQAA AAAAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADrAgAAfwEAAO0C AACNAQAABQAAALsKiAZNCvEGTQoMCrsKowm7CogGJQAAAAwAAAAFAACAKAAA AAwAAAAEAAAAFAAAAAwAAAANAAAAVwAAADAAAADrAgAAfwEAAO0CAACNAQAA BQAAALsKiAZNCvEGTQoMCrsKowm7CogGFAAAAAwAAAANAAAAJQAAAAwAAAAF AACAJwAAABgAAAAEAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAEAAAAEwAAAAwA AAABAAAAVgAAADAAAADYAgAAkAEAAOwCAACUAQAABQAAAHoF2AsaBhALXAoQ C7wJ2At6BdgLJQAAAAwAAAAFAACAKAAAAAwAAAAEAAAAFAAAAAwAAAANAAAA VwAAADAAAADYAgAAkAEAAOwCAACUAQAABQAAAHoF2AsaBhALXAoQC7wJ2At6 BdgLFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAL+/ vwAAAAAAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADYAgAAkAEA AOwCAACUAQAABQAAAHoF2AsaBhALXAoQC7wJ2At6BdgLJQAAAAwAAAAFAACA KAAAAAwAAAAEAAAAFAAAAAwAAAANAAAAVwAAADAAAADYAgAAkAEAAOwCAACU AQAABQAAAHoF2AsaBhALXAoQC7wJ2At6BdgLFAAAAAwAAAANAAAAJQAAAAwA AAAFAACAJwAAABgAAAAEAAAAAAAAAKWlpQAAAAAAJQAAAAwAAAAEAAAAEwAA AAwAAAABAAAAVgAAADAAAADpAgAAkAEAAOwCAACUAQAABQAAAMEJAAxXClYL VwoVC8EJ0wvBCQAMJQAAAAwAAAAFAACAKAAAAAwAAAAEAAAAFAAAAAwAAAAN AAAAVwAAADAAAADpAgAAkAEAAOwCAACUAQAABQAAAMEJAAxXClYLVwoVC8EJ 0wvBCQAMFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAAAAAA AKWlpQAAAAAAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADpAgAA kAEAAOwCAACUAQAABQAAAMEJAAxXClYLVwoVC8EJ0wvBCQAMJQAAAAwAAAAF AACAKAAAAAwAAAAEAAAAFAAAAAwAAAANAAAAVwAAADAAAADpAgAAkAEAAOwC AACUAQAABQAAAMEJAAxXClYLVwoVC8EJ0wvBCQAMFAAAAAwAAAANAAAAJQAA AAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAL+/vwAAAAAAJQAAAAwAAAAEAAAA EwAAAAwAAAABAAAAVgAAADAAAADZAgAAkwEAAOoCAACUAQAABQAAAH8F2AvB CdgLwQkADH8FAAx/BdgLJQAAAAwAAAAFAACAKAAAAAwAAAAEAAAAFAAAAAwA AAANAAAAVwAAADAAAADZAgAAkwEAAOoCAACUAQAABQAAAH8F2AvBCdgLwQkA DH8FAAx/BdgLFAAAAAwAAAANAAAAJQAAAAwAAAAFAACAJwAAABgAAAAEAAAA AAAAAL+/vwAAAAAAJQAAAAwAAAAEAAAAEwAAAAwAAAABAAAAVgAAADAAAADZ AgAAkwEAAOoCAACUAQAABQAAAH8F2AvBCdgLwQkADH8FAAx/BdgLJQAAAAwA AAAFAACAKAAAAAwAAAAEAAAAFAAAAAwAAAANAAAAVwAAADAAAADZAgAAkwEA AOoCAACUAQAABQAAAH8F2AvBCdgLwQkADH8FAAx/BdgLFAAAAAwAAAANAAAA JQAAAAwAAAAFAACAJwAAABgAAAAEAAAAAAAAAAAA/wAAAAAAJQAAAAwAAAAE AAAAEwAAAAwAAAABAAAAVgAAADAAAADdAgAAgQEAAOoCAACLAQAABQAAAPMJ UAfJBlAHyQaoCfMJqAnzCVAHJQAAAAwAAAAFAACAFAAAAAwAAAANAAAAVwAA ADAAAADdAgAAgQEAAOoCAACLAQAABQAAAPMJUAfJBlAHyQaoCfMJqAnzCVAH FAAAAAwAAAANAAAAGwAAABAAAAAcCQAAmAoAADYAAAAQAAAA5AkAAJgKAAAU AAAADAAAAA0AAAAmAAAAHAAAAAMAAAAGAAAAQQAAAAAAAAAAAAAAJQAAAAwA AAADAAAAKAAAAAwAAAAGAAAAJQAAAAwAAAAFAACAJQAAAAwAAAACAAAAEwAA AAwAAAABAAAAVgAAAKABAADfAgAAggEAAOkCAACLAQAAYQAAAG4H9AhzBw0J hwcmCZYHPwmgB04JtAdiCcMHZwnXB3EJ6wdxCf8HcQkTCHEJIghnCTYIYglF CE4JWQg/CWgITgl8CGIJiwhnCZoIcQmzCHEJxwhxCdEIcQnqCGcJ/ghiCQ0J TgkcCT8JKwkmCToJDQlECfQITgn0CF0J9AhsCfQIdgnqCIUJ4AiUCcwImQm9 CJ4JqQijCZAIowl8CKMJaAieCU8ImQk7CJQJLAiFCRgIdgkOCGwJBAhdCQQI TgkECEQJBAg6CesHKwnSBxwJuQcNCaoH/giWB+oIkQfRCIcHxwiHB7MIhwea CIcHiwiRB3wIlgdoCKoHWQi5B0UIqgc2CJYHIgiRBxMIhwf/B4cH6weHB9cH hwfDB5EHtAeWB6AHqgeWB7kHhwfSB3MH6wduBwQIXwcECFUHBAhGBwQIPAcO CC0HGAgeBywIGQc7CA8HTwgPB2gIDwd8CA8HkAgPB6kIGQe9CB4HzAgtB+AI PAfqCEYH9AhVB/QIXwf0CG4H9AglAAAADAAAAAUAAIAUAAAADAAAAA0AAABX AAAAoAEAAN8CAACCAQAA6QIAAIsBAABhAAAAbgf0CHMHDQmHByYJlgc/CaAH Tgm0B2IJwwdnCdcHcQnrB3EJ/wdxCRMIcQkiCGcJNghiCUUITglZCD8JaAhO CXwIYgmLCGcJmghxCbMIcQnHCHEJ0QhxCeoIZwn+CGIJDQlOCRwJPwkrCSYJ OgkNCUQJ9AhOCfQIXQn0CGwJ9Ah2CeoIhQngCJQJzAiZCb0IngmpCKMJkAij CXwIowloCJ4JTwiZCTsIlAksCIUJGAh2CQ4IbAkECF0JBAhOCQQIRAkECDoJ 6wcrCdIHHAm5Bw0Jqgf+CJYH6giRB9EIhwfHCIcHswiHB5oIhweLCJEHfAiW B2gIqgdZCLkHRQiqBzYIlgciCJEHEwiHB/8HhwfrB4cH1weHB8MHkQe0B5YH oAeqB5YHuQeHB9IHcwfrB24HBAhfBwQIVQcECEYHBAg8Bw4ILQcYCB4HLAgZ BzsIDwdPCA8HaAgPB3wIDweQCA8HqQgZB70IHgfMCC0H4Ag8B+oIRgf0CFUH 9AhfB/QIbgf0CBQAAAAMAAAADQAAACYAAAAcAAAABgAAAAYAAAAeAAAAAAAA AAAAAAAlAAAADAAAAAYAAAAoAAAADAAAAAMAAAAlAAAADAAAAAUAAIAnAAAA GAAAAAMAAAAAAAAAf39/AAAAAAAlAAAADAAAAAMAAAATAAAADAAAAAEAAABW AAAAMAAAABMDAACLAQAAJQMAAI0BAAAFAAAAyhQ0ClEVsgk0GbIJrRg0CsoU NAolAAAADAAAAAUAAIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAMAAA ABMDAACLAQAAJQMAAI0BAAAFAAAAyhQ0ClEVsgk0GbIJrRg0CsoUNAoUAAAA DAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAAv7+/AAAAAAAl AAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAMAAAABMDAACMAQAAIwMAAJAB AAAFAAAAyhQvCrIYLwqyGO0KyhTtCsoULwolAAAADAAAAAUAAIAoAAAADAAA AAMAAAAUAAAADAAAAA0AAABXAAAAMAAAABMDAACMAQAAIwMAAJABAAAFAAAA yhQvCrIYLwqyGO0KyhTtCsoULwoUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAn AAAAGAAAAAMAAAAAAAAApaWlAAAAAAAlAAAADAAAAAMAAAATAAAADAAAAAEA AABWAAAAMAAAACIDAACLAQAAJQMAAJABAAAFAAAAshjyCjQZZgo0GbIJshg0 CrIY8golAAAADAAAAAUAAIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAA MAAAACIDAACLAQAAJQMAAJABAAAFAAAAshjyCjQZZgo0GbIJshg0CrIY8goU AAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAA2NjYAAAA AAAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAMAAAABMDAAB/AQAAJAMA AIEBAAAFAAAAzxTxBjgVjQYbGY0GshjxBs8U8QYlAAAADAAAAAUAAIAoAAAA DAAAAAMAAAAUAAAADAAAAA0AAABXAAAAMAAAABMDAAB/AQAAJAMAAIEBAAAF AAAAzxTxBjgVjQYbGY0GshjxBs8U8QYUAAAADAAAAA0AAAAlAAAADAAAAAUA AIAnAAAAGAAAAAMAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAMAAAATAAAADAAA AAEAAABWAAAAMAAAABMDAACAAQAAIwMAAI0BAAAFAAAAyhQMCrIYDAqyGOwG yhTsBsoUDAolAAAADAAAAAUAAIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABX AAAAMAAAABMDAACAAQAAIwMAAI0BAAAFAAAAyhQMCrIYDAqyGOwGyhTsBsoU DAoUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAApaWl AAAAAAAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAMAAAACIDAAB/AQAA JAMAAI0BAAAFAAAAGxmIBq0Y8QatGAwKGxmjCRsZiAYlAAAADAAAAAUAAIAo AAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAMAAAACIDAAB/AQAAJAMAAI0B AAAFAAAAGxmIBq0Y8QatGAwKGxmjCRsZiAYUAAAADAAAAA0AAAAlAAAADAAA AAUAAIAnAAAAGAAAAAMAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAMAAAATAAAA DAAAAAEAAABWAAAAMAAAAA8DAACQAQAAIwMAAJQBAAAFAAAA2hPYC3oUEAu8 GBALHBjYC9oT2AslAAAADAAAAAUAAIAoAAAADAAAAAMAAAAUAAAADAAAAA0A AABXAAAAMAAAAA8DAACQAQAAIwMAAJQBAAAFAAAA2hPYC3oUEAu8GBALHBjY C9oT2AsUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAA v7+/AAAAAAAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAMAAAAA8DAACQ AQAAIwMAAJQBAAAFAAAA2hPYC3oUEAu8GBALHBjYC9oT2AslAAAADAAAAAUA AIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAMAAAAA8DAACQAQAAIwMA AJQBAAAFAAAA2hPYC3oUEAu8GBALHBjYC9oT2AsUAAAADAAAAA0AAAAlAAAA DAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAApaWlAAAAAAAlAAAADAAAAAMAAAAT AAAADAAAAAEAAABWAAAAMAAAACADAACQAQAAIwMAAJQBAAAFAAAAIRgADLcY Vgu3GBULIRjTCyEYAAwlAAAADAAAAAUAAIAoAAAADAAAAAMAAAAUAAAADAAA AA0AAABXAAAAMAAAACADAACQAQAAIwMAAJQBAAAFAAAAIRgADLcYVgu3GBUL IRjTCyEYAAwUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMAAAAA AAAApaWlAAAAAAAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAMAAAACAD AACQAQAAIwMAAJQBAAAFAAAAIRgADLcYVgu3GBULIRjTCyEYAAwlAAAADAAA AAUAAIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAMAAAACADAACQAQAA IwMAAJQBAAAFAAAAIRgADLcYVgu3GBULIRjTCyEYAAwUAAAADAAAAA0AAAAl AAAADAAAAAUAAIAnAAAAGAAAAAMAAAAAAAAAv7+/AAAAAAAlAAAADAAAAAMA AAATAAAADAAAAAEAAABWAAAAMAAAAA8DAACTAQAAIQMAAJQBAAAFAAAA3xPY CyEY2AshGAAM3xMADN8T2AslAAAADAAAAAUAAIAoAAAADAAAAAMAAAAUAAAA DAAAAA0AAABXAAAAMAAAAA8DAACTAQAAIQMAAJQBAAAFAAAA3xPYCyEY2Ash GAAM3xMADN8T2AsUAAAADAAAAA0AAAAlAAAADAAAAAUAAIAnAAAAGAAAAAMA AAAAAAAAv7+/AAAAAAAlAAAADAAAAAMAAAATAAAADAAAAAEAAABWAAAAMAAA AA8DAACTAQAAIQMAAJQBAAAFAAAA3xPYCyEY2AshGAAM3xMADN8T2AslAAAA DAAAAAUAAIAoAAAADAAAAAMAAAAUAAAADAAAAA0AAABXAAAAMAAAAA8DAACT AQAAIQMAAJQBAAAFAAAA3xPYCyEY2AshGAAM3xMADN8T2AsUAAAADAAAAA0A AAAlAAAADAAAAAUAAIAlAAAADAAAAAQAAAATAAAADAAAAAEAAABWAAAAMAAA ABQDAACBAQAAIQMAAIsBAAAFAAAAUxhQBykVUAcpFagJUxioCVMYUAclAAAA DAAAAAUAAIAUAAAADAAAAA0AAABXAAAAMAAAABQDAACBAQAAIQMAAIsBAAAF AAAAUxhQBykVUAcpFagJUxioCVMYUAcUAAAADAAAAA0AAAAbAAAAEAAAAHwX AACYCgAANgAAABAAAABEGAAAmAoAABQAAAAMAAAADQAAACYAAAAcAAAAAwAA AAYAAABBAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAoAAAADAAAAAYAAAAlAAAA DAAAAAUAAIAlAAAADAAAAAIAAAATAAAADAAAAAEAAABWAAAAoAEAABUDAACC AQAAIAMAAIsBAABhAAAAzhX0CNMVDQnnFSYJ9hU/CQAWTgkUFmIJIxZnCTcW cQlLFnEJXxZxCXMWcQmCFmcJlhZiCaUWTgm5Fj8JyBZOCdwWYgnrFmcJ+hZx CRMXcQknF3EJMRdxCUoXZwleF2IJbRdOCXwXPwmLFyYJmhcNCaQX9AiuF/QI vRf0CMwX9AjWF+oI5RfgCPQXzAj5F70I/hepCAMYkAgDGHwIAxhoCP4XTwj5 FzsI9BcsCOUXGAjWFw4IzBcECL0XBAiuFwQIpBcECJoX6weLF9IHfBe5B20X qgdeF5YHSheRBzEXhwcnF4cHExeHB/oWhwfrFpEH3BaWB8gWqge5FrkHpRaq B5YWlgeCFpEHcxaHB18WhwdLFocHNxaHByMWkQcUFpYHABaqB/YVuQfnFdIH 0xXrB84VBAi/FQQItRUECKYVBAicFQ4IjRUYCH4VLAh5FTsIbxVPCG8VaAhv FXwIbxWQCG8VqQh5Fb0IfhXMCI0V4AicFeoIphX0CLUV9Ai/FfQIzhX0CCUA AAAMAAAABQAAgBQAAAAMAAAADQAAAFcAAACgAQAAFQMAAIIBAAAgAwAAiwEA AGEAAADOFfQI0xUNCecVJgn2FT8JABZOCRQWYgkjFmcJNxZxCUsWcQlfFnEJ cxZxCYIWZwmWFmIJpRZOCbkWPwnIFk4J3BZiCesWZwn6FnEJExdxCScXcQkx F3EJShdnCV4XYgltF04JfBc/CYsXJgmaFw0JpBf0CK4X9Ai9F/QIzBf0CNYX 6gjlF+AI9BfMCPkXvQj+F6kIAxiQCAMYfAgDGGgI/hdPCPkXOwj0FywI5RcY CNYXDgjMFwQIvRcECK4XBAikFwQImhfrB4sX0gd8F7kHbReqB14XlgdKF5EH MReHBycXhwcTF4cH+haHB+sWkQfcFpYHyBaqB7kWuQelFqoHlhaWB4IWkQdz FocHXxaHB0sWhwc3FocHIxaRBxQWlgcAFqoH9hW5B+cV0gfTFesHzhUECL8V BAi1FQQIphUECJwVDgiNFRgIfhUsCHkVOwhvFU8IbxVoCG8VfAhvFZAIbxWp CHkVvQh+FcwIjRXgCJwV6gimFfQItRX0CL8V9AjOFfQIFAAAAAwAAAANAAAA JgAAABwAAAAGAAAABgAAAB4AAAAAAAAAAAAAACUAAAAMAAAABgAAACgAAAAM AAAAAwAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAB/f38AAAAAACUA AAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAAwAAAA9gIAAKcBAAAIAwAAqgEA AAUAAABKDbQR0Q0yEbQRMhEtEbQRSg20ESUAAAAMAAAABQAAgCgAAAAMAAAA AwAAABQAAAAMAAAADQAAAFcAAAAwAAAA9gIAAKcBAAAIAwAAqgEAAAUAAABK DbQR0Q0yEbQRMhEtEbQRSg20ERQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcA AAAYAAAAAwAAAAAAAAC/v78AAAAAACUAAAAMAAAAAwAAABMAAAAMAAAAAQAA AFYAAAAwAAAA9gIAAKkBAAAGAwAArQEAAAUAAABKDa8RMhGvETIRbRJKDW0S Sg2vESUAAAAMAAAABQAAgCgAAAAMAAAAAwAAABQAAAAMAAAADQAAAFcAAAAw AAAA9gIAAKkBAAAGAwAArQEAAAUAAABKDa8RMhGvETIRbRJKDW0SSg2vERQA AAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAClpaUAAAAA ACUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAAwAAAABQMAAKcBAAAIAwAA rQEAAAUAAAAyEXIStBHmEbQRMhEyEbQRMhFyEiUAAAAMAAAABQAAgCgAAAAM AAAAAwAAABQAAAAMAAAADQAAAFcAAAAwAAAABQMAAKcBAAAIAwAArQEAAAUA AAAyEXIStBHmEbQRMhEyEbQRMhFyEhQAAAAMAAAADQAAACUAAAAMAAAABQAA gCcAAAAYAAAAAwAAAAAAAADY2NgAAAAAACUAAAAMAAAAAwAAABMAAAAMAAAA AQAAAFYAAAAwAAAA9gIAAJsBAAAIAwAAngEAAAUAAABPDXEOuA0NDpsRDQ4y EXEOTw1xDiUAAAAMAAAABQAAgCgAAAAMAAAAAwAAABQAAAAMAAAADQAAAFcA AAAwAAAA9gIAAJsBAAAIAwAAngEAAAUAAABPDXEOuA0NDpsRDQ4yEXEOTw1x DhQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAC/v78A AAAAACUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAAwAAAA9gIAAJ0BAAAG AwAAqgEAAAUAAABKDYwRMhGMETIRbA5KDWwOSg2MESUAAAAMAAAABQAAgCgA AAAMAAAAAwAAABQAAAAMAAAADQAAAFcAAAAwAAAA9gIAAJ0BAAAGAwAAqgEA AAUAAABKDYwRMhGMETIRbA5KDWwOSg2MERQAAAAMAAAADQAAACUAAAAMAAAA BQAAgCcAAAAYAAAAAwAAAAAAAAClpaUAAAAAACUAAAAMAAAAAwAAABMAAAAM AAAAAQAAAFYAAAAwAAAABQMAAJsBAAAIAwAAqgEAAAUAAACbEQgOLRFxDi0R jBGbESMRmxEIDiUAAAAMAAAABQAAgCgAAAAMAAAAAwAAABQAAAAMAAAADQAA AFcAAAAwAAAABQMAAJsBAAAIAwAAqgEAAAUAAACbEQgOLRFxDi0RjBGbESMR mxEIDhQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAC/ v78AAAAAACUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAAwAAAA8wIAAK0B AAAGAwAAsAEAAAUAAABaDFgT+gyQEjwRkBKcEFgTWgxYEyUAAAAMAAAABQAA gCgAAAAMAAAAAwAAABQAAAAMAAAADQAAAFcAAAAwAAAA8wIAAK0BAAAGAwAA sAEAAAUAAABaDFgT+gyQEjwRkBKcEFgTWgxYExQAAAAMAAAADQAAACUAAAAM AAAABQAAgCcAAAAYAAAAAwAAAAAAAAC/v78AAAAAACUAAAAMAAAAAwAAABMA AAAMAAAAAQAAAFYAAAAwAAAA8wIAAK0BAAAGAwAAsAEAAAUAAABaDFgT+gyQ EjwRkBKcEFgTWgxYEyUAAAAMAAAABQAAgCgAAAAMAAAAAwAAABQAAAAMAAAA DQAAAFcAAAAwAAAA8wIAAK0BAAAGAwAAsAEAAAUAAABaDFgT+gyQEjwRkBKc EFgTWgxYExQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAA AAClpaUAAAAAACUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAAwAAAAAwMA AK0BAAAGAwAAsQEAAAUAAAChEIATNxHWEjcRlRKhEFMToRCAEyUAAAAMAAAA BQAAgCgAAAAMAAAAAwAAABQAAAAMAAAADQAAAFcAAAAwAAAAAwMAAK0BAAAG AwAAsQEAAAUAAAChEIATNxHWEjcRlRKhEFMToRCAExQAAAAMAAAADQAAACUA AAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAClpaUAAAAAACUAAAAMAAAAAwAA ABMAAAAMAAAAAQAAAFYAAAAwAAAAAwMAAK0BAAAGAwAAsQEAAAUAAAChEIAT NxHWEjcRlRKhEFMToRCAEyUAAAAMAAAABQAAgCgAAAAMAAAAAwAAABQAAAAM AAAADQAAAFcAAAAwAAAAAwMAAK0BAAAGAwAAsQEAAAUAAAChEIATNxHWEjcR lRKhEFMToRCAExQAAAAMAAAADQAAACUAAAAMAAAABQAAgCcAAAAYAAAAAwAA AAAAAAC/v78AAAAAACUAAAAMAAAAAwAAABMAAAAMAAAAAQAAAFYAAAAwAAAA 8wIAALABAAAEAwAAsQEAAAUAAABfDFgToRBYE6EQgBNfDIATXwxYEyUAAAAM AAAABQAAgCgAAAAMAAAAAwAAABQAAAAMAAAADQAAAFcAAAAwAAAA8wIAALAB AAAEAwAAsQEAAAUAAABfDFgToRBYE6EQgBNfDIATXwxYExQAAAAMAAAADQAA ACUAAAAMAAAABQAAgCcAAAAYAAAAAwAAAAAAAAC/v78AAAAAACUAAAAMAAAA AwAAABMAAAAMAAAAAQAAAFYAAAAwAAAA8wIAALABAAAEAwAAsQEAAAUAAABf DFgToRBYE6EQgBNfDIATXwxYEyUAAAAMAAAABQAAgCgAAAAMAAAAAwAAABQA AAAMAAAADQAAAFcAAAAwAAAA8wIAALABAAAEAwAAsQEAAAUAAABfDFgToRBY E6EQgBNfDIATXwxYExQAAAAMAAAADQAAACUAAAAMAAAABQAAgCUAAAAMAAAA BAAAABMAAAAMAAAAAQAAAFYAAAAwAAAA+AIAAJ4BAAAFAwAAqAEAAAUAAADT ENAOqQ3QDqkNKBHTECgR0xDQDiUAAAAMAAAABQAAgBQAAAAMAAAADQAAAFcA AAAwAAAA+AIAAJ4BAAAFAwAAqAEAAAUAAADTENAOqQ3QDqkNKBHTECgR0xDQ DhQAAAAMAAAADQAAABsAAAAQAAAA/A8AABgSAAA2AAAAEAAAAMQQAAAYEgAA FAAAAAwAAAANAAAAJgAAABwAAAADAAAABgAAAEEAAAAAAAAAAAAAACUAAAAM AAAAAwAAACgAAAAMAAAABgAAACUAAAAMAAAABQAAgCUAAAAMAAAAAgAAABMA AAAMAAAAAQAAAFYAAACgAQAA+QIAAJ8BAAAEAwAApwEAAGEAAABODnQQUw6N EGcOphB2Dr8QgA7OEJQO4hCjDucQtw7xEMsO8RDfDvEQ8w7xEAIP5xAWD+IQ JQ/OEDkPvxBID84QXA/iEGsP5xB6D/EQkw/xEKcP8RCxD/EQyg/nEN4P4hDt D84Q/A+/EAsQphAaEI0QJBB0EC4QdBA9EHQQTBB0EFYQahBlEGAQdBBMEHkQ PRB+ECkQgxAQEIMQ/A+DEOgPfhDPD3kQuw90EKwPZRCYD1YQjg9MEIQPPRCE Dy4QhA8kEIQPGhBrDwsQUg/8DzkP7Q8qD94PFg/KDxEPsQ8HD6cPBw+TDwcP eg8HD2sPEQ9cDxYPSA8qDzkPOQ8lDyoPFg8WDwIPEQ/zDgcP3w4HD8sOBw+3 DgcPow4RD5QOFg+ADioPdg45D2cOUg9TDmsPTg6EDz8OhA81DoQPJg6EDxwO jg8NDpgP/g2sD/kNuw/vDc8P7w3oD+8N/A/vDRAQ7w0pEPkNPRD+DUwQDQ5g EBwOahAmDnQQNQ50ED8OdBBODnQQJQAAAAwAAAAFAACAFAAAAAwAAAANAAAA VwAAAKABAAD5AgAAnwEAAAQDAACnAQAAYQAAAE4OdBBTDo0QZw6mEHYOvxCA Ds4QlA7iEKMO5xC3DvEQyw7xEN8O8RDzDvEQAg/nEBYP4hAlD84QOQ+/EEgP zhBcD+IQaw/nEHoP8RCTD/EQpw/xELEP8RDKD+cQ3g/iEO0PzhD8D78QCxCm EBoQjRAkEHQQLhB0ED0QdBBMEHQQVhBqEGUQYBB0EEwQeRA9EH4QKRCDEBAQ gxD8D4MQ6A9+EM8PeRC7D3QQrA9lEJgPVhCOD0wQhA89EIQPLhCEDyQQhA8a EGsPCxBSD/wPOQ/tDyoP3g8WD8oPEQ+xDwcPpw8HD5MPBw96DwcPaw8RD1wP Fg9IDyoPOQ85DyUPKg8WDxYPAg8RD/MOBw/fDgcPyw4HD7cOBw+jDhEPlA4W D4AOKg92DjkPZw5SD1MOaw9ODoQPPw6EDzUOhA8mDoQPHA6ODw0OmA/+DawP +Q27D+8Nzw/vDegP7w38D+8NEBDvDSkQ+Q09EP4NTBANDmAQHA5qECYOdBA1 DnQQPw50EE4OdBAUAAAADAAAAA0AAAAiAAAADAAAAP////8oAAAADAAAAAcA AAAlAAAADAAAAAcAAIAlAAAADAAAAAAAAIBLAAAAEAAAAAAAAAAFAAAAKAAA AAwAAAACAAAAKAAAAAwAAAAEAAAAKAAAAAwAAAADAAAARgAAABQAAAAIAAAA R0RJQwMAAAAiAAAADAAAAP////8iAAAADAAAAP////8lAAAADAAAAA0AAIAo AAAADAAAAAUAAAAiAAAADAAAAP////9GAAAANAEAACgBAABFTUYrJkAAABAA AAAEAAAAAAAAACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEIK2wHBAJ6H QwhABQYwAAAAJAAAAAEQwNurqio+AAAAAAAAAAAAAAAABQAAAEEAUgBJAEEA TAAAADZABYDAAAAAtAAAAAAAAP8BAAAAAQAAAA4AAABTAHQAYQB0AGkAYwAg AE4AZQB0AHcAbwByAGsAd7cmPjMzMz5mRow+MzMzPrz7oz4zMzM+EXHTPjMz Mz5mJus+MzMzPrwb/j4zMzM+M2MUPzMzMz7ePSA/MzMzPt4NPz8zMzM+ichW PzMzMz4zo2I/MzMzPpq5gD8zMzM+75aMPzMzMz6asZM/MzMzPgAAgD8AAAAA AAAAAAAAgD8AAAAAAAAAABIAAAAMAAAAAQAAABgAAAAMAAAAAAAAABYAAAAM AAAAGAAAACUAAAAMAAAAAQAAAFQAAACgAAAACAAAABEBAABuAAAAIgEAAAEA AAAAQNNBAPDSQQgAAAAgAQAADgAAAEwAAAAAAAAAAAAAAAAAAAD///////// /2gAAABTAHQAYQB0AGkAYwAgAE4AZQB0AHcAbwByAGsACwAAAAQAAAAJAAAA BAAAAAQAAAAIAAAABAAAAAwAAAAJAAAABAAAAAwAAAAJAAAABQAAAAAAAAAl AAAADAAAAA0AAIBGAAAAxAAAALgAAABFTUYrK0AAAAwAAAAAAAAAKkAAACQA AAAYAAAA8mDAQgAAAAAAAAAA6KnAQgrbAcEAnodDNkAFgIQAAAB4AAAAAAAA /wEAAAABAAAACAAAADEAMAAuADEALgAxAC4AMAARMcw+AADAPmam+z4AAMA+ 3o0VPwAAwD6JaCE/AADAPjMjOT8AAMA+3v1EPwAAwD6JuFw/AADAPjOTaD8A AMA+AACAPwAAAAAAAAAAAACAPwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwA AAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAAAHwAAAAeAAAAJAEA AFUAAAA1AQAAAQAAAABA00EA8NJBHgAAADMBAAAIAAAATAAAAAAAAAAAAAAA AAAAAP//////////XAAAADEAMAAuADEALgAxAC4AMAAJAAAACQAAAAQAAAAJ AAAABAAAAAkAAAAEAAAAAAAAACUAAAAMAAAADQAAgEYAAAAwAQAAJAEAAEVN RisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADoqcBCBLUo RDSZk0MIQAUGMAAAACQAAAABEMDbq6oqPgAAAAAAAAAAAAAAAAUAAABBAFIA SQBBAEwAAAA2QAWAwAAAALQAAAAAAAD/AQAAAAEAAAAOAAAAUwB0AGEAdABp AGMAIABOAGUAdAB3AG8AcgBrAHe3Jj5I4Ro+ZkaMPkjhGj68+6M+SOEaPhFx 0z5I4Ro+ZibrPkjhGj68G/4+SOEaPjNjFD9I4Ro+3j0gP0jhGj7eDT8/SOEa PonIVj9I4Ro+M6NiP0jhGj6auYA/SOEaPu+WjD9I4Ro+mrGTP0jhGj4AAIA/ AAAAAAAAAAAAAIA/AAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAW AAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAAoAAAALICAAAnAQAAGAMAADgB AAABAAAAAEDTQQDw0kGyAgAANgEAAA4AAABMAAAAAAAAAAAAAAAAAAAA//// //////9oAAAAUwB0AGEAdABpAGMAIABOAGUAdAB3AG8AcgBrAAsAAAAEAAAA CQAAAAQAAAAEAAAACAAAAAQAAAAMAAAACQAAAAQAAAAMAAAACQAAAAUAAAAA AAAAJQAAAAwAAAANAACARgAAAMQAAAC4AAAARU1GKytAAAAMAAAAAAAAACpA AAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEIEtShENJmTQzZABYCEAAAAeAAA AAAAAP8BAAAAAQAAAAgAAAAxADAALgAxAC4AMQAuADAAETHMPgrXsz5mpvs+ CtezPt6NFT8K17M+iWghPwrXsz4zIzk/CtezPt79RD8K17M+ibhcPwrXsz4z k2g/CtezPgAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAABIAAAAMAAAAAQAAABgA AAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAAB8AAAAyQIA ADoBAAAAAwAASwEAAAEAAAAAQNNBAPDSQckCAABJAQAACAAAAEwAAAAAAAAA AAAAAAAAAAD//////////1wAAAAxADAALgAxAC4AMQAuADAACQAAAAkAAAAE AAAACQAAAAQAAAAJAAAABAAAAAAAAAAlAAAADAAAAA0AAIBGAAAAHAEAABAB AABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA6KnA Qj6ELURI/EZDCEAFBjAAAAAkAAAAARDA26uqKj4AAAAAAAAAAAAAAAAFAAAA QQBSAEkAQQBMAAAANkAFgKwAAACgAAAAAAAA/wEAAAABAAAADAAAAFIAZQBk AGkAcwB0AHIAaQBiAHUAdABlAEREUj6amZk9IsKmPpqZmT13N9Y+mpmZPWbW Aj+amZk9EVEMP5qZmT1mpiE/mpmZPRGBLT+amZk9ZrY7P5qZmT0RMUU/mpmZ PbzrXD+amZk9ZqZ0P5qZmT2JQIA/mpmZPQAAgD8AAAAAAAAAAAAAgD8AAAAA AAAAABIAAAAMAAAAAQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAM AAAAAQAAAFQAAACUAAAAygIAAL8AAAAeAwAA0AAAAAEAAAAAQNNBAPDSQcoC AADOAAAADAAAAEwAAAAAAAAAAAAAAAAAAAD//////////2QAAABSAGUAZABp AHMAdAByAGkAYgB1AHQAZQAMAAAACQAAAAkAAAAEAAAACAAAAAQAAAAFAAAA BAAAAAkAAAAJAAAABAAAAAAAAAAlAAAADAAAAA0AAIBGAAAAFAEAAAgBAABF TUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA6KnAQj6E LURI/EZDNkAFgNQAAADIAAAAAAAA/wEAAAABAAAAEAAAAHMAdABhAHQAaQBj ACAAaQBuAHQAbwAgAG8AcwBwAGYAiUjwPc3MjD6aeU0+zcyMPkTkfD7NzIw+ d+etPs3MjD7NnMU+zcyMPiKS2D7NzIw+Zp4BP83MjD4ReQ0/zcyMPrzzFj/N zIw+Zq4uP83MjD4RiTo/zcyMPrxDUj/NzIw+Zh5eP83MjD4R2XU/zcyMPjOX hT/NzIw+iXSRP83MjD4AAIA/AAAAAAAAAAAAAIA/AAAAAAAAAAASAAAADAAA AAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAA rAAAAMECAADSAAAAJQMAAOMAAAABAAAAAEDTQQDw0kHBAgAA4QAAABAAAABM AAAAAAAAAAAAAAAAAAAA//////////9sAAAAcwB0AGEAdABpAGMAIABpAG4A dABvACAAbwBzAHAAZgAIAAAABAAAAAkAAAAEAAAABAAAAAgAAAAEAAAABAAA AAkAAAAEAAAACQAAAAQAAAAJAAAACAAAAAkAAAAAAAAAJQAAAAwAAAANAACA RgAAAKgAAACcAAAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAAAPJgwEIA AAAAAAAAAOipwEI+hC1ESPxGQzZABYBoAAAAXAAAAAAAAP8BAAAAAQAAAAUA AABpAG4AIABSADMAIjLpPjMz8z53J/w+MzPzPmbOFT8zM/M+EakhPzMz8z4R eUA/MzPzPgAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAAAAAEgAAAAwAAAABAAAA GAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAAAGwAAADi AgAA5gAAAAYDAAD3AAAAAQAAAABA00EA8NJB4gIAAPUAAAAFAAAATAAAAAAA AAAAAAAAAAAAAP//////////WAAAAGkAbgAgAFIAMwAAAAQAAAAJAAAABAAA AAwAAAAAAAAAJQAAAAwAAAANAACARgAAABwBAAAQAQAARU1GKytAAAAMAAAA AAAAACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEIK2wHB2qIbQwhABQYw AAAAJAAAAAEQwNurqio+AAAAAAAAAAAAAAAABQAAAEEAUgBJAEEATAAAADZA BYCsAAAAoAAAAAAAAP8BAAAAAQAAAAwAAABSAGUAZABpAHMAdAByAGkAYgB1 AHQAZQBERFI+mpmZPSLCpj6amZk9dzfWPpqZmT1m1gI/mpmZPRFRDD+amZk9 ZqYhP5qZmT0RgS0/mpmZPWa2Oz+amZk9ETFFP5qZmT2861w/mpmZPWamdD+a mZk9iUCAP5qZmT0AAIA/AAAAAAAAAAAAAIA/AAAAAAAAAAASAAAADAAAAAEA AAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAAlAAA AAwAAACUAAAAYAAAAKUAAAABAAAAAEDTQQDw0kEMAAAAowAAAAwAAABMAAAA AAAAAAAAAAAAAAAA//////////9kAAAAUgBlAGQAaQBzAHQAcgBpAGIAdQB0 AGUADAAAAAkAAAAJAAAABAAAAAgAAAAEAAAABQAAAAQAAAAJAAAACQAAAAQA AAAAAAAAJQAAAAwAAAANAACARgAAABQBAAAIAQAARU1GKytAAAAMAAAAAAAA ACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEIK2wHB2qIbQzZABYDUAAAA yAAAAAAAAP8BAAAAAQAAABAAAABzAHQAYQB0AGkAYwAgAGkAbgB0AG8AIABv AHMAcABmAIlI8D3NzIw+mnlNPs3MjD5E5Hw+zcyMPnfnrT7NzIw+zZzFPs3M jD4iktg+zcyMPmaeAT/NzIw+EXkNP83MjD688xY/zcyMPmauLj/NzIw+EYk6 P83MjD68Q1I/zcyMPmYeXj/NzIw+Edl1P83MjD4zl4U/zcyMPol0kT/NzIw+ AACAPwAAAAAAAAAAAACAPwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAA AAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAAAKwAAAADAAAApwAAAGcA AAC4AAAAAQAAAABA00EA8NJBAwAAALYAAAAQAAAATAAAAAAAAAAAAAAAAAAA AP//////////bAAAAHMAdABhAHQAaQBjACAAaQBuAHQAbwAgAG8AcwBwAGYA CAAAAAQAAAAJAAAABAAAAAQAAAAIAAAABAAAAAQAAAAJAAAABAAAAAkAAAAE AAAACQAAAAgAAAAJAAAAAAAAACUAAAAMAAAADQAAgEYAAACoAAAAnAAAAEVN RisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADoqcBCCtsB wdqiG0M2QAWAaAAAAFwAAAAAAAD/AQAAAAEAAAAFAAAAaQBuACAAUgAyACIy 6T4zM/M+dyf8PjMz8z5mzhU/MzPzPhGpIT8zM/M+EXlAPzMz8z4AAIA/AAAA AAAAAAAAAIA/AAAAAAAAAAAAABIAAAAMAAAAAQAAABgAAAAMAAAAAAAAABYA AAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAABsAAAAJAAAALoAAABIAAAAywAA AAEAAAAAQNNBAPDSQSQAAADJAAAABQAAAEwAAAAAAAAAAAAAAAAAAAD///// /////1gAAABpAG4AIABSADIAAAAEAAAACQAAAAQAAAAMAAAAAAAAACUAAAAM AAAADQAAgEYAAACUAAAAiAAAAEVNRisrQAAADAAAAAAAAAAIQAMCPAAAADAA AAABEMDbAAAAAE4AAAAAAAAApz52PwIAAAACAAAAAgAAAAIAAAABEMDbAAAA AAAAAP8IQAQDLAAAACAAAAABEMDbAgAAAAAAAACRJbBCYZ5QQwoGHkOqeJhD AAEAABVABAAQAAAABAAAAAMAAAAkAAAAJAAAAAAAgD0AAAAAAAAAAAAAgD0A AAAAAAAAAAIAAABfAAAAOAAAAAUAAAA4AAAAAAAAADgAAAAAAAAAAAABAA8A AAAAAAAAAAAAAAAAAAAAAAAAAAAAACUAAAAMAAAABQAAACUAAAAMAAAABQAA gFcAAAAkAAAAVgAAAM8AAACgAAAAMwEAAAIAAACCBQoN4QkQEyUAAAAMAAAA BwAAgCUAAAAMAAAAAAAAgCQAAAAkAAAAAACAQQAAAAAAAAAAAACAQQAAAAAA AAAAAgAAACgAAAAMAAAABQAAAEYAAABcAAAAUAAAAEVNRisIQAQDPAAAADAA AAABEMDbBAAAAAAAAABD4yBDH8KWQ10GJUO3Sp1D8OIZQ9dOmUND4yBDH8KW QwABAYEUQASAEAAAAAQAAAAAAAD/JAAAACQAAAAAAIA9AAAAAAAAAAAAAIA9 AAAAAAAAAAACAAAAJwAAABgAAAAFAAAAAAAAAAAAAAAAAAAAJQAAAAwAAAAF AAAAEwAAAAwAAAABAAAAJQAAAAwAAAAIAACAVgAAACwAAACZAAAALQEAAKYA AAA7AQAABAAAAA8K2RJRCqoTnwkqEw8K2RIlAAAADAAAAAcAAIATAAAADAAA AAEAAAAlAAAADAAAAAAAAIAkAAAAJAAAAAAAgEEAAAAAAAAAAAAAgEEAAAAA AAAAAAIAAABGAAAAiAAAAHwAAABFTUYrCEADAjwAAAAwAAAAARDA2wAAAABO AAAAAAAAAKc+dj8CAAAAAgAAAAIAAAACAAAAARDA2wAAAAAAAAD/CEAEAywA AAAgAAAAARDA2wIAAAAAAAAAd1MyRF4deENfkR5EO22sQwABAAAVQAQAEAAA AAQAAAADAAAAJAAAACQAAAAAAIA9AAAAAAAAAAAAAIA9AAAAAAAAAAACAAAA XwAAADgAAAADAAAAOAAAAAAAAAA4AAAAAAAAAAAAAQAPAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAlAAAADAAAAAMAAAAlAAAADAAAAAUAAIBXAAAAJAAAAHgC AAD2AAAAywIAAFsBAAACAAAAlSyCD6UnjhUlAAAADAAAAAcAAIAlAAAADAAA AAAAAIAkAAAAJAAAAAAAgEEAAAAAAAAAAAAAgEEAAAAAAAAAAAIAAAAoAAAA DAAAAAMAAABGAAAAXAAAAFAAAABFTUYrCEAEAzwAAAAwAAAAARDA2wQAAAAA AAAAopMfRDxhrUP1rhxENQqxQ9LmHUSBoqpDopMfRDxhrUMAAQGBFEAEgBAA AAAEAAAAAAAA/yQAAAAkAAAAAACAPQAAAAAAAAAAAACAPQAAAAAAAAAAAgAA ACUAAAAMAAAABQAAABMAAAAMAAAAAQAAACUAAAAMAAAACAAAgFYAAAAsAAAA cgIAAFUBAAB/AgAAYwEAAAQAAADlJ60VLCciFnonVRXlJ60VJQAAAAwAAAAH AACAEwAAAAwAAAABAAAAJQAAAAwAAAAAAACAJAAAACQAAAAAAIBBAAAAAAAA AAAAAIBBAAAAAAAAAAACAAAARgAAAIgAAAB8AAAARU1GKwhAAwI8AAAAMAAA AAEQwNsAAAAATgAAAAAAAACnPnY/AgAAAAIAAAACAAAAAgAAAAEQwNsAAAAA AAAA/whABAMsAAAAIAAAAAEQwNsCAAAAAAAAAE2KOUQ5/KZDTYo5RIgBrUMA AQAAFUAEABAAAAAEAAAAAwAAACQAAAAkAAAAAACAPQAAAAAAAAAAAACAPQAA AAAAAAAAAgAAAF8AAAA4AAAAAwAAADgAAAAAAAAAOAAAAAAAAAAAAAEADwAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAJQAAAAwAAAADAAAAJQAAAAwAAAAFAACA VwAAACQAAADkAgAATAEAAOgCAABcAQAAAgAAAGMu4BRjLqEVJQAAAAwAAAAH AACAJQAAAAwAAAAAAACAJAAAACQAAAAAAIBBAAAAAAAAAAAAAIBBAAAAAAAA AAACAAAAKAAAAAwAAAADAAAARgAAAFwAAABQAAAARU1GKwhABAM8AAAAMAAA AAEQwNsEAAAAAAAAAFOfOkTQdqxDTYo5RG73skNGdThE0HasQ1OfOkTQdqxD AAEBgRRABIAQAAAABAAAAAAAAP8kAAAAJAAAAAAAgD0AAAAAAAAAAAAAgD0A AAAAAAAAAAIAAAAlAAAADAAAAAUAAAATAAAADAAAAAEAAAAlAAAADAAAAAgA AIBWAAAALAAAAOECAABYAQAA6wIAAGYBAAAEAAAAqC6PFWMuXxYeLo8VqC6P FSUAAAAMAAAABwAAgBMAAAAMAAAAAQAAACUAAAAMAAAAAAAAgCQAAAAkAAAA AACAQQAAAAAAAAAAAACAQQAAAAAAAAAAAgAAAEYAAACIAAAAfAAAAEVNRisI QAMCPAAAADAAAAABEMDbAAAAAE4AAAAAAAAApz52PwIAAAACAAAAAgAAAAIA AAABEMDbAAAAAAAAAP8IQAQDLAAAACAAAAABEMDbAgAAAAAAAADHq4lCGECR Q4NyVULKfqhDAAEAABVABAAQAAAABAAAAAMAAAAkAAAAJAAAAAAAgD0AAAAA AAAAAAAAgD0AAAAAAAAAAAIAAABfAAAAOAAAAAMAAAA4AAAAAAAAADgAAAAA AAAAAAABAA8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAACUAAAAMAAAAAwAAACUA AAAMAAAABQAAgFcAAAAkAAAAMwAAACEBAABHAAAAUwEAAAIAAABOBCgSVgMQ FSUAAAAMAAAABwAAgCUAAAAMAAAAAAAAgCQAAAAkAAAAAACAQQAAAAAAAAAA AACAQQAAAAAAAAAAAgAAACgAAAAMAAAAAwAAAEYAAABcAAAAUAAAAEVNRisI QAQDPAAAADAAAAABEMDbBAAAAAAAAADfPWdCp6qoQ/pjRkJhJq5D+mNGQrlL p0PfPWdCp6qoQwABAYEUQASAEAAAAAQAAAAAAAD/JAAAACQAAAAAAIA9AAAA AAAAAAAAAIA9AAAAAAAAAAACAAAAJQAAAAwAAAAFAAAAEwAAAAwAAAABAAAA JQAAAAwAAAAIAACAVgAAACwAAAAxAAAATgEAADoAAABdAQAABAAAAJ0DFhUa A8UVGgPqFJ0DFhUlAAAADAAAAAcAAIATAAAADAAAAAEAAAAlAAAADAAAAAAA AIAkAAAAJAAAAAAAgEEAAAAAAAAAAAAAgEEAAAAAAAAAAAIAAABGAAAAWAEA AEwBAABFTUYrKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA6KnAQomnmkODaiFD CEAFBjAAAAAkAAAAARDA26uqKj4AAAAAAAAAAAAAAAAFAAAAQQBSAEkAQQBM AAAANkAFgPQAAADoAAAAAAAA/wEAAAABAAAAEwAAAE8AbgBsAHkAIABvAG4A ZQAgAG8AcwBwAGYAIAByAG8AdQB0AGUAmtkwPgrXozzNzJo+CtejPCJCyj4K 16M8dzfdPgrXozwR8QM/CtejPLzLDz8K16M8ZoYnPwrXozwRQT8/CtejPLz7 Vj8K16M8ZtZiPwrXozwRkXo/CtejPDPzhz8K16M8idCTPwrXozzevZk/Ctej PDOrnz8K16M83sWmPwrXozwzo7I/CtejPImAvj8K16M83m3EPwrXozwAAIA/ AAAAAAAAAAAAAIA/AAAAAAAAAAAAABIAAAAMAAAAAQAAABgAAAAMAAAAAAAA ABYAAAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAADAAAAARgEAAJQAAADOAQAA pQAAAAEAAAAAQNNBAPDSQUYBAACjAAAAEwAAAEwAAAAAAAAAAAAAAAAAAAD/ /////////3QAAABPAG4AbAB5ACAAbwBuAGUAIABvAHMAcABmACAAcgBvAHUA dABlAE0GDAAAAAkAAAAEAAAACAAAAAQAAAAJAAAACQAAAAkAAAAEAAAACQAA AAgAAAAJAAAABAAAAAQAAAAFAAAACQAAAAkAAAAEAAAAAAAAACUAAAAMAAAA DQAAgEYAAAAMAQAAAAEAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADy YMBCAAAAAAAAAADoqcBCiaeaQ4NqIUM2QAWAzAAAAMAAAAAAAAD/AQAAAAEA AAAPAAAAZgBvAHIAIAAxADAALgAxAC4AMQAuADAALwAyADQAIqK5Pq5HYT53 V9E+rkdhPmZmAD+uR2E+vJsOP65HYT5mdho/rkdhPhExMj+uR2E+vOtJP65H YT5mxlU/rkdhPhGBbT+uR2E+vFt5P65HYT4zi4g/rkdhPol4jj+uR2E+3lWa P65HYT4zQ6A/rkdhPokgrD+uR2E+AACAPwAAAAAAAAAAAACAPwAAAAAAAAAA AAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAADAAA AAEAAABUAAAAqAAAAFgBAACoAAAAuwEAALkAAAABAAAAAEDTQQDw0kFYAQAA twAAAA8AAABMAAAAAAAAAAAAAAAAAAAA//////////9sAAAAZgBvAHIAIAAx ADAALgAxAC4AMQAuADAALwAyADQATAYEAAAACQAAAAUAAAAEAAAACQAAAAkA AAAEAAAACQAAAAQAAAAJAAAABAAAAAkAAAAEAAAACQAAAAAAAAAlAAAADAAA AA0AAIBGAAAAIAEAABQBAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA 8mDAQgAAAAAAAAAA6KnAQomnmkODaiFDNkAFgOAAAADUAAAAAAAA/wEAAAAB AAAAEQAAAHMAaABvAHcAaQBuAGcAIABpAG4AIABSADEAIAAoAG4AbwDvjm4+ PQrXPiLyoT49Ctc+d2fRPj0K1z5mbgA/PQrXPmY+Hz89Ctc+EbkoPz0K1z68 c0A/PQrXPmYuWD89Ctc+EQlkPz0K1z68g20/PQrXPjOfgj89Ctc+iYyIPz0K 1z6J9Jc/PQrXPt7Roz89Ctc+M7+pPz0K1z7e2bA/PQrXPjO3vD89Ctc+AACA PwAAAAAAAAAAAACAPwAAAAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAA AAAWAAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAAtAAAAEwBAAC7AAAAywEA AMwAAAABAAAAAEDTQQDw0kFMAQAAygAAABEAAABMAAAAAAAAAAAAAAAAAAAA //////////9wAAAAcwBoAG8AdwBpAG4AZwAgAGkAbgAgAFIAMQAgACgAbgBv AHYFCAAAAAkAAAAJAAAADAAAAAQAAAAJAAAACQAAAAQAAAAEAAAACQAAAAQA AAAMAAAACQAAAAQAAAAFAAAACQAAAAAAAAAlAAAADAAAAA0AAIBGAAAAqAAA AJwAAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA 6KnAQomnmkODaiFDNkAFgGgAAABcAAAAAAAA/wEAAAABAAAABQAAAEUAQwBN AFAAKQARqSE/UrgeP2YePj9SuB4/Zu5cP1K4Hj+JPIA/UrgePzN3jj9SuB4/ AACAPwAAAAAAAAAAAACAPwAAAAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAA AAAAAAAWAAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAAbAAAAHIBAADOAAAA pAEAAN8AAAABAAAAAEDTQQDw0kFyAQAA3QAAAAUAAABMAAAAAAAAAAAAAAAA AAAA//////////9YAAAARQBDAE0AUAApADMECwAAAAwAAAANAAAACwAAAAAA AAAlAAAADAAAAA0AAIBGAAAAlAAAAIgAAABFTUYrK0AAAAwAAAAAAAAACEAD AjwAAAAwAAAAARDA2wAAAABOAAAAAAAAAKc+dj8CAAAAAgAAAAIAAAACAAAA ARDA2wAAAAAAAAD/CEAEAywAAAAgAAAAARDA2wIAAAAAAAAAGlK8Q56zaEMa UrxD9selQwABAAAVQAQAEAAAAAQAAAADAAAAJAAAACQAAAAAAIA9AAAAAAAA AAAAAIA9AAAAAAAAAAACAAAAXwAAADgAAAADAAAAOAAAAAAAAAA4AAAAAAAA AAAAAQAPAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAlAAAADAAAAAMAAAAlAAAA DAAAAAUAAIBXAAAAJAAAAHcBAADnAAAAewEAAE0BAAACAAAAixeMDosXuRQl AAAADAAAAAcAAIAlAAAADAAAAAAAAIAkAAAAJAAAAAAAgEEAAAAAAAAAAAAA gEEAAAAAAAAAAAIAAAAoAAAADAAAAAMAAABGAAAAXAAAAFAAAABFTUYrCEAE AzwAAAAwAAAAARDA2wQAAAAAAAAAJ3y+Qz49pUMaUrxD272rQw0oukM+PaVD J3y+Qz49pUMAAQGBFEAEgBAAAAAEAAAAAAAA/yQAAAAkAAAAAACAPQAAAAAA AAAAAACAPQAAAAAAAAAAAgAAACUAAAAMAAAABQAAABMAAAAMAAAAAQAAACUA AAAMAAAACAAAgFYAAAAsAAAAdAEAAEoBAAB9AQAAWAEAAAQAAADQF6gUixd4 FUUXqBTQF6gUJQAAAAwAAAAHAACAEwAAAAwAAAABAAAAJQAAAAwAAAAAAACA JAAAACQAAAAAAIBBAAAAAAAAAAAAAIBBAAAAAAAAAAACAAAARgAAAFgBAABM AQAARU1GKypAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEJrwX9D1dQXRAhA BQYwAAAAJAAAAAEQwNurqio+AAAAAAAAAAAAAAAABQAAAEEAUgBJAEEATAAA ADZABYD0AAAA6AAAAAAAAP8BAAAAAQAAABMAAABDAG8AbgBmAGkAZwB1AHIA YQB0AGkAbwBuACAAYQB0ACAAUgAxAI5zTz+2YEu/jkNuP7ZgS78c/4I/tmBL v3Lcjj+2YEu/x8mUP7ZgS78ch5k/tmBLv3JkpT+2YEu/x0GxP7ZgS79yXLg/ tmBLv8c5xD+2YEu/HCfKP7ZgS79y5M4/tmBLv8fB2j+2YEu/HJ/mP7ZgS79y jOw/tmBLv8dp+D+2YEu/HFf+P7ZgS785IgJAtmBLvznWCUC2YEu/AACAPwAA AAAAAAAAAACAPwAAAAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAW AAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAAwAAAAE4BAAAEAgAA1AEAABUC AAABAAAAAEDTQQDw0kFOAQAAEwIAABMAAABMAAAAAAAAAAAAAAAAAAAA//// //////90AAAAQwBvAG4AZgBpAGcAdQByAGEAdABpAG8AbgAgAGEAdAAgAFIA MQAAAAwAAAAJAAAACQAAAAQAAAAEAAAACQAAAAkAAAAFAAAACQAAAAQAAAAE AAAACQAAAAkAAAAEAAAACQAAAAQAAAAEAAAADAAAAAAAAAAlAAAADAAAAA0A AIBGAAAADAEAAAABAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDA QgAAAAAAAAAA6KnAQmvBf0PV1BdENkAFgMwAAADAAAAAAAAA/wEAAAABAAAA DwAAAHIAbwB1AHQAZQByACAAbwBzAHAAZgAgADEAMAAwAI7Ddz+DLRi/cvyC P4MtGL/H2Y4/gy0Yvxy3mj+DLRi/cqSgP4MtGL/Hgaw/gy0Yv3Kcsz+DLRi/ x4m5P4MtGL8cZ8U/gy0Yv8cR0D+DLRi/HO/bP4MtGL9y3OE/gy0Yv8fJ5z+D LRi/HKfzP4MtGL9yhP8/gy0YvwAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAAAAA EgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwAAAAB AAAAVAAAAKgAAABdAQAAFwIAAMUBAAAoAgAAAQAAAABA00EA8NJBXQEAACYC AAAPAAAATAAAAAAAAAAAAAAAAAAAAP//////////bAAAAHIAbwB1AHQAZQBy ACAAbwBzAHAAZgAgADEAMAAwAIMEBQAAAAkAAAAJAAAABAAAAAkAAAAFAAAA BAAAAAkAAAAIAAAACQAAAAQAAAAEAAAACQAAAAkAAAAAAAAAJQAAAAwAAAAN AACARgAAAHgBAABsAQAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAAAPJg wEIAAAAAAAAAAOipwEJrwX9D1dQXRDZABYA4AQAALAEAAAAAAP8BAAAAAQAA ABoAAAAgAHIAbwB1AHQAZQByAC0AaQBkACAAMgAwADAALgAyADAAMAAuADIA MAAwAC4AMgAwADAA5NALP5/0yb6Oqxc/n/TJvuTgJT+f9Mm+jps9P5/0yb45 VlU/n/TJvuQwYT+f9Mm+jut4P5/0yb5ykIM/n/TJvhyrij+f9Mm+cmiPP5/0 yb7HRZs/n/TJvhwzoT+f9Mm+chCtP5/0yb7H7bg/n/TJvhzLxD+f9Mm+crjK P5/0yb7HldY/n/TJvhxz4j+f9Mm+clDuP5/0yb7HPfQ/n/TJvo4NAECf9Mm+ OfwFQJ/0yb7k6gtAn/TJvo7hDkCf9Mm+OdAUQJ/0yb7kvhpAn/TJvgAAgD8A AAAAAAAAAAAAgD8AAAAAAAAAABIAAAAMAAAAAQAAABgAAAAMAAAAAAAAABYA AAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAADoAAAANAEAACoCAADtAQAAOwIA AAEAAAAAQNNBAPDSQTQBAAA5AgAAGgAAAEwAAAAAAAAAAAAAAAAAAAD///// /////4AAAAAgAHIAbwB1AHQAZQByAC0AaQBkACAAMgAwADAALgAyADAAMAAu ADIAMAAwAC4AMgAwADAABAAAAAUAAAAJAAAACQAAAAQAAAAJAAAABQAAAAUA AAAEAAAACQAAAAQAAAAJAAAACQAAAAkAAAAEAAAACQAAAAkAAAAJAAAABAAA AAkAAAAJAAAACQAAAAQAAAAJAAAACQAAAAAAAAAlAAAADAAAAA0AAIBGAAAA UAEAAEQBAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAA AAAA6KnAQmvBf0PV1BdENkAFgBABAAAEAQAAAAAA/wEAAAABAAAAFgAAACAA bABvAGcALQBhAGQAagBhAGMAZQBuAGMAeQAtAGMAaABhAG4AZwBlAHMA5PAl P3IcR76OyzE/chxHvjlGOz9yHEe+5ABTP3IcR76Ou2o/chxHvuTweD9yHEe+ x1WIP3IcR74cM5Q/chxHvnLwmD9yHEe+x82kP3IcR75yeK8/chxHvsdVuz9y HEe+HDPHP3IcR77H3dE/chxHvnKI3D9yHEe+HKPjP3IcR77HTe4/chxHvhwr +j9yHEe+OQQDQHIcR77k8ghAchxHvo7hDkByHEe+OdAUQHIcR74AAIA/AAAA AAAAAAAAAIA/AAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAA DAAAABgAAAAlAAAADAAAAAEAAABUAAAA0AAAAD4BAAA+AgAA5gEAAE8CAAAB AAAAAEDTQQDw0kE+AQAATQIAABYAAABMAAAAAAAAAAAAAAAAAAAA//////// //94AAAAIABsAG8AZwAtAGEAZABqAGEAYwBlAG4AYwB5AC0AYwBoAGEAbgBn AGUAcwAEAAAABAAAAAkAAAAJAAAABQAAAAkAAAAJAAAABAAAAAkAAAAIAAAA CQAAAAkAAAAIAAAACAAAAAUAAAAIAAAACQAAAAkAAAAJAAAACQAAAAkAAAAA AAAAJQAAAAwAAAANAACARgAAABgCAAAMAgAARU1GKytAAAAMAAAAAAAAACpA AAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEJrwX9D1dQXRDZABYDYAQAAzAEA AAAAAP8BAAAAAQAAACoAAAAgAHIAZQBkAGkAcwB0AHIAaQBiAHUAdABlACAA cwB0AGEAdABpAGMAIABtAGUAdAByAGkAYwAtAHQAeQBwAGUAIAAxACAAcwB1 AGIAbgBlAHQAcwA5jmM9YQu2O3Kc0D1hC7Y7jiMhPmELtjscB4A+YQu2O3J8 rz5hC7Y7x3HCPmELtjtyHO0+YQu2O+RoAj9hC7Y7OZ4QP2ELtjvkGBo/YQu2 O47TMT9hC7Y7OY5JP2ELtjvkaFU/YQu2O44jbT9hC7Y7Of54P2ELtjvHKYc/ YQu2OxwXjT9hC7Y7cvSYP2ELtjvH4Z4/YQu2Oxyfoz9hC7Y7x0muP2ELtjsc N7Q/YQu2O3L8xT9hC7Y7x9nRP2ELtjscx9c/YQu2O8fh3j9hC7Y7HJ/jP2EL tjvHSe4/YQu2O3Jk9T9hC7Y7x1H7P2ELtjs5/gJAYQu2O+TsCEBhC7Y7jtsO QGELtjs50hFAYQu2O+TAF0BhC7Y7jrcaQGELtjvkDCBAYQu2O477JUBhC7Y7 OeorQGELtjvk2DFAYQu2O47HN0BhC7Y7Ob46QGELtjsAAIA/AAAAAAAAAAAA AIA/AAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgA AAAlAAAADAAAAAEAAABUAAAASAEAAAUBAABRAgAAGgIAAGICAAABAAAAAEDT QQDw0kEFAQAAYAIAACoAAABMAAAAAAAAAAAAAAAAAAAA//////////+gAAAA IAByAGUAZABpAHMAdAByAGkAYgB1AHQAZQAgAHMAdABhAHQAaQBjACAAbQBl AHQAcgBpAGMALQB0AHkAcABlACAAMQAgAHMAdQBiAG4AZQB0AHMABAAAAAUA AAAJAAAACQAAAAQAAAAIAAAABAAAAAUAAAAEAAAACQAAAAkAAAAEAAAACQAA AAQAAAAIAAAABAAAAAkAAAAEAAAABAAAAAgAAAAEAAAADQAAAAkAAAAEAAAA BQAAAAQAAAAIAAAABQAAAAQAAAAIAAAACQAAAAkAAAAEAAAACQAAAAQAAAAI AAAACQAAAAkAAAAJAAAACQAAAAQAAAAAAAAAJQAAAAwAAAANAACARgAAAKAB AACUAQAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAA AOipwEJrwX9D1dQXRDZABYBgAQAAVAEAAAAAAP8BAAAAAQAAAB4AAAAgAHIA ZQBkAGkAcwB0AHIAaQBiAHUAdABlACAAbwBzAHAAZgAgADMAMAAwACAAcwB1 AGIAbgBlAHQAcwAcx94+KH1SPnJ89j4ofVI+jnMJPyh9Uj45LiE/KH1SPuTo OD8ofVI+jmNCPyh9Uj7kuFc/KH1SPo6TYz8ofVI+5MhxPyh9Uj6OQ3s/KH1S Phx/iT8ofVI+clyVPyh9Uj7HSZs/KH1SPhwnpz8ofVI+chStPyh9Uj7H8bg/ KH1SPnKcwz8ofVI+x3nPPyh9Uj4cZ9U/KH1SPnJU2z8ofVI+xzHnPyh9Uj4c D/M/KH1SPnLs/j8ofVI+5GwCQCh9Uj45wgdAKH1SPuSwDUAofVI+jp8TQCh9 Uj45jhlAKH1SPuR8H0AofVI+jnMiQCh9Uj4AAIA/AAAAAAAAAAAAAIA/AAAA AAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAA DAAAAAEAAABUAAAAAAEAACoBAABkAgAA+AEAAHUCAAABAAAAAEDTQQDw0kEq AQAAcwIAAB4AAABMAAAAAAAAAAAAAAAAAAAA//////////+IAAAAIAByAGUA ZABpAHMAdAByAGkAYgB1AHQAZQAgAG8AcwBwAGYAIAAzADAAMAAgAHMAdQBi AG4AZQB0AHMABAAAAAUAAAAJAAAACQAAAAQAAAAIAAAABAAAAAUAAAAEAAAA CQAAAAkAAAAEAAAACQAAAAQAAAAJAAAACAAAAAkAAAAEAAAABAAAAAkAAAAJ AAAACQAAAAQAAAAIAAAACQAAAAkAAAAJAAAACQAAAAQAAAAAAAAAJQAAAAwA AAANAACARgAAAMgBAAC8AQAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAA APJgwEIAAAAAAAAAAOipwEJrwX9D1dQXRDZABYCIAQAAfAEAAAAAAP8BAAAA AQAAACIAAAAgAG4AZQB0AHcAbwByAGsAIAAxADUALgAxAC4AMQAuADAAIAAw AC4AMAAuADAALgAyADUANQAgAGEAcgBlAGEAIAAwAByXhD76pM8+ckycPvqk zz7Hwcs++qTPPhw3+z76pM8+OXYJP/qkzz45Rig/+qTPPuQAQD/6pM8+OTZO P/qkzz6Oi2M/+qTPPjlmbz/6pM8+cpCDP/qkzz7HbY8/+qTPPhxblT/6pM8+ cjihP/qkzz7HJac/+qTPPhwDsz/6pM8+cvC4P/qkzz7HzcQ/+qTPPhy7yj/6 pM8+cpjWP/qkzz7Hhdw/+qTPPhxj6D/6pM8+clDuP/qkzz7HLfo/+qTPPo4N AED6pM8+OfwFQPqkzz7k6gtA+qTPPo7ZEUD6pM8+OdAUQPqkzz7kvhpA+qTP PjlMHkD6pM8+5DokQPqkzz6OKSpA+qTPPjkgLUD6pM8+AACAPwAAAAAAAAAA AACAPwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAY AAAAJQAAAAwAAAABAAAAVAAAABgBAAAZAQAAdwIAAAcCAACIAgAAAQAAAABA 00EA8NJBGQEAAIYCAAAiAAAATAAAAAAAAAAAAAAAAAAAAP//////////kAAA ACAAbgBlAHQAdwBvAHIAawAgADEANQAuADEALgAxAC4AMAAgADAALgAwAC4A MAAuADIANQA1ACAAYQByAGUAYQAgADAABAAAAAkAAAAJAAAABAAAAAwAAAAJ AAAABQAAAAgAAAAEAAAACQAAAAkAAAAEAAAACQAAAAQAAAAJAAAABAAAAAkA AAAEAAAACQAAAAQAAAAJAAAABAAAAAkAAAAEAAAACQAAAAkAAAAJAAAABAAA AAkAAAAFAAAACQAAAAkAAAAEAAAAAAAAACUAAAAMAAAADQAAgEYAAAAMAQAA AAEAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADo qcBCa8F/Q9XUF0Q2QAWAzAAAAMAAAAAAAAD/AQAAAAEAAAAPAAAAcgBvAHUA dABlAHIAIABvAHMAcABmACAAMwAwADAAjsN3P+Q4Tj9y/II/5DhOP8fZjj/k OE4/HLeaP+Q4Tj9ypKA/5DhOP8eBrD/kOE4/cpyzP+Q4Tj/Hibk/5DhOPxxn xT/kOE4/xxHQP+Q4Tj8c79s/5DhOP3Lc4T/kOE4/x8nnP+Q4Tj8cp/M/5DhO P3KE/z/kOE4/AACAPwAAAAAAAAAAAACAPwAAAAAAAAAAAAASAAAADAAAAAEA AAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAAqAAA AF0BAACeAgAAxQEAAK8CAAABAAAAAEDTQQDw0kFdAQAArQIAAA8AAABMAAAA AAAAAAAAAAAAAAAA//////////9sAAAAcgBvAHUAdABlAHIAIABvAHMAcABm ACAAMwAwADAAmAkFAAAACQAAAAkAAAAEAAAACQAAAAUAAAAEAAAACQAAAAgA AAAJAAAABAAAAAQAAAAJAAAACQAAAAAAAAAlAAAADAAAAA0AAIBGAAAAeAEA AGwBAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA 6KnAQmvBf0PV1BdENkAFgDgBAAAsAQAAAAAA/wEAAAABAAAAGgAAACAAcgBv AHUAdABlAHIALQBpAGQAIAAxADUAMAAuADEANQAwAC4AMQA1ADAALgAxADUA MADk0As/C7aAP46rFz8LtoA/5OAlPwu2gD+Omz0/C7aAPzlWVT8LtoA/5DBh Pwu2gD+O63g/C7aAP3KQgz8LtoA/HKuKPwu2gD9yaI8/C7aAP8dFmz8LtoA/ HDOhPwu2gD9yEK0/C7aAP8ftuD8LtoA/HMvEPwu2gD9yuMo/C7aAP8eV1j8L toA/HHPiPwu2gD9yUO4/C7aAP8c99D8LtoA/jg0AQAu2gD85/AVAC7aAP+Tq C0ALtoA/juEOQAu2gD850BRAC7aAP+S+GkALtoA/AACAPwAAAAAAAAAAAACA PwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAA JQAAAAwAAAABAAAAVAAAAOgAAAA0AQAAsQIAAO0BAADCAgAAAQAAAABA00EA 8NJBNAEAAMACAAAaAAAATAAAAAAAAAAAAAAAAAAAAP//////////gAAAACAA cgBvAHUAdABlAHIALQBpAGQAIAAxADUAMAAuADEANQAwAC4AMQA1ADAALgAx ADUAMAAEAAAABQAAAAkAAAAJAAAABAAAAAkAAAAFAAAABQAAAAQAAAAJAAAA BAAAAAkAAAAJAAAACQAAAAQAAAAJAAAACQAAAAkAAAAEAAAACQAAAAkAAAAJ AAAABAAAAAkAAAAJAAAAAAAAACUAAAAMAAAADQAAgEYAAABQAQAARAEAAEVN RisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADoqcBCa8F/ Q9XUF0Q2QAWAEAEAAAQBAAAAAAD/AQAAAAEAAAAWAAAAIABsAG8AZwAtAGEA ZABqAGEAYwBlAG4AYwB5AC0AYwBoAGEAbgBnAGUAcwDk8CU/pU+aP47LMT+l T5o/OUY7P6VPmj/kAFM/pU+aP467aj+lT5o/5PB4P6VPmj/HVYg/pU+aPxwz lD+lT5o/cvCYP6VPmj/HzaQ/pU+aP3J4rz+lT5o/x1W7P6VPmj8cM8c/pU+a P8fd0T+lT5o/cojcP6VPmj8co+M/pU+aP8dN7j+lT5o/HCv6P6VPmj85BANA pU+aP+TyCEClT5o/juEOQKVPmj850BRApU+aPwAAgD8AAAAAAAAAAAAAgD8A AAAAAAAAABIAAAAMAAAAAQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAAACUA AAAMAAAAAQAAAFQAAADQAAAAPgEAAMQCAADmAQAA1QIAAAEAAAAAQNNBAPDS QT4BAADTAgAAFgAAAEwAAAAAAAAAAAAAAAAAAAD//////////3gAAAAgAGwA bwBnAC0AYQBkAGoAYQBjAGUAbgBjAHkALQBjAGgAYQBuAGcAZQBzAAQAAAAE AAAACQAAAAkAAAAFAAAACQAAAAkAAAAEAAAACQAAAAgAAAAJAAAACQAAAAgA AAAIAAAABQAAAAgAAAAJAAAACQAAAAkAAAAJAAAACQAAAAAAAAAlAAAADAAA AA0AAIBGAAAA6AEAANwBAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA 8mDAQgAAAAAAAAAA6KnAQmvBf0PV1BdENkAFgKgBAACcAQAAAAAA/wEAAAAB AAAAJQAAACAAcgBlAGQAaQBzAHQAcgBpAGIAdQB0AGUAIABjAG8AbgBuAGUA YwB0AGUAZAAgAG0AZQB0AHIAaQBjAC0AdAB5AHAAZQAgADEA5Ng4Pj/psz+O Q2g+P+mzP3KMkD4/6bM/xwHAPj/psz8cd+8+P+mzPzk2AT8/6bM/josWPz/p sz85ZiI/P+mzP46bMD8/6bM/ORY6Pz/psz/k0FE/P+mzP46LaT8/6bM/OWZ1 Pz/psz9ykIY/P+mzP8d9jD8/6bM/ciiXPz/psz/HBaM/P+mzPxzjrj8/6bM/ csC6Pz/psz/HncY/P+mzP3JI0T8/6bM/xzXXPz/psz8cE+M/P+mzP3Lw7j8/ 6bM/x930Pz/psz+OUQNAP+mzPzlACUA/6bM/5DYMQD/psz85xA9AP+mzP+Qi EkA/6bM/OXgXQD/psz+OBRtAP+mzPzn8HUA/6bM/jlEjQD/psz85QClAP+mz P+QuL0A/6bM/jiUyQD/psz8AAIA/AAAAAAAAAAAAAIA/AAAAAAAAAAAAABIA AAAMAAAAAQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAAAQAA AFQAAAAsAQAAEQEAANgCAAANAgAA6QIAAAEAAAAAQNNBAPDSQREBAADnAgAA JQAAAEwAAAAAAAAAAAAAAAAAAAD//////////5gAAAAgAHIAZQBkAGkAcwB0 AHIAaQBiAHUAdABlACAAYwBvAG4AbgBlAGMAdABlAGQAIABtAGUAdAByAGkA YwAtAHQAeQBwAGUAIAAxAAgNBAAAAAUAAAAJAAAACQAAAAQAAAAIAAAABAAA AAUAAAAEAAAACQAAAAkAAAAEAAAACQAAAAQAAAAIAAAACQAAAAkAAAAJAAAA CQAAAAgAAAAEAAAACQAAAAkAAAAEAAAADQAAAAkAAAAEAAAABQAAAAQAAAAI AAAABQAAAAQAAAAIAAAACQAAAAkAAAAEAAAAAAAAACUAAAAMAAAADQAAgEYA AAC8AAAAsAAAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAA AAAAAADoqcBCa8F/Q9XUF0Q2QAWAfAAAAHAAAAAAAAD/AQAAAAEAAAAHAAAA cwB1AGIAbgBlAHQAcwDHRZ4/2ILNP3LwqD/Ygs0/x820P9iCzT8cq8A/2ILN P3KIzD/Ygs0/x2XYP9iCzT8cU94/2ILNPwAAgD8AAAAAAAAAAAAAgD8AAAAA AAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAA AAwAAAABAAAAVAAAAHgAAAB3AQAA6wIAAK0BAAD8AgAAAQAAAABA00EA8NJB dwEAAPoCAAAHAAAATAAAAAAAAAAAAAAAAAAAAP//////////XAAAAHMAdQBi AG4AZQB0AHMAYQ0IAAAACQAAAAkAAAAJAAAACQAAAAQAAAAAAAAAJQAAAAwA AAANAACARgAAAKABAACUAQAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAA APJgwEIAAAAAAAAAAOipwEJrwX9D1dQXRDZABYBgAQAAVAEAAAAAAP8BAAAA AQAAAB4AAAAgAHIAZQBkAGkAcwB0AHIAaQBiAHUAdABlACAAbwBzAHAAZgAg ADEAMAAwACAAcwB1AGIAbgBlAHQAcwAcx94+chznP3J89j5yHOc/jnMJP3Ic 5z85LiE/chznP+ToOD9yHOc/jmNCP3Ic5z/kuFc/chznP46TYz9yHOc/5Mhx P3Ic5z+OQ3s/chznPxx/iT9yHOc/clyVP3Ic5z/HSZs/chznPxwnpz9yHOc/ chStP3Ic5z/H8bg/chznP3Kcwz9yHOc/x3nPP3Ic5z8cZ9U/chznP3JU2z9y HOc/xzHnP3Ic5z8cD/M/chznP3Ls/j9yHOc/5GwCQHIc5z85wgdAchznP+Sw DUByHOc/jp8TQHIc5z85jhlAchznP+R8H0ByHOc/jnMiQHIc5z8AAIA/AAAA AAAAAAAAAIA/AAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAA DAAAABgAAAAlAAAADAAAAAEAAABUAAAAAAEAACoBAAD+AgAA+AEAAA8DAAAB AAAAAEDTQQDw0kEqAQAADQMAAB4AAABMAAAAAAAAAAAAAAAAAAAA//////// //+IAAAAIAByAGUAZABpAHMAdAByAGkAYgB1AHQAZQAgAG8AcwBwAGYAIAAx ADAAMAAgAHMAdQBiAG4AZQB0AHMABAAAAAUAAAAJAAAACQAAAAQAAAAIAAAA BAAAAAUAAAAEAAAACQAAAAkAAAAEAAAACQAAAAQAAAAJAAAACAAAAAkAAAAE AAAABAAAAAkAAAAJAAAACQAAAAQAAAAIAAAACQAAAAkAAAAJAAAACQAAAAQA AAAAAAAAJQAAAAwAAAANAACARgAAAMgBAAC8AQAARU1GKytAAAAMAAAAAAAA ACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEJrwX9D1dQXRDZABYCIAQAA fAEAAAAAAP8BAAAAAQAAACIAAAAgAG4AZQB0AHcAbwByAGsAIAAzADcALgAx AC4AMQAuADAAIAAwAC4AMAAuADAALgAyADUANQAgAGEAcgBlAGEAIAAwAByX hD4GWwBAckycPgZbAEDHwcs+BlsAQBw3+z4GWwBAOXYJPwZbAEA5Rig/BlsA QOQAQD8GWwBAOTZOPwZbAECOi2M/BlsAQDlmbz8GWwBAcpCDPwZbAEDHbY8/ BlsAQBxblT8GWwBAcjihPwZbAEDHJac/BlsAQBwDsz8GWwBAcvC4PwZbAEDH zcQ/BlsAQBy7yj8GWwBAcpjWPwZbAEDHhdw/BlsAQBxj6D8GWwBAclDuPwZb AEDHLfo/BlsAQI4NAEAGWwBAOfwFQAZbAEDk6gtABlsAQI7ZEUAGWwBAOdAU QAZbAEDkvhpABlsAQDlMHkAGWwBA5DokQAZbAECOKSpABlsAQDkgLUAGWwBA AACAPwAAAAAAAAAAAACAPwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAA AAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAAABgBAAAZAQAAEgMAAAcC AAAjAwAAAQAAAABA00EA8NJBGQEAACEDAAAiAAAATAAAAAAAAAAAAAAAAAAA AP//////////kAAAACAAbgBlAHQAdwBvAHIAawAgADMANwAuADEALgAxAC4A MAAgADAALgAwAC4AMAAuADIANQA1ACAAYQByAGUAYQAgADAABAAAAAkAAAAJ AAAABAAAAAwAAAAJAAAABQAAAAgAAAAEAAAACQAAAAkAAAAEAAAACQAAAAQA AAAJAAAABAAAAAkAAAAEAAAACQAAAAQAAAAJAAAABAAAAAkAAAAEAAAACQAA AAkAAAAJAAAABAAAAAkAAAAFAAAACQAAAAkAAAAEAAAAAAAAACUAAAAMAAAA DQAAgEYAAAA8AQAAMAEAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADy YMBCAAAAAAAAAADoqcBCgxATRKs9B0QIQAUGMAAAACQAAAABEMDbq6oqPgAA AAAAAAAAAAAAAAUAAABBAFIASQBBAEwAAAA2QAWAzAAAAMAAAAAAAAD/AQAA AAEAAAAPAAAAcgBvAHUAdABlAHIAIABvAHMAcABmACAAMQAwADAAMzOUPm7b Nj7enbA+bts2PjMT4D5u2zY+RMQHP27bNj7vnhM/bts2PppZKz9u2zY+7445 P27bNj6aaUU/bts2PkQkXT9u2zY+mnlyP27bNj4iGoU/bts2PncHiz9u2zY+ zfSQP27bNj4i0pw/bts2PnevqD9u2zY+AACAPwAAAAAAAAAAAACAPwAAAAAA AAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAA DAAAAAEAAABUAAAAqAAAAGgCAAAfAgAA0AIAADACAAABAAAAAEDTQQDw0kFo AgAALgIAAA8AAABMAAAAAAAAAAAAAAAAAAAA//////////9sAAAAcgBvAHUA dABlAHIAIABvAHMAcABmACAAMQAwADAAFAAFAAAACQAAAAkAAAAEAAAACQAA AAUAAAAEAAAACQAAAAgAAAAJAAAABAAAAAQAAAAJAAAACQAAAAAAAAAlAAAA DAAAAA0AAIBGAAAADAEAAAABAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAY AAAA8mDAQgAAAAAAAAAA6KnAQoMQE0SrPQdENkAFgMwAAADAAAAAAAAA/wEA AAABAAAADwAAACAAbwBzAHAAZgAgAHIAbwB1AHQAZQByAC0AaQBkAIn4qz4d 1ME+3q3DPh3UwT4zI/M+HdTBPu/mDj8d1ME+mqEmPx3UwT5EfDI/HdTBPu9W Pj8d1ME+RIxMPx3UwT7vRmQ/HdTBPpoBfD8d1ME+Iu6DPx3UwT53y48/HdTB PiLmlj8d1ME+zQCePx3UwT4ivqI/HdTBPgAAgD8AAAAAAAAAAAAAgD8AAAAA AAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAA AAwAAAABAAAAVAAAAKgAAABtAgAAMgIAAMwCAABDAgAAAQAAAABA00EA8NJB bQIAAEECAAAPAAAATAAAAAAAAAAAAAAAAAAAAP//////////bAAAACAAbwBz AHAAZgAgAHIAbwB1AHQAZQByAC0AaQBkAEUGBAAAAAkAAAAIAAAACQAAAAQA AAAEAAAABQAAAAkAAAAJAAAABAAAAAkAAAAFAAAABQAAAAQAAAAAAAAAJQAA AAwAAAANAACARgAAAOQAAADYAAAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAA GAAAAPJgwEIAAAAAAAAAAOipwEKDEBNEqz0HRDZABYCkAAAAmAAAAAAAAP8B AAAAAQAAAAsAAAA1ADAALgA1ADAALgA1ADAALgA1ADAA3s3RPkIdFD+aoQA/ Qh0UP0RcGD9CHRQ/7zYkP0IdFD+a8Ts/Qh0UP0SsUz9CHRQ/74ZfP0IdFD+a QXc/Qh0UPyJ+hz9CHRQ/d2uNP0IdFD/NSJk/Qh0UPwAAgD8AAAAAAAAAAAAA gD8AAAAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAY AAAAJQAAAAwAAAABAAAAVAAAAJAAAAB0AgAARgIAAMYCAABXAgAAAQAAAABA 00EA8NJBdAIAAFUCAAALAAAATAAAAAAAAAAAAAAAAAAAAP//////////ZAAA ADUAMAAuADUAMAAuADUAMAAuADUAMAAAAAkAAAAJAAAABAAAAAkAAAAJAAAA BAAAAAkAAAAJAAAABAAAAAkAAAAAAAAAJQAAAAwAAAANAACARgAAADwBAAAw AQAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOip wEKDEBNEqz0HRDZABYD8AAAA8AAAAAAAAP8BAAAAAQAAABQAAAAgAHIAZQBk AGkAcwB0AHIAaQBiAHUAdABlACAAcwB0AGEAdABpAGMAEbFJPnVQRz+8G3k+ dVBHP4n4mD51UEc/3m3IPnVQRz8z4/c+dVBHP0RsBT91UEc/msEaP3VQRz9E nCY/dVBHP5rRND91UEc/REw+P3VQRz/vBlY/dVBHP5rBbT91UEc/RJx5P3VQ Rz93q4g/dVBHP82Yjj91UEc/d0OZP3VQRz/NMJ8/dVBHPyIOqz91UEc/d/uw P3VQRz/NuLU/dVBHPwAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAABIAAAAMAAAA AQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAADE AAAAXwIAAFkCAADZAgAAagIAAAEAAAAAQNNBAPDSQV8CAABoAgAAFAAAAEwA AAAAAAAAAAAAAAAAAAD//////////3QAAAAgAHIAZQBkAGkAcwB0AHIAaQBi AHUAdABlACAAcwB0AGEAdABpAGMABAAAAAUAAAAJAAAACQAAAAQAAAAIAAAA BAAAAAUAAAAEAAAACQAAAAkAAAAEAAAACQAAAAQAAAAIAAAABAAAAAkAAAAE AAAABAAAAAAAAAAlAAAADAAAAA0AAIBGAAAA+AAAAOwAAABFTUYrK0AAAAwA AAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA6KnAQoMQE0SrPQdENkAF gLgAAACsAAAAAAAA/wEAAAABAAAADQAAAG0AZQB0AHIAaQBjAC0AdAB5AHAA ZQAgADEA3p28PqiDej+a2QE/qIN6P0SUGT+og3o/724lP6iDej9EpDM/qIN6 P+8ePT+og3o/RHRSP6iDej+aqWA/qIN6P0SEbD+og3o/zeyAP6iDej8iyow/ qIN6P3enmD+og3o/zZSeP6iDej8AAIA/AAAAAAAAAAAAAIA/AAAAAAAAAAAA ABIAAAAMAAAAAQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAA AQAAAFQAAACcAAAAcAIAAGwCAADHAgAAfQIAAAEAAAAAQNNBAPDSQXACAAB7 AgAADQAAAEwAAAAAAAAAAAAAAAAAAAD//////////2gAAABtAGUAdAByAGkA YwAtAHQAeQBwAGUAIAAxACAFDQAAAAkAAAAEAAAABQAAAAQAAAAIAAAABQAA AAQAAAAIAAAACQAAAAkAAAAEAAAAAAAAACUAAAAMAAAADQAAgEYAAAAgAQAA FAEAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADo qcBCgxATRKs9B0Q2QAWA4AAAANQAAAAAAAD/AQAAAAEAAAARAAAAIABuAGUA dAB3AG8AcgBrACAAMQA1AC4AMQAuADEALgAwAGbGVz5u25Y/iZiDPm7blj/e DbM+btuWPzOD4j5u25Y/iTj6Pm7blj9E7Bs/btuWP++mMz9u25Y/RNxBP27b lj+aMVc/btuWP0QMYz9u25Y/78Z6P27blj/NQIk/btuWPyIujz9u25Y/dwub P27blj/N+KA/btuWPyLWrD9u25Y/d8OyP27blj8AAIA/AAAAAAAAAAAAAIA/ AAAAAAAAAAAAABIAAAAMAAAAAQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAA ACUAAAAMAAAAAQAAAFQAAAC0AAAAYQIAAH8CAADYAgAAkAIAAAEAAAAAQNNB APDSQWECAACOAgAAEQAAAEwAAAAAAAAAAAAAAAAAAAD//////////3AAAAAg AG4AZQB0AHcAbwByAGsAIAAxADUALgAxAC4AMQAuADAANAcEAAAACQAAAAkA AAAEAAAADAAAAAkAAAAFAAAACAAAAAQAAAAJAAAACQAAAAQAAAAJAAAABAAA AAkAAAAEAAAAAAAAACUAAAAMAAAADQAAgEYAAAAUAQAACAEAAEVNRisrQAAA DAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADoqcBCgxATRKs9B0Q2 QAWA1AAAAMgAAAAAAAD/AQAAAAEAAAAQAAAAMAAuADAALgAwAC4AMgA1ADUA IABhAHIAZQBhACAAMAC823g+B3WwPzPjqz4HdbA/iZjDPgd1sD/eDfM+B3Ww P5phBT8HdbA/RBwdPwd1sD/v9ig/B3WwP5qxQD8HdbA/RGxYPwd1sD/vJnA/ B3WwP5oBfD8HdbA/It6JPwd1sD/N+JA/B3WwPyLWnD8HdbA/d7OoPwd1sD/N oK4/B3WwPwAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAABIAAAAMAAAAAQAAABgA AAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAACsAAAAZAIA AJMCAADVAgAApAIAAAEAAAAAQNNBAPDSQWQCAACiAgAAEAAAAEwAAAAAAAAA AAAAAAAAAAD//////////2wAAAAwAC4AMAAuADAALgAyADUANQAgAGEAcgBl AGEAIAAwAAkAAAAEAAAACQAAAAQAAAAJAAAABAAAAAkAAAAJAAAACQAAAAQA AAAJAAAABQAAAAkAAAAJAAAABAAAAAAAAAAlAAAADAAAAA0AAIBGAAAAPAEA ADABAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA 6KnAQluf1kKsxwxECEAFBjAAAAAkAAAAARDA26uqKj4AAAAAAAAAAAAAAAAF AAAAQQBSAEkAQQBMAAAANkAFgMwAAADAAAAAAAAA/wEAAAABAAAADwAAAHIA bwB1AHQAZQByACAAbwBzAHAAZgAgADEAMAAwAM3MDj4pXI++IqJHPilcj75m RpM+KVyPvry7wj4pXI++EXHaPilcj74z8wQ/KVyPvokoEz8pXI++MwMfPylc j77evTY/KVyPvjMTTD8pXI++3s1jPylcj76JqG8/KVyPvjODez8pXI++756J Pylcj75EfJU/KVyPvgAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAAAAAEgAAAAwA AAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAA AKgAAAB5AAAACQIAAOEAAAAaAgAAAQAAAABA00EA8NJBeQAAABgCAAAPAAAA TAAAAAAAAAAAAAAAAAAAAP//////////bAAAAHIAbwB1AHQAZQByACAAbwBz AHAAZgAgADEAMAAwADgKBQAAAAkAAAAJAAAABAAAAAkAAAAFAAAABAAAAAkA AAAIAAAACQAAAAQAAAAEAAAACQAAAAkAAAAAAAAAJQAAAAwAAAANAACARgAA ANgAAADMAAAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAAAPJgwEIAAAAA AAAAAOipwEJbn9ZCrMcMRDZABYCYAAAAjAAAAAAAAP8BAAAAAQAAAAoAAAAg AHIAbwB1AHQAZQByAC0AaQBkALyruz4K16O9EWHTPgrXo728y+8+CtejvYmg Dz8K16O9M1snPwrXo73eNTM/CtejvYnwSj8K16O93iVZPwrXo70zW2c/Ctej vd7VcD8K16O9AACAPwAAAAAAAAAAAACAPwAAAAAAAAAAEgAAAAwAAAABAAAA GAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAAAIgAAACP AAAAHAIAAMwAAAAtAgAAAQAAAABA00EA8NJBjwAAACsCAAAKAAAATAAAAAAA AAAAAAAAAAAAAP//////////YAAAACAAcgBvAHUAdABlAHIALQBpAGQABAAA AAUAAAAJAAAACQAAAAQAAAAJAAAABQAAAAUAAAAEAAAAAAAAACUAAAAMAAAA DQAAgEYAAAAMAQAAAAEAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADy YMBCAAAAAAAAAADoqcBCW5/WQqzHDEQ2QAWAzAAAAMAAAAAAAAD/AQAAAAEA AAAPAAAAMQAxADAALgAxADEAMAAuADEAMQAwAC4AMQAxADAAmlmYPY/C9T13 Fys+j8L1PREBhT6PwvU9Zna0Po/C9T28K8w+j8L1PRGh+z6PwvU9M4sVP4/C 9T3eRS0/j8L1PYkgOT+PwvU9M9tQP4/C9T3elWg/j8L1PUQogD+PwvU9mhWG P4/C9T3v8pE/j8L1PUTQnT+PwvU9AACAPwAAAAAAAAAAAACAPwAAAAAAAAAA AAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgAAAAlAAAADAAA AAEAAABUAAAAqAAAAHIAAAAwAgAA6AAAAEECAAABAAAAAEDTQQDw0kFyAAAA PwIAAA8AAABMAAAAAAAAAAAAAAAAAAAA//////////9sAAAAMQAxADAALgAx ADEAMAAuADEAMQAwAC4AMQAxADAAAAAJAAAACQAAAAkAAAAEAAAACQAAAAkA AAAJAAAABAAAAAkAAAAJAAAACQAAAAQAAAAJAAAACQAAAAAAAAAlAAAADAAA AA0AAIBGAAAADAEAAAABAABFTUYrK0AAAAwAAAAAAAAAKkAAACQAAAAYAAAA 8mDAQgAAAAAAAAAA6KnAQluf1kKsxwxENkAFgMwAAADAAAAAAAAA/wEAAAAB AAAADwAAACAAbABvAGcALQBhAGQAagBhAGMAZQBuAGMAeQAtACIiCj4K16M+ zYw5PgrXoz53d18+CtejPhExnz4K16M+ZqbOPgrXoz4REes+CtejPjNDDT8K 16M+3v0kPwrXoz6JeC4/CtejPjMzRj8K16M+iYhbPwrXoz4zQ3M/CtejPu9+ hT8K16M+mimQPwrXoz5E1Jo/CtejPgAAgD8AAAAAAAAAAAAAgD8AAAAAAAAA AAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwA AAABAAAAVAAAAKgAAAB4AAAAQwIAAOMAAABUAgAAAQAAAABA00EA8NJBeAAA AFICAAAPAAAATAAAAAAAAAAAAAAAAAAAAP//////////bAAAACAAbABvAGcA LQBhAGQAagBhAGMAZQBuAGMAeQAtAAAABAAAAAQAAAAJAAAACQAAAAUAAAAJ AAAACQAAAAQAAAAJAAAACAAAAAkAAAAJAAAACAAAAAgAAAAAAAAAJQAAAAwA AAANAACARgAAALwAAACwAAAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAA APJgwEIAAAAAAAAAAOipwEJbn9ZCrMcMRDZABYB8AAAAcAAAAAAAAP8BAAAA AQAAAAcAAABjAGgAYQBuAGcAZQBzAGYWxT64HgU/EcHvPrgeBT8zmw8/uB4F P95VJz+4HgU/iRA/P7geBT8zy1Y/uB4FP96Fbj+4HgU/AACAPwAAAAAAAAAA AACAPwAAAAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAA ABgAAAAlAAAADAAAAAEAAABUAAAAeAAAAJAAAABWAgAAywAAAGcCAAABAAAA AEDTQQDw0kGQAAAAZQIAAAcAAABMAAAAAAAAAAAAAAAAAAAA//////////9c AAAAYwBoAGEAbgBnAGUAcwAAAAgAAAAJAAAACQAAAAkAAAAJAAAACQAAAAAA AAAlAAAADAAAAA0AAIBGAAAA+AAAAOwAAABFTUYrK0AAAAwAAAAAAAAAKkAA ACQAAAAYAAAA8mDAQgAAAAAAAAAA6KnAQluf1kKsxwxENkAFgLgAAACsAAAA AAAA/wEAAAABAAAADQAAACAAcgBlAGQAaQBzAHQAcgBpAGIAdQB0AGUAvHuH PuxROD8RMZ8+7FE4P7ybuz7sUTg/ERHrPuxROD8zQw0/7FE4P969Fj/sUTg/ MxMsP+xROD/e7Tc/7FE4PzMjRj/sUTg/3p1PP+xROD+JWGc/7FE4PzMTfz/s UTg/73aFP+xROD8AAIA/AAAAAAAAAAAAAIA/AAAAAAAAAAAAABIAAAAMAAAA AQAAABgAAAAMAAAAAAAAABYAAAAMAAAAGAAAACUAAAAMAAAAAQAAAFQAAACc AAAAhQAAAGkCAADWAAAAegIAAAEAAAAAQNNBAPDSQYUAAAB4AgAADQAAAEwA AAAAAAAAAAAAAAAAAAD//////////2gAAAAgAHIAZQBkAGkAcwB0AHIAaQBi AHUAdABlAACABAAAAAUAAAAJAAAACQAAAAQAAAAIAAAABAAAAAUAAAAEAAAA CQAAAAkAAAAEAAAAAAAAACUAAAAMAAAADQAAgEYAAAAoAQAAHAEAAEVNRisr QAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADoqcBCW5/WQqzH DEQ2QAWA6AAAANwAAAAAAAD/AQAAAAEAAAASAAAAcwB0AGEAdABpAGMAIABt AGUAdAByAGkAYwAtAHQAeQBwAGUAmtmPPR+Faz8iQh0+H4VrP82sTD4fhWs/ vMuVPh+Faz8Rga0+H4VrP2Z2wD4fhWs/ESHrPh+Faz8zawE/H4VrP971JD8f hWs/ibA8Px+Faz8zi0g/H4VrP4nAVj8fhWs/MztgPx+Faz+JkHU/H4VrP+/i gT8fhWs/RNCHPx+Faz/vepI/H4VrP0RYnj8fhWs/AACAPwAAAAAAAAAAAACA PwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAA JQAAAAwAAAABAAAAVAAAALgAAAByAAAAfQIAAOcAAACOAgAAAQAAAABA00EA 8NJBcgAAAIwCAAASAAAATAAAAAAAAAAAAAAAAAAAAP//////////cAAAAHMA dABhAHQAaQBjACAAbQBlAHQAcgBpAGMALQB0AHkAcABlAAgAAAAEAAAACQAA AAQAAAAEAAAACAAAAAQAAAANAAAACQAAAAQAAAAFAAAABAAAAAgAAAAFAAAA BAAAAAgAAAAJAAAAAAAAACUAAAAMAAAADQAAgEYAAADQAAAAxAAAAEVNRisr QAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADoqcBCW5/WQqzH DEQ2QAWAkAAAAIQAAAAAAAD/AQAAAAEAAAAJAAAAMQAgAHMAdQBiAG4AZQB0 AHMAEWGtPilcjz9m1tw+KVyPP7yL9D4pXI8/M5sPPylcjz/eVSc/KVyPP4kQ Pz8pXI8/M8tWPylcjz/ehW4/KVyPP4lgej8pXI8/AACAPwAAAAAAAAAAAACA PwAAAAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAAAAAWAAAADAAAABgA AAAlAAAADAAAAAEAAABUAAAAhAAAAIwAAACQAgAAzwAAAKECAAABAAAAAEDT QQDw0kGMAAAAnwIAAAkAAABMAAAAAAAAAAAAAAAAAAAA//////////9gAAAA MQAgAHMAdQBiAG4AZQB0AHMAAAAJAAAABAAAAAgAAAAJAAAACQAAAAkAAAAJ AAAABAAAAAAAAAAlAAAADAAAAA0AAIBGAAAAxAAAALgAAABFTUYrK0AAAAwA AAAAAAAAKkAAACQAAAAYAAAA8mDAQgAAAAAAAAAA6KnAQluf1kKsxwxENkAF gIQAAAB4AAAAAAAA/wEAAAABAAAACAAAACAAbgBlAHQAdwBvAHIAawBmJsU+ w/WoP7zb3D7D9ag/iSgGP8P1qD8z4x0/w/WoP969KT/D9ag/3o1IP8P1qD+J SGA/w/WoP959bj/D9ag/AACAPwAAAAAAAAAAAACAPwAAAAAAAAAAEgAAAAwA AAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAA AHwAAACQAAAAowIAAMsAAAC0AgAAAQAAAABA00EA8NJBkAAAALICAAAIAAAA TAAAAAAAAAAAAAAAAAAAAP//////////XAAAACAAbgBlAHQAdwBvAHIAawAE AAAACQAAAAkAAAAEAAAADAAAAAkAAAAFAAAAAAAAACUAAAAMAAAADQAAgEYA AADsAAAA4AAAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAA AAAAAADoqcBCW5/WQqzHDEQ2QAWArAAAAKAAAAAAAAD/AQAAAAEAAAAMAAAA MQAwAC4AMgAuADEAMgA5AC4AMQAzADIAzYxaPlyPwj+8u5w+XI/CPxExzD5c j8I/ZubjPlyPwj/erQk/XI/CP4mIFT9cj8I/M0MtP1yPwj/e/UQ/XI/CP4m4 XD9cj8I/M5NoP1yPwj/vJoA/XI/CP0QEjD9cj8I/AACAPwAAAAAAAAAAAACA PwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAA JQAAAAwAAAABAAAAVAAAAJQAAACAAAAAtwIAANsAAADIAgAAAQAAAABA00EA 8NJBgAAAAMYCAAAMAAAATAAAAAAAAAAAAAAAAAAAAP//////////ZAAAADEA MAAuADIALgAxADIAOQAuADEAMwAyAAkAAAAJAAAABAAAAAkAAAAEAAAACQAA AAkAAAAJAAAABAAAAAkAAAAJAAAAAAAAACUAAAAMAAAADQAAgEYAAAAAAQAA 9AAAAEVNRisrQAAADAAAAAAAAAAqQAAAJAAAABgAAADyYMBCAAAAAAAAAADo qcBCW5/WQqzHDEQ2QAWAwAAAALQAAAAAAAD/AQAAAAEAAAAOAAAAMAAuADAA LgAwAC4AMwAgAGEAcgBlAGEAIAAwAM0sPj72KNw/vIuOPvYo3D8RQaY+9ijc P2a21T72KNw/vGvtPvYo3D+JcA4/9ijcPzNLGj/2KNw/3gUyP/Yo3D+J4D0/ 9ijcPzObVT/2KNw/idBjP/Yo3D8zi3s/9ijcP++iiT/2KNw/RJCPP/Yo3D8A AIA/AAAAAAAAAAAAAIA/AAAAAAAAAAASAAAADAAAAAEAAAAYAAAADAAAAAAA AAAWAAAADAAAABgAAAAlAAAADAAAAAEAAABUAAAAoAAAAH0AAADKAgAA3AAA ANsCAAABAAAAAEDTQQDw0kF9AAAA2QIAAA4AAABMAAAAAAAAAAAAAAAAAAAA //////////9oAAAAMAAuADAALgAwAC4AMwAgAGEAcgBlAGEAIAAwAAkAAAAE AAAACQAAAAQAAAAJAAAABAAAAAkAAAAEAAAACQAAAAUAAAAJAAAACQAAAAQA AAAAAAAAJQAAAAwAAAANAACARgAAACABAAAUAQAARU1GKytAAAAMAAAAAAAA ACpAAAAkAAAAGAAAAPJgwEIAAAAAAAAAAOipwEJbn9ZCrMcMRDZABYDgAAAA 1AAAAAAAAP8BAAAAAQAAABEAAAAgAG4AZQB0AHcAbwByAGsAIAAzADcALgAx AC4AMQAuADAAM7N4PY/C9T/vLts9j8L1PyKCTD6PwvU/ZraVPo/C9T+8a60+ j8L1P7wL6z6PwvU/iUANP4/C9T/edRs/j8L1PzPLMD+PwvU/3qU8P4/C9T+J YFQ/j8L1PzMbbD+PwvU/3vV3P4/C9T9E2Ic/j8L1P5rFjT+PwvU/76KZP4/C 9T9EkJ8/j8L1PwAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAAAAAEgAAAAwAAAAB AAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAYAAAAJQAAAAwAAAABAAAAVAAAALQA AABxAAAA3QIAAOgAAADuAgAAAQAAAABA00EA8NJBcQAAAOwCAAARAAAATAAA AAAAAAAAAAAAAAAAAP//////////cAAAACAAbgBlAHQAdwBvAHIAawAgADMA NwAuADEALgAxAC4AMAAAAAQAAAAJAAAACQAAAAQAAAAMAAAACQAAAAUAAAAI AAAABAAAAAkAAAAJAAAABAAAAAkAAAAEAAAACQAAAAQAAAAAAAAAJQAAAAwA AAANAACARgAAABQBAAAIAQAARU1GKytAAAAMAAAAAAAAACpAAAAkAAAAGAAA APJgwEIAAAAAAAAAAOipwEJbn9ZCrMcMRDZABYDUAAAAyAAAAAAAAP8BAAAA AQAAABAAAAAwAC4AMAAuADAALgAyADUANQAgAGEAcgBlAGEAIAAwAESEvj0U rgdAzSw+PhSuB0B3l20+FK4HQBFBpj4UrgdAZva9PhSuB0C8a+0+FK4HQImQ Aj8UrgdAM0saPxSuB0DeBTI/FK4HQInAST8UrgdAM5tVPxSuB0DeVW0/FK4H QDOLez8UrgdA76KJPxSuB0BEgJU/FK4HQJptmz8UrgdAAACAPwAAAAAAAAAA AACAPwAAAAAAAAAAEgAAAAwAAAABAAAAGAAAAAwAAAAAAAAAFgAAAAwAAAAY AAAAJQAAAAwAAAABAAAAVAAAAKwAAAB0AAAA8AIAAOUAAAABAwAAAQAAAABA 00EA8NJBdAAAAP8CAAAQAAAATAAAAAAAAAAAAAAAAAAAAP//////////bAAA ADAALgAwAC4AMAAuADIANQA1ACAAYQByAGUAYQAgADAACQAAAAQAAAAJAAAA BAAAAAkAAAAEAAAACQAAAAkAAAAJAAAABAAAAAkAAAAFAAAACQAAAAkAAAAE AAAAAAAAACUAAAAMAAAADQAAgEYAAAAcAAAAEAAAAEVNRisrQAAADAAAAAAA AABMAAAAZAAAAOX///+VAAAAPAMAACMDAADl////lQAAAFgDAACPAgAAKQCq AAAAAAAAAAAAAACAPwAAAAAAAAAAAACAPwAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAIgAAAAwAAAD/////RgAAABwAAAAQAAAARU1GKwJAAAAM AAAAAAAAAA4AAAAUAAAAAAAAABAAAAAUAAAA//////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////AP////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////gQEAAIIBAACDAQAAhAEAAIUBAACGAQAA hwEAAIgBAACJAQAAigEAAIsBAACMAQAAjQEAAI4BAACPAQAAkAEAAJEBAACS AQAAkwEAAJQBAACVAQAAlgEAAJcBAACYAQAAmQEAAJoBAACbAQAAnAEAAJ0B AACeAQAAnwEAAKABAAChAQAAogEAAKMBAACkAQAApQEAAKYBAACnAQAAqAEA AKkBAACqAQAAqwEAAKwBAACtAQAArgEAAK8BAACwAQAAsQEAALIBAACzAQAA tAEAALUBAAC2AQAAtwEAALgBAAC5AQAAugEAALsBAAC8AQAAvQEAAL4BAAC/ AQAAwAEAAMEBAADCAQAAwwEAAMQBAADFAQAAxgEAAMcBAADIAQAAyQEAAMoB AADLAQAAzAEAAM0BAADOAQAAzwEAANABAADRAQAA0gEAANMBAADUAQAA/v// //3////XAQAA2AEAANkBAADaAQAA2wEAANwBAADdAQAA3gEAAN8BAADgAQAA 4QEAAOIBAADjAQAA5AEAAOUBAADmAQAA5wEAAOgBAADpAQAA6gEAAOsBAADs AQAA7QEAAO4BAADvAQAA8AEAAPEBAADyAQAA8wEAAPQBAAD1AQAA9gEAAPcB AAD4AQAA+QEAAPoBAAD7AQAA/AEAAP0BAAD+AQAA/wEAAAACAAD+/wAABAAC AAAAAAAAAAAAAAAAAAAAAAABAAAA4IWf8vlPaBCrkQgAKyez2TAAAABQ8AAA CAAAAAEAAABIAAAABAAAAFAAAAAGAAAAaAAAAAUAAAB0AAAAAwAAAIAAAAAH AAAAjAAAAAIAAAD8AAAAEQAAACQBAAACAAAAtQMAAB4AAAAOAAAAQWRtaW5p c3RyYXRvcgAAAB4AAAABAAAAAAAAAB4AAAABAAAAAAAAAB4AAAABAAAAAAAA AB4AAABoAAAAQzpcUHJvZ3JhbSBGaWxlc1xNaWNyb3NvZnQgT2ZmaWNlXFZp c2lvMTBcMTAzM1xTb2x1dGlvbnNcTmV0d29ya1xMb2dpY2FsIE5ldHdvcmsg RGlhZ3JhbSAoVVMgdW5pdHMpLnZzdAAeAAAAIAAAAE9TUEYgTXVsdGkgUHJv Y2VzcyBFQ01QIHByb2JsZW0ARwAAACTvAAD/////DgAAAAEAAACMAAAAAAAA AAAAAAB8AAAAoQAAAAAAAAAAAAAAygwAAJUQAAAgRU1GAAABABzvAAADAAAA AQAAAA8AAABsAAAAAAAAAAAFAAAABAAAUgEAAA4BAAAAAAAAAAAAAAAAAABQ KAUAsB4EAFYASQBTAEkATwAAAEQAcgBhAHcAaQBuAGcAAAAAAAAATAAAAHzu AAAAAAAAAAAAAHwAAAChAAAAAAAAAAAAAAB9AAAAogAAACAAzAAAAAAAAAAA AAAAgD8AAAAAAAAAAAAAgD8AAAAAAAAAAP///wAAAAAAZAAAACgAAACMAAAA 8O0AACgAAAB9AAAAogAAAAEAGAAAAAAA8O0AAAAAAAAAAAAAAAAAAAAAAAD/ //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////8A//////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////AP////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /wD///////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////8A//////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////AP////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////wD///////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////8A//////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////AP////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////wD///////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////8A//////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////AP////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////wD///////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////8A//////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////AP// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////wD///////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////8A//////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// AP////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////wD///////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////8A//////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////AP////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////wD///////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////8A//////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////AP////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////wD///////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////8A//////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////AP////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////wD///////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////8A//// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////AP////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////wD///////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////8A //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////AP////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////wD///////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //8A//////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////AP////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////wD///////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////8A//////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////AP////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////wD///////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////8A//////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA//////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////AP////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////+fn5////0BAQP///0BAQP///0BAQP///0BAQP// /0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BA QP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////wD///// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////8A//////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////AAAA////AAAA////AAAA////AAAA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA//////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////AP////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////4eHh////0BAQP///0BAQP// /0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BA QP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///7e3t/////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////////wD/ //////////////////////////////////////////////////////////// //////////8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8A AAD///8AAAD///8AAAD///////////////////////////////////////// //////////////////////////////////////////////////+/v7////+/ v7////+/v7////+/v7////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////8A//////////////////////////////////////// ////////////////////////////5+fn////QEBA////QEBA////QEBA//// QEBA////QEBA////QEBA////QEBA////QEBA////h4eH//////////////// //////////////////////////////////////////////////////////// ////////ICAg////AAAA////AAAA////AAAA////YGBg//////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////AP////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////0BAQP///0BAQP///0BAQP///0BA QP////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /wD///////////////////////////////////////////////////////// //////////9gYGD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8A AAD///8AAAD///8AAAD///8AAAD////////////////////n5+f///+/v7// //+/v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/ v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/v7// //+/v7////+/v7////+/v7/////39/f///////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////8A//////////////////////////////////// ////////////////////////////////////QEBA////QEBA////QEBA//// QEBA////QEBA////QEBA////QEBA////QEBA////QEBA//////////////// ////////////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////AP////////////// //////////////////////////////////////////////////////////// /8fHx////7+/v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/ v////////////////////////////8/Pz////39/f////39/f////39/f/// /39/f////39/f////39/f////39/f////39/f////39/f////39/f////39/ f////39/f////39/f////39/f////39/f////39/f////39/f////39/f/// /39/f////+/v7/////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////wD///////////////////////////////////////////////////// //////////////////////////8AAAD///8AAAD///8AAAD///8AAAD///8A AAD///8AAAD///8AAAD///+/v7////////////////////////////////// ///////////////////f39////+/v7////+/v7////+/v7////+/v7////+/ v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/v7// ///v7+////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////8A//////////////////////////////// ////////////////////////////////////////////WFhY////QEBA//// QEBA////QEBA////QEBA////QEBA////QEBA////QEBA//////////////// ////////////////////////////////////////////AAAA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA//////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////AP////////// //////////////////////////////////////////////////////////// /////////7+/v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/ v/////////////////////////////////////////////////////////// /7+/v////39/f////39/f////39/f////39/f////39/f////39/f////39/ f////39/f////39/f////39/f////39/f////39/f////9/f3/////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////wD///////////////////////////////////////////////// //////////////////////////+fn5////8AAAD///8AAAD///8AAAD///8A AAD///8AAAD///8AAAD///8gICD///////////////////////////////// //////////////////9/f3////9/f3////9/f3////9/f3////9/f3////9/ f3////9/f3////9/f3////9/f3////9/f3////9/f3////9/f3////9/f3// //9/f3////+fn5////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////8A//////////////////////////// ////////////////////////////////////////////////////QEBA//// QEBA////QEBA////QEBA////QEBA////QEBA////QEBA//////////////// ////////////////////////////////////////////AAAA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA//////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////AP////// //////////////////////////////////////////////////////////// /////////////////////////7+/v////7+/v////7+/v////7+/v////+/v 7/////////////////////////////////////////////////////////// /39/f////39/f////39/f////39/f////39/f////39/f////39/f////39/ f////39/f////39/f////39/f////39/f////39/f////39/f////5+fn/// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////wD///////////////////////////////////////////// //////////////////////////////////////////8gICD///8AAAD///8A AAD///8AAAD///8AAAD///////////////////////////////////////// //////////////////////////////////////////////////9/f3////9/ f3////9/f3////9/f3////9/f3////9/f3////9/f3////9/f3////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////8A//////////////////////// //////////////////////////////////////////////////////////// ////////f39/////f39/////f39/////f39/////39/f//////////////// //////////////////////////////////////////////////////////// ////////QEBA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////YGBg//////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////AP// //////////////////////////////////////////////////////////// /////////////////////////7+/v////7+/v////7+/v////7+/v////7+/ v/////////////////////////////////////////////////////////// /////////////////////////////////7+/v////7+/v////7+/v////7+/ v////7+/v////7+/v////7+/v////7+/v/////////////////////////// //////////////////////////////////////////////////////////// /////////////0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP// /0BAQP///0BAQP///0BAQP////////////////////////////////////// /////////////////wD///////////////////////////////////////// //////////////////////////////////////////9gYGD///8AAAD///8A AAD///8AAAD///8AAAD///8AAAD///////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////9AQED///8AAAD///8AAAD///8AAAD/ //8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///////////////////// //////////////////////////////////////8A//////////////////// //////////////////////////////////////////////////////////// ////////f39/////f39/////f39/////f39/////f39///////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// AP////////////////////////////////////////////////////////// /////////7+/v////39/f////39/f////39/f////39/f////39/f////39/ f////39/f////39/f////4+Pj/////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP// /0BAQP///0BAQP///0BAQP///7e3t/////////////////////////////// /////////////////////wD///////////////////////////////////// //////////////////////////////////8AAAD///8AAAD///8AAAD///8A AAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///////////////// ///////////v7+////9/f3////9/f3////9/f3////9/f3////9/f3////9/ f3////9/f3////9/f3////9/f3////9/f3////9/f3////9/f3////9/f3// //9/f3////9/f3////9/f3////9/f3////9/f3////////////////////// //////////////////////////////+/v7////8AAAD///8AAAD///8AAAD/ //8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///////////// //////////////////////////////////////////8A//////////////// ////////////////////////////////////////////////////v7+///// f39/////f39/////f39/////f39/////f39/////f39/////f39/////f39/ ////j4+P////////////////////////////AAAA////AAAA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//// AAAA//////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////AP////////////////////////////////////////////////////// /////////////////////////6+vr////39/f////39/f////39/f////39/ f////39/f////39/f/////////////////////////////////////f39/// /7+/v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/ v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/v/// /7+/v////7+/v////7+/v/////////////////////////////////////// /////////////////////////////wAAAP///wAAAP///wAAAP///wAAAP// /wAAAP///wAAAP///wAAAP////////////////////////////////////// /////////////////////////wD///////////////////////////////// //////////////////////////////////////////////////8AAAD///8A AAD///8AAAD///8AAAD///8AAAD///8AAAD////f39////////////////// //////////////////////////+Hh4f///9AQED///9AQED///9AQED///9A QED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED/ //9AQED///9AQED///9AQED///9AQED///+3t7f///////////////////// //////////////////////////////////////////////8AAAD///8AAAD/ //8AAAD///8AAAD///8AAAD///8AAAD///8AAAD////f39////////////// //////////////////////////////////////////////8A//////////// //////////////////////////////////////////////////////////// ////////19fX////v7+/////v7+/////v7+/////v7+/////v7+/////v7+/ ////////////////////////////////////////////////////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////AP////////////////////////////////////////////////// /////////////////////////////////////////39/f////39/f////39/ f////39/f////8/Pz/////////////////////////////////////////// /////////9fX1////7+/v////7+/v////7+/v////7+/v////7+/v////7+/ v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/v////7+/v/// /7+/v////7+/v////+fn5/////////////////////////////////////// /////////////////////wAAAP///wAAAP///wAAAP///wAAAP///wAAAP// /wAAAP///wAAAP///wAAAP///wAAAP///2BgYP////////////////////// /////////////////////////////wD///////////////////////////// //////////////////////////////////////////////////////////8g ICD///8AAAD///8AAAD///8AAAD///8AAAD///////////////////////// //////////9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9A QED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED/ //9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9A QED///////////////////////////////////+fn5////9AQED///9AQED/ //9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///// //////////////////////////////////////////////////8A//////// //////////////////////////////////////////////////////////// ////////////////////////v7+/////v7+/////v7+/////v7+/////5+fn ////////////////////////////39/f////AAAA////AAAA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//// AAAA////AAAA////AAAA//////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////AP////////////////////////////////////////////// /////////////////////////4eHh////0BAQP///0BAQP///0BAQP///0BA QP///0BAQP///0BAQP///0BAQP///0BAQP////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////wAAAP///wAAAP///wAAAP// /wAAAP///wAAAP///wAAAP///39/f/////////////////////////////// /////////////////////////////////wD///////////////////////// //////////////////////////////////////////////////8AAAD///8A AAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///////// //////////////////////////////////////////////////9AQED///9A QED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED/ //9AQED///9AQED///9AQED///9AQED///////////////////////////// //////////////////////////////////////////////////////+3t7f/ //9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///////////// //////////////////////////////////////////////////////8A//// //////////////////////////////////////////////////////////// ////////19fX////v7+/////v7+/////v7+/////v7+/////v7+/////v7+/ ////v7+/////v7+///////////////////////////////////////////// ////////f39/////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////v7+///// //////////////////////////////////////////////////////////// ////////////39/f////v7+/////v7+/////v7+/////v7+/////v7+///// v7+/////v7+///////////////////////////////////////////////// ////////////////AP////////////////////////////////////////// /////////////////////////5+fn////0BAQP///0BAQP///0BAQP///0BA QP///0BAQP///0BAQP///0BAQP///0BAQP///1hYWP////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////wAAAP///wAAAP///wAAAP// /wAAAP///wAAAP///wAAAP///wAAAP///2BgYP////////////////////// /////////////////////////////////////wD///////////////////// //////////////////////////////////////////////////8AAAD///8A AAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD/ //////////////////////////////////////////////////8AAAD///8A AAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD/ //8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///////////////////// //////////////////////////////////////////////////////+fn5// //9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///// //////////////////////////////////////////////////////////8A //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//// QEBA//////////////////////////////////////////////////////// ////////////9/f3////v7+/////v7+/////v7+/////v7+/////v7+///// v7+/////v7+/////v7+///////////////////////////////////////// ////////////////////AP////////////////////////////////////// /////////////////////////////////////////////////0BAQP///0BA QP///0BAQP///0BAQP///0BAQP////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////wAAAP///wAAAP///wAAAP// /wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///7+/v/////////////// /////////////////////////////////////////wD///////////////// //////////////////////////////////////////////////////////// ///////f39////8AAAD///8AAAD///8AAAD///8AAAD///9gYGD///////// //////////////////////////////////////////////////////////// //////////////9AQED///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD/ //8AAAD///8AAAD///9gYGD///////////////////////////////////// ///////////////////////////////////////////////////////n5+f/ //9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9A QED///////////////////////////////////////////////////////// //8A//////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////QEBA////QEBA////QEBA ////QEBA////QEBA////QEBA////QEBA////QEBA//////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////AP////////////////////////////////// /////////////////////////////////////////0BAQP///0BAQP///0BA QP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////wD///////////// //////////////////////////////////////////////////////////9/ f3////8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD/ //8gICD///////////////////////////////////////////////////// ///////f39////8AAAD///8AAAD///8AAAD///8AAAD///8AAAD///8AAAD/ //8AAAD///8AAAD///8AAAD///8AAAD///////////////////////////// //////////////////////////////////////////////////////////// //////////+/v7////9/f3////////////////////////////////////// //////////////////////////////////////////////////////////// //////8A//////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////QEBA////QEBA////QEBA////QEBA ////QEBA////QEBA////QEBA////QEBA////QEBA////QEBA////WFhY//// //////////////////////////////////////////////////////////// ////////////////////////////////////AAAA////YGBg//////////// //////////////////////////////////////////////////////////// ////////////////////////////AP////////////////////////////// //////////////////////////////////////////////////////////// /////////////39/f////9/f3/////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////+/v7////7+/v/////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////7+/v/// /39/f/////////////////////////////////////////////////////// /////////////////////////////////////////////////wD///////// //////////////////////////////////////////////////////////// //////////////////////////////8gICD///8AAAD///////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////8AAAD///8gICD/ //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////8A//////////////////////////////////////////////// ////////////////////////////////////////////////////////v7+/ ////7+/v//////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////z8/P////QEBA//////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////AP////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////wD///// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ///////////Hx8e/v7+/v7+/v7+/v7/Hx8f///////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ///////////Pz8+/v7+/v7+/v7+/v7+/v7/f39////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////8A//////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////v7+/f39/SEhIEBAQAAAAODg4QEBAQEBAQEBAQEBAODg4AAAA EBAQSEhIf39/v7+///////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////9/f3r6+vf39/QEBAEBAQAAAAMDAwQEBAQEBAQEBAQEBAQEBAICAg AAAAICAgUFBQf39/x8fH//////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////AP////////////////////// //////////////////////////////////////////////////////////// /////////////////////9/f33h4eCAgIEBAQH9/f7e3t+/v7/////////// /////////////////////+/v77e3t39/f0BAQCAgIIeHh9/f3/////////// //////////////////////////////////////////////////////////// /////////////////////7e3t2BgYBAQEFBQUH9/f7+/v+/v7/////////// /////////////////////////9/f36+vr39/fzg4OCAgIHh4eNfX1/////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////////wD/ //////////////////////////////////////////////////////////// ///////////////////////////////////n5+dwcHAgICCHh4ff39////// ///////////////////////////////////////////////////////////f 3994eHggICBwcHDv7+////////////////////////////////////////// //////////////////////////////////+vr68wMDBISEifn5/39/f///// //////////////////////////////////////////////////////////// ///f39+Hh4coKChQUFDPz8////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////8A//////////////////////////////////////// ////////////////////////////////////////////////////j4+PKCgo j4+P//////////////////////////////////////////////////////// ////////////////////////////////j4+PICAgj4+P//////////////// ////////////////////////////////////////////////x8fHMDAwUFBQ z8/P//////////////////////////////////////////////////////// ////////////////////////////////////r6+vMDAwUFBQ7+/v//////// //////////////////////////////////////////////////////////// ////////////////////////////////////////AP////////////////// //////////////////////////////////////////////////////////// //////f392hoaHBwcO/v7/////////////////////////////////////// /////////////////////////////////////////////////////////+/v 73BwcGhoaP////////////////////////////////////////////////// /////5eXl0BAQM/Pz/////////////////////////////////////////// //////////////////////////////////////////////////////////// /////6+vrzAwMM/Pz/////////////////////////////////////////// //////////////////////////////////////////////////////////// /wD///////////////////////////////////////////////////////// ///////////////////////v7+9ISEiXl5f///////////////////////// //////////////////////////////////////////////////////////// //////////////////////////+Xl5dISEj39/f///////////////////// //////////////////////9oaGhoaGj39/f///////////////////////// //////////////////////////////////////////////////////////// ///////////////////////////////f3984ODi3t7f///////////////// //////////////////////////////////////////////////////////// //////////////////////8A////AAAAAAAAAAAA////////AAAAAAAAAAAA ////////////////////////////////////////////SEhIt7e3//////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////p6enWFhY ////////////////////////////////////////cHBwh4eH//////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////9/f3 SEhIv7+/////////////////////////////////////////////////AAAA AAAAAAAA////////AAAAAAAAAAAA////////////////AAAAAP////////// /wAAAAAAAP///////////wAAAP////////////////////////////////// /3h4eIeHh/////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////3BwcI+Pj////////////////////////////////5eX l2hoaP////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////+/v7zg4ONfX1/////////////////////////// /////////////wAAAP///////////wAAAAAAAP///////////wAAAP////// /////wD///////////8AAAAAAAAAAAD///////////////8AAAD///////// ///////////////////Pz88wMDD///////////8AAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAD///////////////////////////////////////////// //////////////////////////////////////8AAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAD////Pz88wMDD///////////////////////////////////// //////////////////////////////////////////////////8AAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAD////////////////////////Pz88wMDD///// //////////////////////////////8AAAD///////////8AAAAAAAAAAAD/ //////////////8AAAD///////8A////////////AAAAAAAAAAAAAAAA//// ////////AAAAAAAAAAAA////////////////////UFBQr6+v////AAAAAAAA v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/AAAAAAAA//////////////// ////////////////////////////////////////////////AAAAAAAAAAAA v7+/a2trVFRUv7+/v7+/v7+/v7+/v7+/AAAAAAAAAAAA//////////////// //////////////////////////////////////////////////////////// AAAAAAAAAAAAv7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/AAAAAAAAAAAA//// ////////////UFBQr6+v////////////////////////AAAAAAAAAAAA//// ////////AAAAAAAAAAAAAAAA////////////AAAAAAAAAAAAAP////////// /wAAAAAAAAAAAP///////////////////////wAAAP///////////////+fn 5yAgIPf39wAAAL+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/ v7+/vwAAAP////////////////////////////////////////////////// /////wAAAL+/v7+/v7+/v7+/v7m5uRISErOzs7+/v7+/v7+/v7+/v62trRgY GLm5uQAAAP////////////////////////////////////////////////// /////////////////wAAAL+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/ v7+/v7+/v7+/v7+/vwAAAP///////////8fHxzg4OP////////////////// /////wAAAP///////////////////wAAAAAAAAAAAP////////////////// /////wAAAAD///////////8AAP8AAAAAAAAAAP////////////////////// //8AAAC/v7+/v7+/v7+Dg4M8PDy/v78AAAC/v7+/v7+/v7+/v78AAAAAAAAA AAAAAAAAAAC/v7+/v7+/v7+/v78AAADPz8+/v7+/v7+/v7+/v7+/v7+/v7+/ v7+/v7+/v7+/v7+/v7+/v7+/v78AAAC/v7+/v7+/v7+/v78AAAAAAAAAAAAA AAAAAAAAAAC/v7+Dg4M8PDy/v78AAAC/v7+/v7+/v7+/v7+/v7+/v7+/v7+/ v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v7+/v78AAAC/v7+/v7+/v7+/v78A AAAAAAAAAAAAAAAAAAAAAAC/v7+/v7+/v7+/v78AAAC/v7+/v7+/v7+/v78G Bga5ubm/v7+/v7+/v7+/v78AAAD///////////////////////8AAP8AAAAA AAAAAP////////////////////////8AAAAAAAAAAAD///////////////// AAAAAAAAAAAA////////////AAAAQEBAQEBAQEBAJCQkHBwcUFBQAAAAAAAA AAAAAAAAAAAAAAAAAAAAv7+/v7+/v7+/AAAAAAAAAAAAAAAAAAAAcHBwQEBA QEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAAAAAAAAAAAAA AAAAAAAAAAAAAAAAd3d3v7+/v7+/AAAAAAAAAAAAAAAAv7+/AAAASEhIQEBA QEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBAQEBA AAAAAAAAAAAAAAAAAAAAAAAAAAAAv7+/v7+/v7+/AAAAAAAAAAAAAAAAv7+/ AAAAQEBAQEBAQEBAQEBADAwMNDQ0QEBAQEBAQEBAQEBAAAAA////////AAAA AAAAAAAAAAD/////////////////AAAAAAAAAAAA////////////AAAAAAAA AAAAAP///////////////wAAAAAAAAAAAAAAAP///////wAAAP////////// /39/f39/f////wAAAL+/vwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AL+/v7+/vwAAAP////////////////////////////////////////////// /////////wAAAL+/vwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAE5O TnFxcQAAAAAAAP////////////////////////////////////////////// /////////////////////wAAAL+/vwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAL+/v7+/vwAAAAAAAP///////////////0BAQL+/v/////////// /////wAAAP///////wAAAAAAAAAAAAAAAP///////////////wAAAAAAAAAA AAAAAP///////wAAAAAAAAAAAAAAAP8AAP8AAP8AAP8AAAAAAAAAAAAAAAD/ //8AAAD///////////////+Xl5doaGj///8AAAC/v7+/v7+/v78AAAAAAAAA AAAAAAAAAAAAAAAAAAC/v7+/v7+/v78AAAD///////////////////////// //////////////////////////////8AAAC/v7+/v7+/v78AAAAAAAAAAAAA AAAAAAAAAAAAAAD///9ra2tUVFS/v78AAAD///////////////////////// //////////////////////////////////////////8AAAC/v7+/v7+/v78A AAAAAAAAAAAAAAAAAAAAAAAAAAD///+/v7+/v7+/v78AAAD///////////// //8YGBjn5+f///////////////////8AAAD///8AAAAAAAAAAAAAAAAAAP8A AP8AAP8AAP8AAAAAAAAAAAAAAAD///8AAAAAAAAAAAAA//////////////// ////////AAAAAAAAAAAAAAAAAAAA////////////////v7+/QEBA//////// AAAAAAAAAAAAAAAAAAAA////AAAA////AAAAAAAAAAAAAAAAAAAA//////// ////////////////////////////////////////////////////////AAAA AAAAAAAAAAAAAAAAAAAAAAAAv7+/AAAAAAAAAAAAAAAAAAAAAAAA//////// //////////////////////////////////////////////////////////// ////////AAAAAAAAAAAAAAAAAAAAAAAAAAAAv7+/AAAAAAAAAAAAAAAAAAAA AAAA////////////////7+/vEBAQ////////////////////////AAAAAAAA AAAAAAAAAAAA////////////////////////AAAAAAAAAAAAAAAAAAAAAP// /////////////////////////////////////wAAAP////////////////// //////f39yAgIOfn5////////////wAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAP////////////////////////////////////////////////////// /////////////////////////wAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA APf39yAgIOfn5/////////////////////////////////////////////// /////////////////////////////////////wAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAP///////////////////////////6+vr1BQUP////////// /////////////////////wAAAP////////////////////////////////// /////wAAAP///////wAAAAD///////////8AAAAAAAD///////////8AAAD/ //////////////////////////////9/f39/f3////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////9wcHCP j4////////////////////////+Pj49wcHD///////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //8wMDDPz8////////////////////////////////////8AAAD///////// //8AAAAAAAD///////////8AAAD///////////8A////AAAAAAAAAAAA//// ////AAAAAAAAAAAA////////////////////////////////////5+fnMDAw 5+fn//////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////39/fKCgo9/f3////////////////////////9/f3KCgo39/f //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////n5+fYGBg//////////////////////////////// ////////////AAAAAAAAAAAA////////AAAAAAAAAAAA//////////////// AP////////////////////////////////////////////////////////// /////////////////7+/v0BAQP////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////f390BAQMfHx/////////////////// /////////////9fX1zg4OO/v7/////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////+/v7////////////8fHx0BAQPf39/////////// /////////////////////////////+/v79fX1/////////////////////// /////////////////////wD////////////////////////////v7+////// //////////////////////////////////////////+Pj49gYGD///////// //////////////////////////////////////////////////////////// ///////////////////////////////////////////////39/dYWFinp6f/ //////////////////////////////////////+3t7dAQEDv7+////////// //////////////////////////////////////////////////////////// ///////////////////////////////////////////Hx8dXV1fv7+/Hx8c4 ODjn5+f///////////////////////////////////////////+SkpJvb2// //////////////////////////////////////////8A//////////////// ////////////n5+fr6+v//////////////////////////////////////// ////////j4+PQEBA7+/v//////////////////////////////////////// //////////////////////////////////////////////////////////// ////7+/vSEhIn5+f////////////////39/f//////////////////////// ////t7e3MDAwz8/P//////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////Q0NDVVVVQEBA39/f//////////////////////////////////// ////////////b29vZ2dn//////////////////////////////////////// ////AP///////////////////////////6CgoCgoKP////////////////// /////////////////////////////////7+/vzAwMK+vr/////////////// //////////////////////////////////////////////////////////// /////////////////////6+vrygoKL+/v////////////////////0FBQf// /////////////////////////////9/f3zg4OHBwcO/v7/////////////// //////////////////////////////////////////////////////////// /////////////////////////8/Pz0hISB0dHbGxsf////////////////// /////////////////////////////////4eHh39/f/////////////////// /////////////////////////wD///////////////////////////////8g ICDf39/////////////////////////////////////////////////////f 399QUFBAQEDHx8f///////////////////////////////////////////// ///////////////////////////////////Hx8dAQEBQUFDn5+f///////// ///////////n5+cEBAT39/f////////////////////////////////39/eP j48gICB/f3/v7+////////////////////////////////////////////// ///////////////////////////////////////f399gYGAwMDC3t7fv7+8w MDDf39/////////////////////////////////Pz8////9/f3////97e3vv 7+9/f3////9/f3////////////////////////////////8A//////////// ////////////////////eHh4h4eH//////////////////////////////// ////////////////////////////v7+/ODg4KCgoj4+P5+fn//////////// ////////////////////////////////////////////39/fh4eHKCgoODg4 v7+/////////////////////////////////AAAA//////////////////// ////////////////////////7+/vf39/EBAQUFBQn5+f7+/v//////////// ////////////////////////////////////////////////39/fj4+PMDAw ICAgn5+f////////////19fXODg47+/v//////////////////////////// ////AAAA////AAAA////AAAA////AAAA////f39///////////////////// ////////AP///////////////////////////////8fHxzg4OP////////// //////////////////////////////////////////////////f39////9fX 13BwcBgYGCgoKGBgYJeXl7+/v7+/v/f39/////////f397+/v7+/v5eXl2Bg YCAgICAgIHh4eNfX1////////////////////////////////////////wAA AP///////////////////////////////////////////////////////6+v r2BgYBAQEDAwMGBgYJeXl7+/v7+/v+/v7////////////9fX17+/v7+/v4eH h1BQUBgYGCAgIHBwcM/Pz////////////////////////7e3t0hISP////// /////////////////////8/Pz////39/f////39/f////39/f////39/f/// /////////////////////////////wD///////////////////////////// //////8gICDf39////////////////////////////////////////////// ///v7+9SUlL39/f////////////////X19efn59oaGhAQEBAQEAICAgAAAAA AAAICAhAQEBAQEBoaGifn5/f39////////////////////////////////// //////////////////8AAAD///////////////////////////////////// ///////////////////////////////Pz8+fn59oaGhAQEBAQEAQEBAAAAAA AAAAAAAoKChAQEBAQEB4eHivr6/n5+f///////////////////////////// //////////+Hh4d4eHj///////////+fn5////9/f3////9/f3////9/f3// //9/f3////9/f3////9/f3////9/f3////////////////////8A//////// ////////v7+/////AAAA////AAAAeHh4AAAA////AAAA//////////////// ////////////////////////d3d3VVVV//////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////AAAA//////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////WFhYp6en////////////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////ICAg//// ////////////AP///////////////////0BAQP///0BAQP///zIyMjg4OEBA QP///5+fn////////////////////////////////8fHxzg4OP////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /wAAAP////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////f3 9zg4OM/Pz////8/Pz////7+/v////7+/v////7+/v////7+/v////7+/v/// /7+/v////7+/v////////////////////wD///////+/v7////+/v7////+/ v7////+/v7////9CQkL39/e/v7////+/v7////+/v7/////////////////v 7+8wMDDf39////////////////+/v7////+/v7////+/v7////+/v7////+/ v7////+/v7////+/v7////+/v7////+/v7////+/v7/////39/f///////// //////////////////////8AAAD///////////////////////////////// //////////////+/v7////+/v7////+/v7////+/v7////+/v7////+/v7// //+/v7////+/v7////+/v7////+/v7/////X19f///////////////////// ///////////////////////f3984ODjn5+f///////////////////////// //////////////////////////////////////////////////////8A//// QEBA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////////////////////UFBQr6+v////////////////QEBA////AAAA//// AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA ////AAAA////////////////////////////////////AAAA//////////// ////////////////////////////////n5+f////AAAA////AAAA////AAAA ////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA////AAAA//// ////////////////////////////////////////////////x8fHQEBA9/f3 //////////////////////////////////////////////////////////// ////////////////AP///////0BAQP///0BAQP///0BAQP///0BAQP///0BA QP///0BAQP///0BAQP///0BAQP///////////5+fn2BgYP////////////// /////////0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BA QP///0BAQP///0BAQP///0BAQP///+fn5/////////////////////////// /////wAAAP///////////////////////////////////////////////0BA QP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP///0BAQP// /0BAQP///0BAQP///4eHh/////////////////////////////////////// /////////////5eXl2hoaP////////////////////////////////////// /////////////////////////////////////wD///////////////////// ///////////////////////////////////////////////////////X19cw MDD39/f///////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////8AAAD///////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////9wcHCPj4////////////// //////////////////////////////////////////////////////////8A //////////////////////////////////////////////////////////// ////////////9/f3MDAw19fX//////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////AAAA//////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////QEBAv7+///////////////////////////////////////////////// ////////////////////AP////////////////////////////////////// /////////////////////////////////3BwcI+Pj/////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////wAAAP////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////+/v7zAwMN/f3/////////////////////// /////////////////////////////////////////wD///////////////// //////////////////////////////////////////////////+vr69QUFD/ //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////8AAAD///////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ///////////////////////////////////////////////////X19cwMDD3 9/f///////////////////////////////////////////////////////// //8A//////////////////////////////////////////////////////// ////////5+fnKCgo7+/v//////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////AAAA//// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////39/f////////////////f39/////f39/////f39///// ////////////////////////AP////////////////////////////////// /////////////////////////////0BAQL+/v/////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////wAAAP////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////2BgYP// /wAAAP///wAAAP///9/f3////////////////////////wD///////////// //////////////////////////////////////////////9/f39/f3////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////+Pj4////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////+/v7////+/v7////+/v7////////////////////// //////8A//////////////////////////////////////////////////// ////x8fHODg4//////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////QEBA////QEBA////QEBA////QEBA////QEBA//// QEBA////QEBA////QEBA////////AP////////////////////////////// /////////////////////+/v7zAwMN/f3/////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////wAAAP///wAAAP///wAAAP///wAAAP////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////39/f////wAAAP///wAAAP// /wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///////////wD///////// //////////////////////////////////////////9QUFCvr6////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////+Hh4f///9AQED///9AQED///9A QED////Pz8////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //+/v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/v7////+/ v7////////8A//////////////////////////////////////////////// ////39/f//////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////QEBA////QEBA////QEBA////QEBA//// QEBA////QEBA////cHBw////////////AP///////////////////////wAA AP///wAAAP///wAAAP////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////5+fn/// /wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAA AP///wAAAP///9/f3/////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////7+/v////wAAAP// /wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///////////////wD///// //////////////+fn5////9AQED///9AQED////Pz8////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////9AQED///9AQED///9AQED///9AQED///9AQED///9A QED///9AQED///9AQED///9AQED///9AQED///////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////8A//////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////39/f////v7+///// v7+/////v7+/////v7+/////v7+/////v7+/////v7+/////7+/v//////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////AP///wAAAP///wAAAP///wAA AP///wAAAP///wAAAP///wAAAP///wAAAP///wAAAP////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAA AP///wAAAP////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////////////////////////////wC3 t7f///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED/ ///Pz8////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////+fn5////9AQED///9AQED///9AQED///9A QED///9AQED///9AQED///9AQED////Pz8////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////8A////9/f3////v7+/////v7+/////v7+/////v7+/ ////v7+/////v7+///////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////7+/v////v7+/////v7+///// v7+/////v7+/////v7+/////v7+/////v7+/////v7+/////v7+/////v7+/ //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////AP///////wAAAP///wAA AP///wAAAP///wAAAP///wAAAP///wAAAP///yAgIP////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAAAP///wAA AP///wAAAP///wAAAP///wAAAP////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /wD////n5+f///9AQED///9AQED///9AQED///9AQED///9AQED///9AQED/ //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ///////////////////f39////9/f3////9/f3////9/f3////9/f3////9/ f3////9/f3////9/f3////9/f3////9/f3////9/f3////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////8A//////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////AP////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////wD///////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////8A//////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////AP////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////wD///////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////8A//////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////AP////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////wD///////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////8A//////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////AP// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////wD///////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////8A//////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// AP////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////wD///////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////8A//////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////AP////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////wD///////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////8A//////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////AP////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////wD///////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////8A//////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////AP////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// /////////////////////////////////wAOAAAAFAAAAAAAAAAQAAAAFAAA AAzsf4xLVo9ogwAkjzaPSI+Oj2yPfo/aj6KPlLh/WEt/DYDB5XAIkX8BBVyR 6n+CbVKfHo8snz6fAISfYp90n9CfmJ+GBd4zuJ9+YEPzCVAHyQZQoW+oCfMJ WKFQBzAt4JJFTq9gozABBAEUApjXChwJBAETm6DkCcTMP1QLBe02MCmYD2EA /24H9AhzBw0J/4cHJgmWBz8J/6AHTgm0B2IJ/8MHZwnXB3EJ9esRsP8RsBMI cQn/IghnCTYIYgn/RQhOCVkIPwlVaCmwfCWwiyGwmh2wVbMdsMcdsNEdsOoh sP3+JbANCU4JHAl/PwkrCSYJOlOwd0QJ9Cmw9AhdZbD1bGWwdkuwhQngCP+U CcwImQm9CH+eCakIowmQh7BVfIewaIOwT3+wO3uwVSx3sBhzsA5vsARrsF0E KbAECESpsDoTsH8rCdIHHAm5+aD/qgf+CJYH6gh/kQfRCIcHx9GwVbPRsJrR sIvNsHzJsP9oCKoHWQi5B1VF6bA2ybAizQECAAACAgAAAwIAAAQCAAAFAgAA BgIAAAcCAAAIAgAACQIAAAoCAAALAgAADAIAAA0CAAAOAgAADwIAABACAAAR AgAAEgIAAP7///////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////Vmlz aW8gKFRNKSBEcmF3aW5nDQoAAAAAAAAGALB4AAAAhAEAFAAAAMRHDxA5dwAA dwEAAFIAAAAAAO1o6vEDGeny////gwD/9vL/8/vwBwH48QCJgOrxFgSAFgAg ABkAgP8AwMDAAObm5v8Azc3NALOzs+8AmpqaIQCAAGb/ZmYATU1NADP/MzMA GhoaAAAHMwAAGAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAYAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAPUY6/D/8vA6AwAAATjo8/r0 Bw8ZD+H69vHq8fUZ4/gM6/ABAFQA/1oAIAAAQXJpa2FsQAIBRguASlYL9QLn 9BZOBhAAAFT/aW1lcyBOZXc/IFJvbWFuQAI6BPiUAVADgwFXaW5nZKrCAHNA AgTn9BVOBBD9AoMATW9ub3R5f3BlIFNvcnTJA+UF5/QNtQaVAFN5bQ9ib2wA NVTr8D/o89A/8v8EBtrq8SDo81lAGgYAABkBLQAxDwAAVRrr8P/y8BLr8Bzo 86D69AcP6/D28erxSuf0BDrr8ETr8AEAVDEB6POA8vFFD1cP8vHg+y0B6vFo 6uf0DuvwUOvwAgBUhRjr8Djf/IwBjAHm9QGukQIAAAMqBAXr8Aaq6/AH6/AI 6/AJ6/AKquvwC+vwDOvwDevwk3rj+C/AAlUBAQG6AX7fAQIAYgEA/gkS9QEP FgIPFgNiAADT/oW4BOrxmfwDMHp/FK5H4XqEP7YDKUDm9b8BAjQTEJECRBaQ EhUuEB0XZxIDUxYSEwSrYgIRFAV9FgZ9FgeyfRYIoRFnEgkhkhaGBLwE6vGV /ARrBjUT8BMGH0B8HCcXmxQQFaoSZxIG0xdKZxIHSywISyzSGIfABALq8en8 A0wT6/CYL6omNRPOqiTgPyDHIPLw70FQ9BhnE6omihMB5SwC5SwBA+UnMS9D JcImVCTLJsIVqn4VCX0WCkEmC9MXoAJ5COD8AwsPuQMJFH4VfBkAfhUsKX4V qh+8G1o/bDIJElULfRYMfRYNfRYOfRZVD30WEH0WEX0WEn0WpRMtEaTIBOrx 7Y4//XwBJdkDAgBhAP6eQlUBpEUCpEUDpEUEpEUVBaRFBqRFB6RA/D8OTqQz T0VKYC4QnkIRpEUSqqRFE6RFFKRFFaRFFiqkRRekRRikQITMBOrxSRGBTzUS o9AE6vFG/AOAk0GrWZw7CRMoURMTz1YCCs9WAyhRp9QEYxWoVMdZabfYBOrx 9/wDICBTFowpb1MWCAg7b1hvBicgQBEWfWD/I5EXix1nEgSQbB0FkRwGYgjw NtFgZSOqkGwJkGwKkGwLkRe99NwE6vGxgVnwP4gTA4gTlj+JXAkX7yPTFEA1 VQJufANLLARLLAVLLEhKL5Ib60K+4ATq8VH8BEECeFVPfwwbohCwTQHBRSQy hNBBv+QE6vHmgV9kjwA0FZ5PsE/CT9RP5k8OTyBPEAhfRE9WT2hJwOgEM3K7 AFhWhbsBUxTAP1ud0GyeWdh+nmKYc1WkkeWilgBmj8SfdI+Gj5iPqo+8j86P QOCP8oEQFf6PEJRnEhCRlkqKExGArBJtlooTE6CsqRR/looTFcCsFoCsFyaA rBggo5WKExkAtwkSVRp9Fht9Fhx9Fh0tEQFpuAS7AfrxlAR8BbMJNRXK6vGU 4/jMU49lECcADTCkklXFTBJZebM/xT8E1zSoevB1FO0yVSOdtuqTVQjolQno lQrolQvolakM6JB6ozC/exHTF2pAvARKv1y/4/jpEeb1yIa3bKovr8YhM8LC 878+xkI+xgGSRNgv6i/8L/6AUUyMcMLEPxOpZD7MBT7HEOg0EBW+GChRa8AE YM9yzwLj+Ijj+IBZx1UtESUBuwUAMQ/i3+jSwwH23eD74RWEBQItASiRB2cR cA2B2bsJ7wlBLvwPDh8gFP2WMRktPh8QXmlcGMO33Rkj6h1V5AEqBMiyhSmb ki80/+L5wi/ULwDyzwTfCj/RP+MyWMFGv4DfBIDPpNZTnM8ZD8DP0scuEADf yeg3xdFcxdHfhg/13wfvCDIYJOVu0SQx50qFpdVu0QCuCcnVZ+9574vvne+v 78HvANPv5e/37wn/wtRu1XUP2x8A4RLXFfMZsg8g6XMRZPAz5QD1VWsb9QsH HxkfKx/7mB7xfMAFHvNVMhzHca0gA6w/rC++JsADW/9t/r4mFIf0+SwC+SwD +SfF0SzhAM3f2B+VD6cPFe+o0nhRPsEANOQ34Y65eFH3eVvpLOF4QSBYL2ov fC89H6jSR6YkQ+jD/wDvpXEa8ySv+wFihQQJRheGH5gfqh8c+3FApi+4IyMP HOPWL239Aj7MFQNC3AQegaRM6Ckh7TAQSClSA8AC65QBRYLZManRAOch7TCc VX9f1p8xhWFR9JQhAUG/3f/v/3e7PYa5OkJAl7RclQWLaUL7tIGmyWZYBa1A MlBDlSjl2DLZMb0yALk1eFJ4UXhRNCI0ITQhL8EA4ZIvwSJyOLE4seZy5nHm cUBKgkqBSoFDkkORQ5EOqdEE1GHUYQ+p0eFh4WE74jvhBDvhG9EAxdG9MTs/ M+BWP4DtH/8fhj8q7zzvTu9g7wABJeo//D8ORJkvwE/1f88vgPZP8y8FPxc/ OodIEVd1M/5VFE8b6LSBTmsCF4MBQE929YeGd0J4NYROAMADm0vBcrlfy1/d X3i6gIQEjL7/dLylt/uzS5bE0nhRAMYfQjAjf5CTvTGenQmfGS8AKy89L08v q3jxg7x+wLpIGQKAhXtdGZWPp4dTpcTS45EAdZ+Hn5mfq5+9n8+f4Z/znwAF rxevKa87r4KPX6+mjzEzAC/Bja+fr18/cT/Vr+ev+a8AC78dvzeR9IA2t2FH CEeFFADABYREjV3djbsUYNGlv7e/ALGvw6/tv/+/Ec8jzzXPN5IQt3lhR0q3 13xR53+N3wuEAQKMQRePkVjC9C9QaQG2YQCtz7/Pyb/bv/XPB98Z3yvfAD3f L79h31O9VBVTpYtCeL9AI2fDcGkB3wEi8T88A/ffAAnvG+/LBjbvSO9a72zv SN+AWt9s337fmd8Fj7Tfxt0EgOyP/o8Qn+tfNJdOB0iRckLc8sFTnn8FWsPB cmrV5CgrDz0PPZiVPZhTNJfTTzYBFiEzFBLzvw5WDlYIw3CMRMzXBunlU2H1 7wf/ABn/K/89/0//Yf9z/4X/l/8Aqf+7/83/MC85ItZKAw8VCACiwKoBJw/A D9IP5A//DwgfABofTVc2G6LAIwGuD5MvSw8AXQ9vD4EMvHeQDQBx3WFPHwBh H3MfhR+XH6kfux/NH98fAPEfAy8VLycvik9LL10vbysAIw8sPz4/UD9iP3Q/ hj+XBgCqD+lP+0+zL8Uv1y/pLygfADoe6mGpP7s/zT/fP/E/A08AFU8nTzlP S09dT29PgU/kbwDx/7dPFQ/bT3Rf/08RX4NpAC9fQVfY0E5fYF9Ff4Rfll8A sV+6X8xf3l/wXHWhA28VbwAnbzlvS29db29vgW+Tb6VvALdvyW/bbz6fpU8R f8lPNX8Azn9Zf2t/fX+Pf6F/s3+YnwCqn+l/+39erx+PMY9Dj52xARFaj2yP fo+Qj6KPtI/GjwDYj+qP/I8OnyCfMp+Vv1afAGifbC+MnyWvN69Jr1uvba8A f6/8Lw4/67/9v7Wfx5/ZnwDrn/2fnQCf0AIAACDc/+H6Af8E//+cADQCQ/// gAUBAAEA6u8KbwhkHwAPAFjdAh4BWAIDHwBBNEAAD0sPCgZfCO/wZQL/egC+ 3v1ESU5VIuvwNJ7v8MOxAsdUD+T3BwCoD8wP3g/wDwIfFB8mHzgfAEofXB9u H4Afkh+kH7YfyB8A2h/sH/4fEC8iLzQvRi9YLwBqL3wvji+gL7IsfRrr8AgA CQAK6/D/ICBEaXNwbGFveQAK3urxBBpVRuvw//LwA+vwROjzZvr0AAPx8g4P ///q8VAfDeD79vHq8Wjn9AHr8C0cUgJUGOvwBN/8VQHUVQHm9ZLj+IPr8AIA 91VJzJgBgCJASXuYmaIAQRZAP+jz28A/qQW/Sejz8D8OuwQkQARFBdUP5w/5 DwhOBfXwZABlcQhZD2sPiQiq5vUKkQMF6/BH6fJvUJECIB8yH3cKqOP4TZED 0f3y8LgRMgMBdhABAdS4EVEC/PbwEZECYAf/TmV0d29yawB9/l0SAWAAAP7Z EgEK3xjrMgHq8SDz8AAAJP7v8pn/FOAD//z/AAA/8AAAB8AvAAABgA8BgxQP JgzhExIBDgEKAQYAH//A/wH//4P/AfCF/wCI/4IAAQ+F/4+CAIX/AgAI/4n4 BQAP//8A/4yPBIAP8AiN//gCD/COjwKAv/CO+ILwhnkAAP6BAfCD+IMAAgj/ 8IMABgj4+PB/8I+FAISPJhIAv4+A8PAID3AA+PcA+PBwAADw8A+7jwCvAA/w ALoAj+63AIT4A68AhAAF3pUAAPCDjyUBA4//AA+EjwOA8Aivg/gHAGkA/+IB G5cP/wDCAP+9APYCgOhoAeoBlgCEWACE/wN5D9wBTwAPh/+GWAAD9P/9Uevw QSBkZXZp/2NlIHRoYXQg/2Nvbm5lY3Rz/yB0d28gb3Igf21vcmUgbmUJAP9y ayBzZWdtZfluBgD98ExheWVy3yAzIG9m+vBlIPdPU0kPAGRlbC4BAADtYuny ZGb18BoSwLvIzP3wdAVA9PRA+vz0wP3xzIASwJj9mR0AQQZAgC3Yn4ItjBZA HQEgAMBV0BUCFBsECjMEEBsEAwLAVUfr8P/y8Arr8ETo80b69AAH6fLy8REN A+/0UCMJ4Pv28erxaAIINOvwLwEAVBjr8Bzf/AkBlAkD6PMB6/AC6/AjAQSq 6/AF6/AG6/Cb4/h/foQAAgBVQNDMsQC/gBJAQJiZugBB3wZAQGRmwwAaIjdA QMixAXQVwAW2AO7LAwVAUHkJzwMWtoUBBHWIAGAg6PPgNz8F/vQCBXUjAP4J ZWCJAQKUAYQBAQSYASAqEpcCKhIlEurxCZMCSBLCQxcEJhVME2YSJRYJBCog EgMmFgUwFgEwFoUSEkgSAUQWoxcFYhatEmYSroUWCQUaiQEHmAEMIloBBpQB 6hPgEgjmH/gUHQYALwEGoIEE6vFgAfisAIAF4vkBEAAAo5SFBOrxgzsj/fLw bCH++HEkSSpjAgIAYBJW/2lzX0Q2MS5j/2htISMyOTI47zIA/kmFAQFgP/9D b3B5cmlnaP90IChjKSAyMP8wMSBNaWNyb/9zb2Z0IENvcv9wb3JhdGlvbn8u ICBBbGwgsiL/cyByZXNlcnY/ZWQuAP6niQTq8d0RZicBAAuLIwEwJwD+hI0E 6vFlOyd8CQMALoESGiaXAgQirRJBM0LwFgHgElA3JRIEIgMEIf8CAoBMAABA /umckQTq8UupBCvYgr8t2DYjQEHdBUD3cRzHsTDxP0DjqziOujADrzXhuDXz VT/lBYCBAxWAMnWGMKtggYQBLYcxG4UBA9bpMnUf/ACANzBAgFVBhzEhhQJy ZhPAGUN3AcCACDBAeg4PRgkBGU8rRiQhFgQiYRx5M0D0BARB/whWIIMUBCIF ChS4/wD+CVkBAgZ1CQBgq3UIBUEihzCiYAGgvQ3r8GIBAKMIMZPu1EATkHWY AGAF/im+lQTq8Rk7JAIXNXoIAQkcNDQTYgNdBP5NGVXm9VBlIT0JrACYAU7p 8m6FBAKQAShaB49B4vmQARc1iAEUhAEzBanj+GJmJ3QhbCH0/wR+BDDiMwBg CyX/UHJvcGVydGmvZXMA/lhSAiAQ/v0n+TJgGE5FVFf/T1JLIFNIQVD/RSBQ Uk9QRVJ/VElFUwCAUYcxBKxRMiXCOyfDX9VTY1EXRJ/wA8CiP8xBE0Fg/xRI aWRlIEVx/3VpcG1lbnQg/0xhYmVsAKMeftlBYBRTaG93bm5xkO9Y/wH+UBFB Y9Ih+3MusGNbMl0AgCYEQED+NzMmFfBNFy8SkvshAcESiAHwq1JdJbMQOyOE AThv1VNWFEPq8Vth1u7xoCVlYRhqYlRl/XjJIW50cm9sIP9IYW5kbGUAo6TG QINgGIdiRH9lm2odAvtTDrBrwWTgEqMW1mN6E5SsUTMFkDsj/MBf0lY6elFr K2BnRWRpdG5vIvbxkyAQnG+ubjPAZY9DDu1FBgBtiQSUAWABXQRUdY+NN8nj +DiFAlSQAUkSpoQyIhOzhF0iFMCEEDMC4zHjMRcyti8oK3GigAHobCHq8Wwl z3HxhTAk9AL/AmAMTWFudWZ/YWN0dXJlcu5UuQD2UfgyAWARHJkgv05hbWUA /rqCBJtgCW9mAP7dgV0lcXDqj/yPMSM6IgJgDuRQ/2R1Y3QgTnVtCWImmzoi AZ+fT59hkjMF4Wtvn4GfMSPjM2ALUBNhcqeX4zIBI6wql9OfhGGSFzWB858F rzEjhXIC+2AToZVEZXNjcuNpcJdyjkKOIAZSbxN1dCaRhXIBoa+zolGvA/5n jQTQQWuP4frQQY6NQZOfgwgx9AH0ATIiF0y0AF0iYAFgATMC9UH1QRcy4BEA 4BFEEvgx+DHtQmQBZAEJAhiFcYVxwUAAAIBhgGERUREABEC2sggxAP8B/wEE IQsAtC8oQnGvBTCpgY4g/whMR0NWSTE1uzYAM0RgF0+yIGn/bmFsIERSVyA/ ZmlsZSBuSpDsIMDRsV0lX2Hhv9BBjiADNtUuEDAjpyMZDcZyZf9sZWFzZSB2 ZQNyc9MhJMMzBetz47/1s/8xMC8yNS85OfsA/vVCAWAPRGF/dGUgR2VuZdAg wOoi0bEXNbVCic64pTUw7zAwNTMwp1VuaX9xdWUgS2V5I8TyRBU80s/kyFN5 bWL7b2zGp1NoYXBlnSChlUxpbiLF7UUz+BnfBTB/MgBgAjE3/LmXRtNJY29u IEnBRCPEeAUY3yrfxqdWTh1FCYhUeXAixaWym4fwi8y6g1Wa9AIBYAxGd00g Q7XAZ29yCNVgsrLr8A5A/9/aVkZNV5j4yKUt4G9nQ2xhc3OgI8S/suv0dKnP cQGdhGD9LsCjLExvZ2lj92FsLDbTcyxSZWxwYOQgLFZvwyxosyCQJpHRscyy 6/APn4P/BCQNQKviAXCRBIgB/wFdBKD01eb1iAFdJYAFquP4UOipBNQGujTA 0w8HAgG27LPAAFVBAgD6DHQY+wGYAbhQBQMzlBASpvFZdKZBmAHkBLejAvoC L2ICAAa3owOy89Px/ojyCGAaUmVwbwNzadIhCY5UQ9dsQT21QrEEK3I/QM32 DeTS8AUO0vADYgVVZsJiqfBVQrCtF48SQT1ZcQIFNQMBScDd9kYH/kbxMiUy mD8EqjRs/gl+9xSSjfMDQtLxAQMk5LLyR+GISMv/3ff1QgcIYA/v+GVxCQl6 E1E/AC0GO0Q9D08CLHRbD9oTYB84hQZDRJUPA/4siLQy8QDDgXlQw4ETvV0l 8zCVhttTAVSHb3xgHSFAVfVDLyNlf4dlAGmi1B1fL1+TiJRwk4hVwiQ7flAn ADKyMhtxvF9hAAHcMd2xP+X8j0F5ULr99IXvHP9PTbKJ9CqUsAN2fH8XAQAA QGRm9PC/GhJAQMjM/fB09wVAQPTxZhoiQB1A/fHMdBUDBfnwDgPXBUBQ5/SA 6/ABwPsDFuvwAgR1AgD3AGAg6PPgPwX+rjcCBXUDPwsaOAEClXQ+AQXr8CA+ AOvw8IZgBQN0VAFuAlQAdgkApGkHWQwBgwdDByTr8ANTAgYxATEBBNEBCTEB 2QJuAerxCQBjAQMDCtEGBdsGA6W86/BA5vXy92YFARrvIkBAzA4BdBVA/wQA CAChFmkNkwAB6vEaAhnr8B8AFv7r8AICdQIAAGC9IOjz8D8F/jACA/N1AzgL KQECAHIJtioBAgA7BAAAWgQB6mIDA2kJEOvwAwIGVOvwYwIAjgIDlAYBv4EB AABAZvPxGv8SQEDfOI7jOP+OA8BAJ/WCHJ/HyBFAQP/w//FAugQEAQwF8z9Q 5/SA9Ovw7/Av6/ACAnVMtwAAYO7xAC3w8GL3CABy6vEDB8AFa4EC6/AJ8PCA QfDwa/4bOAEDPQJ1Hz8A64AuXQYpOAEAcAXu5/QJAMCHBwHAgC0L8PB6Dn0G AYcPmQ7VFjgBBHECIOjz4D9NBdIDBXXeAN0JGlABqwIG6/AMMQEEBRYDqv8C AwUWBb0HQDgBBrN0Bz8AsQF0CDsUgG0i8PCiGOvwoA3r8G9iAQCjoAAAk1oQ ZxOQdAUQPRIF/r1H6/AyHMdx8/Cs4T/y/wQG6PMVBOA/IL4hAP///+9B6PMB ARDc8P3///8BAAsAAAACAGABMAD+3UXc/wAAGuvwAwG1BuvwDOvwAQIJBgMq AwIDCQYACQYB9S/r8CDj+P7////XAAAB6vEY6/ACBfNwBevwBAHwA8CAHx8A AED+vXHr8P3////y8f7c9/Ti+QEACuvwAgDfYAAA/kkRAQFg/z9Db3B5cmln /2h0IChjKSAy/zAwMSBNaWNy/29zb2Z0IENv/3Jwb3JhdGlv/24uICBBbGwg /iYCcyByZXNlcj92ZWQuAP4YAAAA0AAAAAAAAAAAAAAAAAAAAAgAAAAIAAAA AAAAAJsAAAC0+SEQJBYAAJ8AAABHAJgAAADc+iEQwxYAAHEAAABHAJwAAACs +yEQNBcAAO4AAABHAIcAAAA8/SEQIhgAACcAAABHAKcAAACU/SEQSRgAABEA AABFAIQAAAC0/SEQWhgAACEAAABHAKQAAAAM/iEQexgAACcAAABHAKMAAADM /iEQohgAAGgAAABHAAAAAAADAAAAAQAAAAIAAAAEAAAABgAAAAUAAAAHAAAA /ygXAAABAAkA/wADkwsAAAYA64YB6vEF6/AMAmjXDaAWBAcH6/AQBOkB8fAe AQPr8B4ABN7r8AQBDQAYAfwCyP/yLQHr8C3/8PTwAPpFAuT3IkEE8vJNAAYp APJUCwJBBjQDAAC/v1W/PwYDKwIGYQAG//D/JAOBAI8W9wP/ixbEA4AWkQP/ bhZfA1UWLgP/NBb9Ag4WzgL/4BWfAq0VcgL/cxVGAjMVGgL/7hTxAaMUyAH/ UxShAf0TewH/oxNXAUQTNAH/4BITAXcS8wD/ChLWAJkRugD/JRGgAKwQiAD/ MBByALAPXgD/LQ9MAKgOPAD/Hw4vAJQNIwD/Bg0bAHYMFAD/4wsQAE8LDwD/ uwoQACgKFAD/mAkbAAoJIwD/fwgvAPYHPAD/cQdMAO4GXgD/bgZyAPIFiAD/ eQWgAAUFugD/lATWACcE8wD/vgMTAVoDNAH/+wJXAaECewH/SwKhAfsByAH/ sAHxAWsBGgL/KwFGAvEAcgL/vgCfApAAzgL/agD9AkkALgP/MABfAx4AkQN/ EwDEAw8A96cQ/yoEHgBdBDAA/48ESQDABGoA//EEkAAgBb4A/08F8QB8BSsB /6gFawHUBbAB//0F+wEmBksC/00GoQJzBvsC/5cGWgO6Br4D/9sGJwT7BpQE /xgHBQU0B3kF/04H8gVmB24G/3wH7gaQB3EH/6IH9geyB38I/78HCgnLB5gJ /9MHKAraB7sK/94HTwvfB+ML/94HdgzaBwYN/9MHlA3LBx8O/78HqA6yBy0P /6IHsA+QBzAQ/3wHrBBmByUR/04HmRE0BwoS/xgHdxL7BuAS/9sGRBO6BqMT /5cG/RNzBlMU/00GoxQmBu4U//0FMxXUBXMV/6gFrRV8BeAV/08FDhYgBTQW //EEVRbABG4W/48EgBZdBIsWIyoErAFCBSwB8JkEMAECpAElqQ+7D80P3w/x DwMfABUfJx85H0sfXR9vH4Efkx8ApR+3H8kf2x/tH/8fES8jLwA1L0cvWS9r L30vjy+hLy8CdH4PkA8B/fIkA8HPJOkqoyLEQVTBQDQW8LsEDblA4BVOr0B7 3qtCMxXTBaRBoxTVJZtATJdAcpNAlgb/QxO5Bt8S2gZPdxL6BoRBgEEkfUAi eEEvdUBwQWxBp2lAZEEVk2FABV1AdVlAVEFQQSpMQSlJQJlFQAtBQDxBkfc5 QDRBMEFvLUAoQXr4JUAgQRxBJwT6Br+/A9oGWwO5C0CW+gdAcgNATAb7ASXd BvwxawHT8zLxAN176zBOBZHpMGoAB/AESuEw3DHYMdQx0DFXDwD7zzAHCWAa CWBVMwlgUwlgeAlgowlgfdIJYAUFDwA8MWDVdjFgsjFg8DFgMAZXDwBxRWCz RWD1RWBTNgc2YFVgtFVg8FVgXyoIDwBhZWCUZWA1w2Vg7mVgEwkaYHlg1Ux5 YF95YGt5YG8J/xMAogkeANUJ/zAABwpKADgK/2oAaAqRAJgK/74AxgrxAPMK /ysBIAtrAUsL/7ABdQv7AZ0L/0sCxAuhAuoL//sCDgxbAzEM/78DUgwnBHIM /5QEkAwFBawM/3oFxgzyBd4M/28G9AzuBggN/3EHGg33ByoN/38INw0LCUMN /5kJSw0pClIN/7sKVg1PC1cN/+MLVg11DFIN/wUNSw2TDUMN/x8ONw2nDioN /y0PGg2wDwgN/y8Q9AysEN4M/yQRxgyZEawM/woSkAx3EnIM/98SUgxDEzEM /6MTDgz9E+oL/1MUxAujFJ0L/+4UdQszFUsL/3MVIAutFfMK/+AVxgoNFpgK /zQWaApUFjgK/24WBwqAFtUJf4sWogmPFm+JcFVriXBfiXBMiXAziXB9E4lw 7giPFsOlcNWUpXBhpXAqpXDwB1ePFrS5cHa5cDa5cF/1Bo8Ws8lwcclwfTDJ cPAFjxay2XBVdtlwPNlwBdlw0qsgVaOrIHirIFOrIDOrIDUaqyAHqyD7A6wv vicC/vElB18ZXytfPV9PX2FfAHNfhV+XX6lfu1/NX99f8V8AA28VbydvOW9L b11vb2+BbwCTb6Vvt2/Jb9tv7W//bxF/ACN/NX9Hf1l/a399f49/oX8As3/F f9d/6X/7f9NGZAUEAYRWCZoD8HsPOwpas6ABFP7r8CQDCACaCwf/AzkNjgM6 Ee//AQETewIMEhj/AWENGAFBD4HhAYqxDok2twID////ElkGBLcmgrElh7+Z vzO/AEW/uyDSsxKFdr/iv/S/rL8Avr/Qv0zPXs8GzxjPKs88z//hCoMETAn8 A/9uBZsFfwMPBf95BIsGTAmLBg86B/oFEtFuz4DPks8ApMES3yTfzM/ez/DP At9u3wAm3zjfSt9c39jfgN+S36Tf/LbfQMvjA4EBjAX/+gCNCXsCWQv/DQJf CmsDrwU/awOjBwcDmuH23wAI7xrvpMGa76zvVO9m73jvAIrv9u+u78Dv0u/k 72D/CP/wGv8s/z7/QMu7EicG/xwRrgY5DQ8F/1kLmwU/DDgE/xwROAQAD6EE ACIBfv+Q/6L/pMEiDzQP3P8A7v8ADxIPfg82D0gPWg9sDwDoD5APog+0D9xN IhHwRS4T/kLJvQslA2ENrP8DXRENAh8Tmf8CLxJAAX8NQB8BfQ+fAbgRBh8Y HwLeQQWXzrgfyh9kH3YfACMAygemHxIvJC/cGQovhi8uKfxcL0DLDguhBG8J /xoEjAW5BaID/y0FnASuBm8JP64GXQcnBvwh2B8A6h/8H6TB/C8OPzYvSC9a LwDsL1g/ED+iL8A/0j/YL64//xoEnwGvBRgB/7AJmQJ8CysC/4IKjgPcBY4D D8EHJQNAQRw/Lj9APwCkwUBPUk96P4w/nj8wT5xPwFRP5j8EXxZfHE/yT+MS /0UGOhHMBmEN/y0FfAu5BWwM/2AEOhFgBCMPA78EhFFgT3JPhE+08oVfAJde vk/QT+JPdF/gX5hfKl/4SG9ab2BVJwH//x76sfAmaZAxAENvcP95cmlnaHQg qf8gMTk5NiBNaf9jcm9zeXN0Zf9tcyBFbmdpbv9lZXJpbmcgQ/9vbXBhbnkA D/6xZBMAQWxsIFL+v2FzIFJlc2Vyd3ZlZFO0AQADD2QAAQ9xQOv06fbp8gEA Fuvw/wIAdQIAAGAgdub1Bf4GAgF1Aw4LdUDr8Wbz8RoiQO7x7unyAQAW6/AC AHXfAgAAYCDo8/A/uwX+BgIBdQMOBwAHAAX+9UDr8Wbz8RoiQED9zPzxdBVA AQAW/uvwAgB1AgAAYL0g6PPwPwX+BgIBA3UDDgvpQOv06fPM/PF0FUD3AQAW 6/ACAHUCbwAAYCDm9QX+BgL3AXUDDgfwPwX+cUDr9On26fIBABbr8P8CAHUC AABgIHbm9QX+BgIBdQMOCxgAAACcAAAAAAAAAAAAAAAAAAAABgAAAAYAAAAA AAAAiQAAAOQCIhDxIgAAAwAAAEUAigAAAPwCIhD0IgAAIgAAAEcAiwAAAEwD IhAWIwAAMAAAAEcAiwAAAJwDIhBGIwAALwAAAEcAiwAAAOwDIhB1IwAAKwAA AEcAiwAAADwEIhCgIwAAIgAAAEcAAAAAAAEAAAACAAAAAwAAAAQAAAAFAAAA AAAAAAAAAABSb3V0ZXIA/TPd/hAnADIcx21xBAC8P+rxAYD98a7r8AMHBuvw DA0BAhgAAAAaAAAAAAAAAAAAAAAAAAAAAQAAAAEAAAAAAAAAlAAAAPQEIhB1 JAAAIgAAAEcAAAAAACVm6/BA5vXy9iD8/w4O9wEALuvwAgBwBf4rAgAA8APA ohu26/CgD+vwgEzw8KM1DOvwk0YCkP4YAAAAGgAAAAAAAAAAAAAAAAAAAAEA AAABAAAAAAAAAJ4AAABUBiIQzSQAADcAAABHAAAAAADV2OvwROvwiOvwBAA1 Bevw/wAP/wL99BQJquT3AevwBuf0aOvwJP/2IRAKGQAA0O7r8FQADOvwUKpl +wPaRwAXCQAARvsAbOvwTP8hEMLvIwAApOvwVQAO/uvwjAQiEGYkAKsAD+/y aevwpHkAl+p9ADZLAmR1AQUiEAMEJZAFJAFV7uvwSOvwmuvwFfnwfQLp8s4r VgAT6/D9/wwAAACMtwNef9r/BiQAAAH99P8AAFJblU9VXasBaeL5B+f0Ruvw RP/LIRCHCwAAz+7r8NYAI+vwBNAhfxBWDAAABQFAAf0K6/A80SEQWw3nAABZ PwJmAfTPIfMQtG8AYQAARAAryuvw/GsAtW8AQgFGAH1H6/Bk0iEQ+28A6ykI UAFOnQHzIRDfOiUAAIHr8FYAlR7r8ALr8BHz9OnyJQQABPLxAFVG6/D/8vAD 6/BE6PNm+vQAA/HyDg///+rxUB8N4Pv28erxaOf0AevwLRxSAlQY6/AE3/xV AdRVAeb1kuP4g+vwAgC7VUlkAowpQJcDQLcjQD/o88A/qQW/7Uno8/A/uwQk QASARQXVD+cP+Q9OBfXwZABloHEIWQ9rD4kI5vUKkQMFCuvwR+nyb5ECIB8y H3cKFajj+E2RA/3y8LgRMgNNAXYQAQG4EVEC/Pbw/RGRAmAHTmV0d99vcmsA /l0SAWAXAAD+2RIK3xjrhgHq8SDz8AAAeP7v8pH/JMPh////gYD//wCAf/7v AAA//A8DH+AAvwADwAAAAR4BgIWDJA4jHgUaARYBEgQ/7/8AAH8GAsPh2v// ggCC/wPwAP8PiP8I8P//D//wD/8AiP8FD/////Dwg/8BD5+G/wHwhHsAegDw t4b/BnUBAACGAwd4jQOFAYUAAvAAhF4A3IQBXwEBD4WrAoT/+QRuALcEhf8C D/B/hf8E+QAACbgA1s8AAZ96AJ+qAPCCCf8nAN4Ag2AA5QFjAKkA4QyQAHYB jgL7Bv8P/3/wDwCEmQyQ/wEYwgAGGG4A8AD4AoAANhFLD4m1AgFmAIgBAXEA ZwHwh3MDkgEPh2gHDYleBtT//Vjr8EEgc2V0IP9vZiBjb21wdf90ZXJzIHRo Yf90IHdvcmsgdH9vZ2V0aGVyDgD/IHByb3ZpZGX/IGEgV2ViIHP/aXRlIG9y IG/8EwIlAC1iYXNlZP7z8HJ2aWNlLgAA7WLp8pyZ9fAlGcDO9PLZEkD09AMD 2RLtwOnyjBkTA0ATQP+4YAu2YJcdQNskIi0AwiATBBvA1QQdAhUTBBc7BBHA VUfr8P/y8Arr8ETo80b69AAH6fLy8RENA+/0UCMJ4Pv28erxaAIINOvwLwEA VBjr8Bzf/AkBlAkD6PMB6/AC6/AjAQSq6/AF6/AG6/Cb4/h/3oQAAgBVQOny jBn7QECQAQBAE0BA25yZwwAlKcAE2SK4wAW2AMsDEkBQeQnP2wMWhQEEdYgA YCDe6PPgPwX+9AIFdZQjAP4JYIkBApQBhAEBgQSYASoSlwIqEiUS6vEJCJMC SBJDFwQmFUwTZhIlFqsJBCASAyYWBTAWAUgwFoUSSBIBRBajFwViFritEmYS hRYJBRqJAQeKmAEMWgEGlAHqE+ASCHTmH/gUBgAvAQaggQTg6vFgAawAgAXi +QEQAFMAo4UE6vGDOyP98vDibCH+cSRJKmMCAgBg/xJWaXNfRDYx/y5jaG0h IzI5vzAzNAD+SYUBAf9gP0NvcHlyaf9naHQgKGMpIP8yMDAxIE1pY/9yb3Nv ZnQgQ/9vcnBvcmF0af9vbi4gIEFsbP0gsiJzIHJlc2X/cnZlZC4A/qd0iQTq 8RFmJwEAC4sjnwEwAP6EjQTq8WUMOyd8CQAugRIaJpcCBCIIrRJBM/AWAeAS UDclEgQivQMEIQICgEy8AP7pnJEE6vFLqQRjC7a/YAtCKkBB3QVA93Ecx7Ew 8T9A46s4jrowA6814bg181U/5QWAgQMVgDJ1hjCrYIGEAS2HMRuFAQPW6TJ1 H/wAgDcwQIBVQYcxIYUCcmYTwBlDdwHAgAgwQHoOD0YJARlPK0YkIRYEImEc eTNA9AQEQf8IViCDFAQiBQoUuP8A/glZAQIGdQkAYKt1CAVBIrwAomABoL0N 6/BiAQCjCDGT7tRAE5B1mABgBf4pvpUE6vEZOyQCFzV6CAEJHDQ0E2IDXQT+ TRlV5vVQZSE9CawAmAFO6fJuhQQCkAEoWgePQeL5kAEXNYgBVIQBMwWp4/hi Zif8bST0/wR+BDDiMwBgCyX/UHJvcGVydGmvZXMA/lhSAiAQ/v0n+TJgGE5F VFf/T1JLIFNIQVD/RSBQUk9QRVJ/VElFUwCAUYcxBKxRMiXCOyfDX9VTY1EX RJ/wA8CiP8xBE0Fg/xRIaWRlIEVx/3VpcG1lbnQg/0xhYmVsAKMeftlBYBRT aG93bm5xkO9Y/wH+UBFBY9Ih+3MusGNbMl0AgCYEQED+NzMmFfBNFy8Skvsh AcESiAHwq1JdJbMQOyOEAThv1VNWFEPq8Vth1u7xoCVlYRhqYlRl/XjJIW50 cm9sIP9IYW5kbGUAo6TGQINgGIdiRH9lm2odAvtTDrBrwWTgEqMW1mN6E4Ss UTMFkDsjDXENf9VTOnpRaytgZ0VkaXRubyL28ZMgEJxvrm4zwGWPQw7tRQYA bYkElAFgAV0EVHWPjTfJ4/g4hQJUkAFJEqaEMiITs4RdIhTAhBAzAuMx4zG5 ArYvKCtxooCB6Gwh6vFsJdN1LDj0AgL/YAxNYW51ZmG/Y3R1cmVy7lQA3PZR +DIBYBEcmSBO32FtZQD+uoIEYE0Jb2YA/t2BXSVx6o+4/I8xIzoiAmAO5FBk /3VjdCBOdW1ihCabOiIBn59Pn2GSMwVr8G+fgZ8xI+MzYAtQYQlyp5fjMgEj rCqX059hksIXNYbznwWvMSOFcgJg/ROhlURlc2NyafFwl3Ixo+BQV1dXIJ9D bHVzdCaRhXIBGKGvs6JRr/5njQTQQWuPCOH60EGOjZOfgwgx9AH0AQIyIhdR tF0iYAFgATMC9UEA9UG5AuAR4BFEEvgx+DHtQsBkAWQBCQKFcYVxwUAAAIiA YYBhEVEABEC7sggxAFj/Af8BBCEAtC8oQnGv+AUwqYGOIAhMR0NW30kxNDEA M0RgF/1PsiBpbmFsIET/UlcgZmlsZSABbkqQ7CDWsV0lX2Hmv9BBro4gAzYu EDAjpyMZ/hLGcmVsZWFzZR8gdmVyc9MhKcMzBfjrc+i/+rMxMC8yNd8vOTkA /vVCAWD/D0RhdGUgR2UDbmXQIOoi1rEXNbVCjs78j0GOIAY1MDAwM/yuwDKl VW5pcXVlTyBLZXkoxEQVPNfPfunIU3ltYm9sy6e/U2hhcGUgoZVME2luJ8Xt RTMe3wUwfzKfAGACMTe5l0vTST9jb24gSUQoxHgFuB3fL9/Lp1ZORQmIVAN5 cCfFqrKbh5DMuoNVmv70AgFgDEZNIEMOusBnb3IN1bey6/AOQAwE79pWRk1X mM2lMuBvZx9DbGFzcyjExLLr8LlNTezTcQEAPosjNLrAoCzEpCxJbsigbvdl dCw703MsUmVscGDkICxWdMMsaLMgkCaR1rHRsuvwD5+D/wQkDUCw4gFwkQSI Af8BXQSg+dXm9YgBXSWABarj+FDoqQTUBro0wNMPBwIBtvGzwABVQQIA+gx0 GPsBmAG4UAUDM5QQErHxGXSmQZgB5ASOQucw+wEvYgIABtHzA73z3vH+k/II YBpSZXBvA3Np0iEJjlRD12xBPbVCsQQrcj9A2PYN5N3wBQ7d8ANiBVVmwmK0 8FVCsK0XjxJBPVlxAgVAAwFJwOj2UQf+UfEyJTKYPwSqNHf+CYn3FJKY8wNC 3fEBAynkvfJM4YhI1v/o9/VCBwhgD/r4ZXEUCXoTUT8AOAY7REgPWgIsdGYP 2hNrHziQBkNEoA8D/iyNtD3xAMOBeVDDgRi9XSXzMJWG21MBVIdvfGAoIUBV 9UM6I2V/h2UAaafUHV8vX5OIlHCTiFrCJDt+UCcAMrIyG3G8X2EAAdwx4rFE 5fyPQXlQaknsCWsk/yhOt3N/DifEAyK9fxcBAABAnJn08PclGUDy89kSQEA1 mvTyKQME2SIDBfnwrg4DEkBQ5/SA6/AB98ADFuvwAgR1Au8AAGAg6PPgPwVd /jcCBXUDPwsaOAHrAnQ+AQXr8CD9/5p0Ae9gBQN0VAFuDwCTAgBpB1kMAYMH QwckTuvwAwIGMQExAQTRAWUJMQECiAHq8QkAYwErAwPRBgXbBgNlvOvwQOb1 8veamQYA9yUpQAQD2SJABP8ACABvH5AXAEkB6vEaAhnr8B8AFuvw/wICdQIA AGAg3ujz8D8F/jACA3V5AzgLKQECAHIJKgFbAgA7BAAAWgQBYgN1A2kJEOvw AwIG6/AqYwIAjgIDlAYBf4EBAABAmpn08P8lGUBA5ziO4/84jgPAQOnPlj/j uCsgQED/8P/xdUAEBBAMBfM/UOf06YDr8O/wL+vwAgJ1b0wAAGDu8QAt8PDv YggAcurxAwfA1wWBAuvwCfDwgEHW8PD+GzgBAz0CdR/WPwCALl0GKTgBAHDd Bef0CQDAhwcBwFuAC/Dweg59BgGHD6qZDhY4AQRxAiDo8+CbPwXSAwV13gDd CRpWUAECBuvwDDEBBAUWVQP/AgMFFgW9B0A4AWcGdAc/ALEBdAg7FNuAIvDw ohjr8KAN3uvwYgEAo6AAAJPOWhATkHQFED0SBf69R+vwMhzHcfPwrOE/8v8E BujzFQTgPyC+IQD////vQejzAQEQ3PD9////AQALAAAAAgBgATAA/t1F3P8A ABrr8AMBtQbr8Azr8AECCQYDKgMCAwkGAAkGAfUv6/Ag4/j+////1wAAAerx GOvwAgXzcAXr8AQB8APAgB8fAABA/r1x6/D9////8vH+3Pf04vkBAArr8AIA 32AAAP5JEQEBYP8/Q29weXJpZ/9odCAoYykgMv8wMDEgTWljcv9vc29mdCBD b/9ycG9yYXRpb/9uLiAgQWxsIP4mAnMgcmVzZXI/dmVkLgD+GAAAANAAAAAA AAAAAAAAAAAAAAAIAAAACAAAAAAAAACbAAAAFDcmECYxAACVAAAARwCYAAAA hDYmELsxAABwAAAARwCcAAAAxDgmECsyAADvAAAARwCHAAAAFDkmEBozAAAn AAAARwCnAAAARDQmEEEzAAARAAAARQCEAAAAdDUmEFIzAAAhAAAARwCkAAAA ZDkmEHMzAAAnAAAARwCjAAAAnDsmEJozAABoAAAARwAAAAAAAwAAAAEAAAAC AAAABAAAAAYAAAAFAAAABwAAAP+MIgAAAQAJAP8AA0URAAAGAPXG6fIF6/AM Ao8X624fBAcH6/AQBAF08fAeAQPr8B4ABOvw7wQBDQAYAfwCAeTn9CwBLToA 9PAA+gLi5PciQQTy8k0ABgBB8lMMAkEGNAMAAP//lf8/BgMrAgZhAP7xJP8D YQCtBHMRJf8FqRLABcYTYP8GxRQjB6EV8P8HVRbMCNcWqP8JNheYCm0Xfv8L bRdfDEoXT/8N+hYmDm4W7v8OuhW2D94Uav8QuhVGEW4WE/8S+hb5EkoX7v8T bRfFFG0Xpv8VNheMFtcWbf8XVRY1GKEV8/8YxRSdGcYTLv8aqRK6GnMRVf8b mxEEHIwRqf8cXxFOHfsQ5P8dahBXHrEPxf8e0A4BH9sNOP8f1wxMH8kLOP8f sQoBH6gJxf8euAhXHuYH5P8dKAdOHZwGqf8cKQYEHPcFVf8b7QW6GiQGLv8a 5ASdGb0D8/8YuQI1GOwBbf8XLgGMFrEApv8VTQDFFCAA7v8TIAD5Ej4AE/8S nQBGESQBav8QzgG2D6AC7v8OzgEmDiQBT/8NnQBfDD4Afv8LIACYCiAAqP8J TQDMCLEA8P8HLgEjB+wBYP8GuQLABb0DJf8F5AStBCQGA/8E7QVZA/cFtP8C KQYUApwGg/8BKAcLAeYHne8AuAhNvxAlALH+vRDJCyUA1wxN+wDbsRDQDgsB sfsPg58QFAL7ELT/Al8RWQOMEQOHBJsRrAFCBSwF/vElAKkPuw/ND98P8Q8D HxUfJx8AOR9LH10fbx+BH5MfpR+3HwDJH9sf7R//HxEvIy4wAWQJVJABWgUE KwLwewAMQSL/BABmD+wQNgg/LAg2FywI/jHMPeEebQ/vMuwzggl/f3/oPwbs M6ABDqUCBQBq/wY0CvEGsgnUfwqyCU0KNApmQRAwKTRFMAFeQSVjT3VGMAHO fg2/v79LT11ILwpLUgrmQe11QO5BL3lPcItO5k+uT4gDpaWlyU/+XUZSCvIK 1ApmDmVQsglSc0BiUfhPjEv4Yl90UzBfAADY2Nj8R19dRm8G8QbYBn+NBrsK jQZSnUAA4FF2X4xL4F/yU65fxE9YTXkMG1BgYewGagZoYYEMdV+LTmBvKG8+ X0xvu68KiAZNnUBNl2C7BwqjCdxhcm+MS9xv7mP4qm8+b1hLegXYCxr/BhAL XAoQC7wHCdgLWnHwb4xLWn9scwAofzp/TH+Sf3B/gn/ef6Z//LxvWEvBCQAM Vwr/VgtXChULwQkN02GAAAzsf4xLVo9ogwAkjzaPSI+Oj2yPfo/aj6KPlLh/ WEt/DYDB5XAIkX8BBVyR6n+CbVKfHo8snz6fAISfYp90n9CfmJ+GBd4zuJ9+ YEPzCVAHyQZQoW+oCfMJWKFQBzAt4JJFTq9gozABBAEUApjXChwJBAETm6Dk CcTMP1QLBe02MCmYD2EA/24H9AhzBw0J/4cHJgmWBz8J/6AHTgm0B2IJ/8MH ZwnXB3EJ9esRsP8RsBMIcQn/IghnCTYIYgn/RQhOCVkIPwlVaCmwfCWwiyGw mh2wVbMdsMcdsNEdsOohsP3+JbANCU4JHAl/PwkrCSYJOlOwd0QJ9Cmw9Ahd ZbD1bGWwdkuwhQngCP+UCcwImQm9CH+eCakIowmQh7BVfIewaIOwT3+wO3uw VSx3sBhzsA5vsARrsF0EKbAECESpsDoTsH8rCdIHHAm5+aD/qgf+CJYH6gh/ kQfRCIcHx9GwVbPRsJrRsIvNsHzJsP9oCKoHWQi5B1VF6bA2ybAizbAT0bBf /weHB+sBwNcBwP/DB5EHtAeWB9egB6oRwLkBwNIHfXMDwG4HBAhfJcD1VSXA RiXAPAcOCP8tBxgIHgcsCH8ZBzsIDwdPQ8BVaEPAfEPAkEPAqT/AVb07wMw3 wOAzwOovwDX0K8D0J8D0CPShMC8AQiP0rwa/GL8qvzy/Tr9gvwByv4S/lr+o v7q/zL/ev/C/AALPFM8mzzjPSs9cz27LEk/gJE/KozxPxKlaScoUNP8KURWy CTQZsh8JrRg0Cmzh5JvKo8COSWzvrE+qn1jvnuEvCguyGOzh7Xvg9OH2TY7v 8OrvKl+uj9bvBQCyGN/yCjQZZmvwsgkBsnngaPH+74xLaP+oX7pf/lT/BQDP FPEGOP8VjQYbGY0GsgcY8Qbm8Xz/jEvm/yZvyMTv0v+e4Qwh8GYB7AYDyhRu AXBtCA9kD6RvQv/+UA8FABsZiAatevPwrZ0AGxmjCeIBwHgPjEviDyJ/Pg/O DwUA69oT5nAU4nAYEAsPHBjYC2AR9g+MS2AfABavOh9MH5IfcB+CH94fph/8 vA/KHwUAIRgADP+3GFYLtxgVCwchGNNnIOaNAC9cLxqfADYvSC+OL2wvfi/a L6IvuB8uxi8FAN8TICHrEA4xA98TYjHwH4JtWD8kLzI/AEQ/ij9oP3o/1j+e PzavYEO/UxhQBykVSEGoBwlTGFBBYK9ypUZPjK8bfBegpUQYqq+8r86v/uCv YQDOFfQI0/8VDQnnFSYJ9v8VPwkAFk4JFP8WYgkjFmcJN68WcQlLCVBfCVBz qglQggVQlgFQpf1Aua8WPwnI/UDcAVDr+gVQ+glQExdxCSf6OVAxOVBKF2cJ Xt8XYgltF87AFz//CYsXJgmaFw2/CaQX9AiuXVC9+l1QzF1Q1hfqCOX/F+AI 9BfMCPn/F70I/hepCAOrGJB/UHx/UGh7UE+qd1A7c1Asb1AYa1AO6mdQBGNQ BF9QBAik/qFQmhfrB4sX0v8HfBe5B20Xqv8HXheWB0oXkb8HMReHByfJUBP2 yVD6Fp7QFpEH3P8WlgfIFqoHua8WuQel4VCW3VCCqtlQc9VQX9VQS9VQN6rV UCPZUBTdUADhUPb/FbkH5xXSB9P/FesHzhUECL/qHWC1HWCmHWCcFQ7/CI0V GAh+FSz/CHkVOwhvFU+qO2BoO2B8O2CQO2CpqjdgvTNgzC9g4Ctg6monYPQj YPQfYPQI7EEAeM+Kw+xP/k8QXyJfNF9GXwBYX2pffF+OX6Bfsl/EX9ZfAOhf +l8Mbx5vMG9Cb1RvZmvgGO8q7zzvTu86SUoNtP8R0Q0yEbQRMgMRLWuAZIHq P2RLZI92g8ikP7Y/Vo2vbYDkgW0SC0oN7IGvd4+JjuSPrI/8ti/QjzIRchK0 ERnma4Keg3IS9o9kS2Cf8HKTLp/G/zhLTw1xDv+4DQ0OmxENDg8yEXEO3pF0 n2RL3p+Q8JOsn8KPVo2MbYBeoWwXDkoNZqGM9Y+Jjl6v+CavPJ9Kr5sRCA4t uuuQLZWgmxEj5aAIgPGfY0zar+yjqK88r8g7Wv8MWBP6DJASPH8RkBKcEFgT WLEA7q9kS1i/arMmvzi/Sr+Qv8Buv4C/3L+kv7qvyDuhEP+AEzcR1hI3ET+V EqEQUxNUweq/APw7VM9mwyLPNM9Gz4zPas+gfM/Yz6DPtr/IO18LwKEM47AG 0V8MWtHov/o9UN8AHM8q3zzfgt9g33Lfzt+W3/wuT9Az0xDQDqkNXkDhKBHT EEjh0O2q8P45Pu9Q44pHGBL8Dx6YQxgSxBCiT7RPxk/+2E9hAE4OdBBT/w6N EGcOphB2/w6/EIAOzhCU/w7iEKMO5xC3rw7xEMsB8N8B8PP+AfACD+cQFg/i /xAlD84QOQ+/qxBIGfBcFfBrEfB6rw/xEJMt8Kct8LGqLfDKEfDeFfDtGfD8 /h3wCxCmEBoQjb8QJBB0EC5V8D36VfBMVfBWEGoQZesQYF3yeVvwfhAp7xCD EBB38PwPg/8Q6A9+EM8Pef8Quw90EKwPZf8QmA9WEI4PTK8QhA89mfAumfAk +pnwGifwCxBSD/z/DzkP7Q8qD97/DxYPyg8RD7GvDwcPp8Hwk8HweqrB8Gu9 8Fy58Ei18DmqsfAltfAWufACvfDzrw4HD9/t8Mvt8Lf+7fCjDhEPlA4W/w+A DioPdg45/w9nDlIPUw5rvw9ODoQPPxUANfoVACYVABwOjg8N/w6YD/4NrA/5 vw27D+8NzzMA6PozAPwzABAQ7w0p/xD5DT0Q/g1M/xANDmAQHA5qKxAm5eA1 5eA/5eDk4QBwb4Jj5O/27wj/Gv8s/z7/AFD/Yv90/4b/mP+q/7z/zv8A4P/y /wQPFg8oDzoPTA9eC96c1ScB//8cgSYG/w8AMQBDb3B5/3JpZ2h0IKkg/zE5 OTYgTWlj/3Jvc3lzdGVt/3MgRW5naW5l/2VyaW5nIENvf21wYW55AA8VJH8T AEFsbCBSIyH/cyBSZXNlcnY7ZWS75AEAA6vkAAEPcUDr9On26fIBABbr8P8C AHUCAABgIHbm9QX+BgIBdQMOC+1A6/GamfTwJSlA3O7x6fIBABbr8AIAv3UC AABgIOjz8Hc/Bf4GAgF1Aw4HDwAABf7tQOvxmpn08CUpQH7y89kiQAEAFuvw /wIAdQIAAGAg3ujz8D8F/gYCAXUBAw4L2UDr9Onzmpn98Nki70ABABbr8AIA dd8CAABgIOb1Bf7uBgIBdQMOB/A/BQH+cUDr9On26fIBABbr8P8CAHUCAABg IHbm9QX+BgIBdQMOCxgAAACcAAAAAAAAAAAAAAAAAAAABgAAAAYAAAAAAAAA iQAAAKwfBRBVQQAAAwAAAEUAigAAAORfBRBYQQAAIgAAAEcAiwAAAGQyDxB6 QQAAMQAAAEcAiwAAAIR2DxCrQQAALgAAAEcAiwAAAKx8DxDZQQAALQAAAEcA iwAAAOSPDxAGQgAAIgAAAEcAAAAAAAEAAAACAAAAAwAAAAQAAAAFAAAA/RTj +FdXVyBDbD91c3RlcgD9M93+ECcAMhzHbXEEALw/6vEBgP3xruvwAwcG6/AM DQECGAAAABoAAAAAAAAAAAAAAAAAAAABAAAAAQAAAAAAAACUAAAAXGIREN1C AAAiAAAARwAAAAAAJWbr8EDm9fL2IPz/Dg73AQAu6/ACAHAF/isCAADwA8Ci G7br8KAP6/CATPDwozUM6/CTRgKQ/hgAAAAaAAAAAAAAAAAAAAAAAAAAAQAA AAEAAAAAAAAAngAAAMxAJhA1QwAANwAAAEcAAAAAANXY6/BE6/CI6/AEADUF 6/D/AA//Av30FAmq5PcB6/AG5/Ro6/Ak/zMmEAI0AADQ7uvwVAAM6/DAlFLr A9JHAINQAEYAbP7r8Nw8JhAoQgC7AKTr8FUADuvwlK81JhDMawAR6/BG+wBp 6/BkYBEQ/+prADZLAmT38UAmEANsQ5AFJAHV7uvwSOvwmuvwQAD3FQAC6fKf DlYA9Qjr8P8MAAAAlbn/WolavXadAAD9Af30AABjadtCrxZa8cPi+Qfn9Eb+ 6/BcXSUQeiYAuwDK6/DWACPr8Nz/miUQRCcAABL1AUABCuvwRG4lEJ9WKAAA Xz8CZgGkz1MlELVvAGEAAET7ACvr8HR/JRC26m8AQj8CR+vwDG8l6xD4bwAu CADWAE7+VQEyJhCiQwAADYLr8FYAVULp8hHr8BLr8BPr8FUU6/AV6/AW6/AX 6/BVGOvwGevwGuvwG+vwVRzr8B3r8B7r8B/r8AEg3PAAAAAAAACldOvwNO/0 6vEQ6/ABtuT3IUACAyZA4/gC9B8E6vEL6/ActigQX9NEAAA46/BDLQL/lLco EAtFAAAFBuvwQenyAgFVQunyEuvwE+vwFOvwVRXr8Bbr8Bfr8Bjr8FUZ6/Aa 6/Ab6/Ac6/BVHevwHuvwH+vwIOvwASXc8AAAAAAAAKV06/A07/Tq8RDr8AG2 5PchQAIDJkDj+AL0HwTq8Qvr8Ey3KBBfU0UAADjr8EMtAl/8tCgQizcABuvw AUHp8gIBFAAAAEIAAAAAAAAAAAAAAAAAAAADAAAAAAAAAEAAAABssigQEUUA AEIAAABSAAAAAAAAAAAAAAAAAAAAAAAAAEAAAABktigQkUUAAEEAAABSAAVI 6/A83P8EDxYPKAoFSOvwPNz/BA8WDygKFAAAAEIAAAAAAAAAAAAAAAAAAAAD AAAAAAAAAEUAAACktSgQJEYAAA0AAABSAAAAAAAAAAAAAAAAAAAAAAAAAEUA AADUtigQMUYAAA0AAABSAJWq6/Ao6/B23P/h+gLq6/AF5/Qe6/C8yCG/ELsl AACs6/BT+wDJ6/DEryUQZ68mAAAT6/BCIQLc/1wlECREAACv+i8CP+vwFLUo ENJvRQAAUuvwUABPABsAXF0APkZiBRIFFAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAABQAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAYAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAN3j6/ABABXj+Pz//9z/BAXqAgQAuf8K AOb18D9VGAJJe0CqIAH6SkAB3P9c8fDw8QEAZ+vwCzkEYE8PYQ9zD4UP5fY9 Ko8PDDQJwA8AAH9SAQAAAgAd6/D9COf0IP////j/D+MCAgDv8AkCEQ8jDwA1 D0cPWQ9rD30Pjw+hD+X23QLd/kQBCuvwQzr/XFByb2dyYW3/IEZpbGVzXE3/ aWNyb3NvZnT/IE9mZmljZVz/VmlzaW8xMFz/MTAzM1xTb2z/dXRpb25zXE7/ ZXR3b3JrXEf/ZW5lcmljIE3/YW51ZmFjdHX/cmVyIEVxdWn/cG1lbnQgKFX/ UyB1bml0cykfLlZTUwB/RAEAAAIAHevw/Qjn9CD////4/zfjAgLx8AAA7/AQ DwAiDzQPRg9YD2oPfA+OD6APcOT3CgHo8w4FxAEK7/D/QzpcUHJvZ3L/YW0g RmlsZXP/XE1pY3Jvc2//ZnQgT2ZmaWP/ZVxWaXNpbzH/MFwxMDMzXFP/b2x1 dGlvbnP/XE5ldHdvcmv/XEludGVybmX/dCBTeW1ib2z/cyAoVVMgdW7/aXRz KS5WU1MBAH9DAQAAAgAd6/D9COf0IP////j/H+MCAgAD6/Dv8BAPACIPNA9G D1gPag98D44PoA+65PcC3f5EAQq9AEP/OlxQcm9ncmH/bSBGaWxlc1z/TWlj cm9zb2b/dCBPZmZpY2X/XFZpc2lvMTD/XDEwMzNcU2//bHV0aW9uc1z/TmV0 d29ya1z/TG9naWNhbCD/U3ltYm9scyD/KFVTIHVuaXR/cykuVlNTAH9DAQAA AgAd6/D9COf0IP////j/H+MCAgAE6/Dv8BAPACIPNA9GD1gPag98D44PoA96 5PcC3f5EAQoD7PD/OlxQcm9ncmH/bSBGaWxlc1z/TWljcm9zb2b/dCBPZmZp Y2X/XFZpc2lvMTD/XDEwMzNcU2//bHV0aW9uc1x/TmV0d29yaw0V7yBEZXbz AHMgKP9VUyB1bml0cz8pLlZTUwB/OwEAAAIAHevw/Qjn9CD////4/x/jAgIA Bevw7/AQDwAiDzQPRg9YD2oPfA+OD6AP+uT3At3+RAEKBAD/AEM6XFByb2f/ cmFtIEZpbGX/c1xNaWNyb3P/b2Z0IE9mZmn/Y2VcVmlzaW//MTBcMTAzM1z/ U29sdXRpb27/c1xOZXR3b3L/a1xUZWxlY2//bSAoVVMgdW7/aXRzKS5WU1MB AH9GAQAAAgAd6/D9COf0IP////j/H+MCAgAG6/Dv8BAPACIPNA9GD1gPag98 D44PoA/65PcC3f5EAQoFAP8AQzpcUHJvZ/9yYW0gRmlsZf9zXE1pY3Jvc/9v ZnQgT2Zmaf9jZVxWaXNpb/8xMFwxMDMzXP9Tb2x1dGlvbv9zXE5ldHdvcv9r XFBDIGFuZP8gUGVyaXBoZf9yYWxzIChVU/8gdW5pdHMpLg92c3MAf0kBAAAC AB3r8P0I5/Qg////+P8f4wICAAfr8O/wEA8AIg80D0YPWA9qD3wPjg+gD/bk 9wKA3v1EAQoG/wAAQzpcUHJv/2dyYW0gRmls/2VzXE1pY3Jv/3NvZnQgT2Zm /2ljZVxWaXNp/28xMFwxMDMz/1xTb2x1dGlv/25zXE5ldHdv+3Jr1wBpbnRl cv9zIGFuZCBTY/dhbm4bEShVUyD/dW5pdHMpLnYHc3MAXwQBAAAo6/DQ6vEn BeYC3P/o8wgTBOrx/Srr8HS2DxBKR9cAAE7r8EMhArzc1yYQmCsAuC8GHKp/ JhBQSAAArS8GXzyHJBD9TwCoLwb/hJUkEKVJAAD9oC8GDJgkEEVK9wAAoS8G TJkkEPnmhQBSB5yaJBCTj0sAAK4vAurx7/AAov3xA+vx+PHr8Abr8AcABABV Fevw//LwI+vwUOjz/Pr08fIAAA0CBCju4PvwP1UjAklAqg4rAfpKQNz/4vn2 8erx1Ubn9ATr8ETr8AEACVRlAerxA/HyeQ8LAQcF0JAH4PthAerxaG0GAAC9 JOvwAgBUGOvwDNDf/MABwAHm9QHFAgAA9ZLj+DDvAAMAVUHe6fJAVUBB7wEA gLdbQD/o88A/HBW/GAoTIAAuFCRA4gpMH14frHAfZgH3A2ABEcUBA24cFhP+ nHEBDRC9BCkBjBKoExalFhelFssBqqgTGaUWG6UWHKUWHaqlFh6lFhqlFh+l FiAqpRYhpRYipRab7ATq8WWRcQJVMgZNLfA/Xy0YVCgBBesCzwOkEvMB7gK7 AQS4EQIEdfMAYIwcIenyBf6kEuMCnCIFJqIiBXXAAKwpt/AE6vGNT0cjEBBx Ly4UAzgIeQQFPyI/AfMPb+wIhBwhyAQI3/zzAeIl6wWoiuP4TUcj/fLwjjGd AwFm7wABAY4x6wL8C4sU/wBgB05ldHdv33JrAP4KxQEBYBcAAP6NEgq1OHsx PiXRUYg/mj+sMBOwMwlD/29ubmVjdG9ylL45D0IKFUpl8AT0EYwCxQd03/z0 EfQR5w3AAWEBVQXr8Abr8AdUNAnr8KjAMawwzgIN6/AO6/APCOvwpBGNERLr 8A9B7vG4EQTCEcsBFOvw1hEIIeAR6hEhg+P4wDEHEO8BR+ny5kEAPiXyRMAB /UfiJfJEhkH9RwjjBfJEjkFI/kawBfJEVzEAVFeGQe5IlkFUV4pB7kjAMQBU V45B7kieQVRXVzXyRM8BAFRXlkHuSKZBVFfAMe5IqkEAVFeeQe5IrkFUV88F 8kSkEQBUV6ZB7kiNEVRXqkHuSLpBAP1HrkHuSNJB/UekEe5IuBEAVFeNEe5I whFUV7pB7kjLAQBUVw9B7kjWEVRX0kHuSAghAFRX7vHuSOARVFe4Ee5I6hEA VFfCEe5I9BFUV8sB7kj+EQBUV9YR7kgSIVRXCCHuSBwhAFRX4BHuSCYhVFfq Ee5IMCEqVFPJ4AigxQJQYQGNESSNEe8BESYhJiURMCEwIQaGQREl6/BtgbpB 12WAjwCSj6SPto/Ij+f0/UHvAVcxAGUM71M+JfLxhkF8D4gH5PcAVzW4BYpB EgHIBLplqAmYVdC8BYJFOiHm9XlHJD9+/+sa/2tDQEBm/qGQpgtRQEBfWP/C PV96JUBAyK3MtJB0GaiVFbGVCVVAgyoPByMCPiUHmyPEi0IIIgPilvwx75MB p8jgCK4T7TIBnUKzMAEwBwD+hK0IHBEHEHoa6vGxLrkizwFhAQECXqYDkKZB YqKlQmKiAZ4Re1VF/kckJMreBNGWJvdAQcDGlEBxHMd+rKDxP0DjOI61oLUD qqXhs6XzPwEFgADo8z9B7JXqEcgEsAXnCT4lKP9F6kzzAU7p8m2lFsxDoMgE SL88Jy+B5vU4cQJUAGEBukG6QRQSD0EPQ89E0kEQ4wLu8e7xsAJnc6Y8v06/ AmC+k3KznkG4EbgRFBLCEQDCEZciywHMFdUS1hGwAgghAAghe1LgEeARmFLq EeoRtVIAZkVXMv4R/hHvUhIhEiEpYgAcIRwhzwISAXtVMgHIAAm52ub1ZKGR GiKylBVACBmfJaHxA3CHREg/Wj9rTEWq4/gvRyS8lrWkwLufhQfhkP6LAxfR PiUj1SYom6XFlTzfCU/XcYtE5M8U9s9rTJnj+JBHLXbWTi5LAAFS0wEcIKIi AKgteqIiAc4p4D8F/ggj1l6vAQMIIgFL7wEDgNbRPiXi1aqc+d8L7x3r8DBC 4C/vQedyrwEBX+Pu7wMBASyxwtVUUWkEgkMFuYDy8YpBws8lne8F6wVBmSwA xQcuda4maZzvAYJF8wGEmVqSo1d+30Biq5XNtJVA7PPCkPXzy5fXoe8AmDso 0T7T9E4v7P4EsdChFgNpDT0jWQLWEe8AnhHiJeqSpWC9leA11Sf1gq8cx8gR MtZAlAQBADLVza4QoTilHK8up3tVrhMlV8q+bEE0ikEwqNICclDDvYpBqv16 TYltnAPmMumKOyias1ExDwAAiwDfKHwV7PaNG0FVox719ZUTADilfB7bG3tV mrNML/pEW4EA6o/uVEFVC59X/y+fQZ9Tn/hln3efiZzoaY8waIOhMp+fsZ/D n9WaVCUHCDW29JUzNgE9Ng+v2A8zrwhFr1etSucBSoF2o8M2ha8Aatmjr7Wv sA/Zr+uv/a8PvwDyRHixLL+8v86/Yr90v4a/AJi/qr+PT6FP4L/yvwTPFs8A KM86z0zPXs9wz4LPlMVUJQCkz7bPE2nJJdzPrd+/3xLfACTfNt9I31rfbN+N b5DfNG8ARm/G39jfje8gL7Hvw+/V7wBE7bg4X+y4OHnv6N9Lf11/AB7vgX/l 5sM/APU/jxrzyUEAJP828GYTP/+hj2P/df+H/wCZ/6v/vf/P/+H/8/8FDxcP ICkPOw9ND18PcQ9khA+WDwCoD7oPZj/eD/APAh8UHyYfAPmPSh9cHZcxcB8e L5QffgIAqx+9H/mp2x/Jr/8fO78jLwD1gzsvTSTeVVwvqI+AL5Iv8KQvti/I L9osDEu45wdLr07wLwI/FD8mOt5VEQf2gkE5hMYBjsZgPymvUIQ/lj+oNBL2 AhL2A/qxEAzzFNbWP4AAKOw//j8QTwAiTzRPRk9YT2pMh6F9Tw1fAB9fs0/F T9dP6U/7T+Df8t8AMV9DX1VfZ195X4tfnV+vXwDBX9Nf5VXeVfVfB29k+feF AC1v/m8Qf2NvdW+Hb5lvq28AvW/e/+Fvhf+X/xd/KX/ef8BxvwKPFI8mj5V0 AN8BAwxTg+0PAQPKfzl/nA+uDwBvf9IPk3YU3651kB9rg4L1AHmOsbGluZqP w7++j9CP4o8A9I8GnxifIA8iv06fYJ9ynyCEn5afqJ+6nzTVYNWf558A+Z8L r7fPL69Br1OvZa93rwBKL5uvra+/r22/46/1rwe/AENPK78ZT0+/Yb8oTnmB QQUAeYw9Q/UfI2MGXd4/GS1FD6hXDx1cJsm5u0o5PxRAVEnAbzM6PxUyPxUq PxUdIm7LzwMkf8JBAYTStQTRVQrQVgoEdBIEKGQZknHLUgPRVgXbVgpmKwoF dBIFeh2GHBgKMTIrEAHKvFJjAQSwNkBXUMMyT19hX/w/VXrAjxwYwbC7SrFY mF3fZrFWMkDBAV0fkXJ0H5Jx+rECAhZkEmIBzgUDeh3vXy92AC5jPnJ5gYJh 3U/vTwFfE19AJV83X0lfjm9tX39bTI5eBTygXiyyX8Ra+XHVU3uFIQp6hu1f /1l7hgWFhrSGACdvOW9Lb11vb2/4fwqPpW8At2+7SluIQo3fZluGA38VfwAn fzl/uZ+dj29/CvMawYt/AJ1/TKLoMVqpxH9kRd5/8H9oNZ8Uj0LgkbtJgEQ/ FKnAN4VahxQ/FRw/FQQasCvPE9IDAgzx1VMDrwbZlQMDJbbSj/6S1sHEJXj7 4ebPJPMCAoBMoUAB/rE/wz/VP+c/TZ9flBJJAJpBvk8LMrGfzQ8xxTlJCMsA W4Yfz2wfQ81kRVvP96R0zwCGz9XHh0kIz8DPLM/kzyz8kAXfcc8p33jaDhI4 gjNUFrq1UjHJEgour0CvUq8AZK92r4ivmq+sr76v0K1RhqWg6qYYBreaRAg/ FfQAWToctr3RKb87vbXmVL9mvwB4v4q/nL+uv8C/0r/kv/a/AFTfZt9435HP UM987a//188Als/l/4HqBA8WD+vPOg/aJgBVD8H/PN+LD6T/rw/BD5bfAqjc MskSgjHB39Pf5d/338AJ7xvvLe8/71Hv06qwUKT69eqmHvr2+vUO+vXwAJ/v IbGCMbnvO71FJuTv9u8ACP8a/yz/Pv9Q/2L/dP+G/wDkD/YPCB8hD+D/DC5A P2gPACcPdj8SKZQ/pj97D8o/NjcA5j9SP80PHE81P0BPUk8nHwKp2zNUEq8i Uh9kH6x/iB8A0H+sFfohy1G8FK8h1ilck6QiXx2JDvIg9xEzj1CztbxJgGaY UKYLQIDMIqFR8t9mtyChUxKKN2aKZ4gDLd6TryBgdb3Q/2AE5GICAAd1cQ3b UTcQ4lQDgBG3IWUiuREG6VjZWIAStyFRF+oz31HZUgPkUQYhYwEB7lLpUjJk H7OvIUkvO70oVmZ0L/6SnR04WohUHyMpOJZYj1NAllcBR1JUaAJVZwZWZtRj riLfb/Fqe2cuCH8AAQF1ggEjfzV5tSm5EnDGNQcAangBOwCANxBAejO3IWVy EQFrf31/3pJwjyWdcb52Qqp/A71/z36LIb9FP5gvEiAiLndm/iEBriI3gkZy CDeCVWI3ggO7JURXzSJTX1hlX+FBBj9VARo7KnVPAuwkAo9Paz28haFe1Y/5 MwDlQepQ4E/yTwRfFl8oXzpf8ExffY9wX4JapuRRgfelDUmWWNTacPUn05In qFjykxe6X8xfAN5f8F8CbxRvJm84b0ppLpHAWm9sbaemhW+Xb9mUS8kfowJL G0aWWAQmy20Ep6ampwanpvhkLL8+vsynAFm/PnZ0v4a5ZXTzY+13wbIC038B vL/Ov7hzbEWdcQ/GAvu/Aw7P+H8KjxyPyKY3gwCExpeyS4OExlqPbI+wn5CP AKKPtI/Gj9iP6o/2EvKeJt8AJJe7QVEfRJ91H2ifmR+rH7C9H88f4R9j6oA5 cuQwFVSKNTGE7iGW76jqu0EAq69sbXfm1q+GL2fPqi+8LwDOL+Av8i/pzxY/ dE+GT5hPALE/cD8+7XH/9z+2P6f/Q+oAxv/Y/wtP/P+fD3r/Mg9lT6BWD8L/ eg+tT79FGTgbT/9TUEYgUHJvY39lc3MgMTAwhtIBD4Dfkt+k37bfyN/a3+zf mP7fEO/wHUBLNO+89QwCT+78Ye+joR0Re+9sbRUmAKbvuO/K79zv7u8A/xL/ JP8ANv9I/6YPuA/KD+P/ov+0/wARPysP6v9HP+MYZD92Pz0PAJo/Bja1PyE/ jg/rPwQ/D08QIU/oD/oPDBYyFhRx4TKfQDMfVp9XH3qfjJbOISiJF8Bx4aYp ziHpNb4tzJ9VQD+GJdzzpbeeUJhWf6CnPcKghSCoWAKDUxALrx2vL69Br1Ov Za8Ad6+Jr0ppceEZL2xtOGZEL/rorgjZlAxLuOdLW69QCL3AUhq4DNVjBDhm N2cGOGZJtL1vz25dZwDqbz52BX8XeWV00iHqtU52EiTPAU1/X36GQDgtgpwh TFXAdScAAehAsRIViRLfYgAAYjPKcXqDJFbFljUraC/TkhpZZjeDAgSGAazP vs/Qz+LP9M8G3wAY32FPPN8p4YNeko8kl91QAL5P0E/iT/RPBl8YXypfPF/8 Tl/PntDTHmHQQv0keFhAT3uEQQvpGYpYs5MJnFgYLUQPVPshCVoBsl/EX9Zf AOhf+l8Mbx5vn6XrgTxvva2gaKZnb3lvalSDUyoIvhwIq29opmenBmimSbTt r/+uAI2nGr+PtjW/R7m2tMoh6rUEfrZjfwF9v4h/mn+sf75/ANB/4n/0feuB CIM0xhePKY8AO49Nj1+PcY+Dj5WPp48o4giznsLPyTNHZlAvoLEi+oEA94Rl Ua+wGNTvgRiR0iEKnwBO06QvZVF8FbIhVp9omrYptMoptxlFJuQ4M87QS20K eFNmEoBQnJng0BslKd3U2SLd1bmQ6NM7EkCmWgMAp5M464HixxD9UxIBoIsQ YAEwIQBaIuk1HeRuLpw5SMXV/2MLtmALQipAe0Gg+9RAcRzHf+C/8T9A4ziO iOADSn3l4Ybl8/8aZlBl0mYA2KCJF3WZviVfJWVRBcCpJlQ7oMYRE2ZQTmRS bcckoMIhjxGMFBv/7ifJpyg4AsQSVM4hG9Eb0Xoi++H74YI+khRm9JYy9bH1 sekyZxAFhg//If8z/pNF8yrhdUEAdUF6IjShNKOfJI8RljKXQSCXQeky/nH+ cTxCGwsEtiJAveG94TLSZyFnIZMSHjIEBQllUR8/BCrhhxKIEV/CAILRPEmM EBvRHc3f1ejVTt+D//8b0fvl++HCISPcBIBO9jzVsiGVD1rf7id22SwAiRdp 8aYp0iG2Lc4l74G53w1X8N9AmuDVexLt0XsTcLmQhBMA5wgiAACY7SjRPmIU sc56HwQvAG8fA5AXXiPoEpdBalBZ4T6V9MXVjRbgiePAQOnPD5bjuCuJUIjm JCObUICa72RSEeE57CTvOUhiI1cAHc4X0WnxMtEj32b0PN/ULwBg2zLTdN+G 35jfqt+83npSA+1T1N/m3/jfCu9hLy7vCEDvUu/E1mBu74Dvku+k7wC278jv 2u/s7PWx/++P//kPADX/R/9Z/2v/ff+YT6pPs/8Axf/X/+n/+/8NDx8PMQ9D DwBVD2cFJR1/D5EPGJHwQfWxAQ42GMRRSEBl1oZB9bG3DwBIEPBF1A+Ibydf Ch8cHy4fAEAfPT9kH3YfiB+aH6wfvh8A0B/iH/QfYen9MhcvKS87LwBNL6E/ cS+DL2XW84EB0ZPBAPOPgG9kgkaRcoksnw7FRpmgUMdKQl/Pcc/0bZGqyhgS 0qTwWjB1QyzSpQNA1KYadUMCQHvPA2KjWFVwrwCCqUGWl6+posmxsqXVvllE wO2//7zNta9RzD/cUwICI4BMnLFIxZ1RMFXHHV1gnVHcbWlRNDmHz1UAm8tA aAGtz52CAd/WwYCiAXeZANbNZasPlnyvD9WWr9bFDsUAuK8TlNGv468N19rB /XVlrwAdv4mvQb/kXGK/zq+Gv5+q8DZh3sUM0UZoU3RhdP9pYyBOZXR3b39y ayAxMC4xJMABMEyCZVHsYU2MnDNkj3SPgIaP/XWgj7KPxI/Wj2fUpUqyuUAG lWAGmf0yGd9wBBuW/MMAK51Ln12RcZ9dkQaIUQMCZVFFk2VRtSNR0hUDV9YF YdYBh5+Zn6ufBL2fdkI31Z/nnmvTr1EOUwL81gH4nwqvHK8ur0CvUq8AsL/C v9S/5r+sr76vle81vwD0r8vv+Mnp7/vvSL8f/3NwADT/oO9Y/6W/g+/o76D/ 7b8I/78RzyPFcY1IRE9WTw70hZnNaJDK+iGWanlhB/4AqdNJ8OT/N9YG/lHS BecF6BpR0gHvDwEBTYFyVU2PAEnPW89tz5WPkc+jz7XPx89a74qgBpXQVRmV KhCeARoinzSaVgFFn1edHSZwnwCCn5LfpN+238qf2t/unxjvACrvrB9O72Dv cu+fD+L/9P8Auu89P+sWWD9M/wv/jj/pGgCtP78/YP/jP2y/YT8ZT7r/gD1P qT9hTwIPFAW9ZNE1Uv9lZGlzdHJpYt91dGUgcywDaW5/dG8gb3NwZgJQByBS MygSmnElHzcfSR+AWx9tH38fkR+jH7Uf648WAv6FVuMf9R8V3ykvOd9bmCEC mnQkIGGTP2YFYy91LwCHL5kvqy+9L88v4S/zLwU/ABc/KT+HT5lPq0+9T4M/ 6D8ASn8MT25/3T/vP6R/tn8lTwA0f0lPWH8Qj39PNI+gf1iPcMdP2U/rT/1P IFIyHlIA6QMdXy9fQV9TX2VfyHRCwQEoUAc5YZFZQsFaCQF5tV9+iYSAREBA zczNkLWs2FYy+sSZFetVIgLclQUHLzth9tEhLzMtDabUTC80YoVaaCOEgzB6 fxSuR+F6hD+NjBcCAA1SAAHbEVlhFAUAZW93b4lvXZimb7hvym/cbwDubwB/ En8kf4KPlI+mj7iPAH5/bbXbkHi/AI+cv9F/wLgA3pXTv+W/Go8Jz4T1I8+P vwBqj1nPcr99z4/PxI8ckSAI/09ubHkgb25ldgdTcm/ugWZvcjkG/y8yNCBz aG9392luZwxSMSAobn9vIEVDTVApFJIBGyGUFZ8nn3TSUbGC3Se/IGSfdp+I nyi/rJwOclDhf//AVEBA7w1J8afzMyuVCB3jG8m3UP/RVchE3STuv65vUYgD Le+idY1QYPt1CjDABORiAgDXB3UNV+ELXOYDgJURlmEiUAEGZehV6IBFEpZh FynDW+FV4gNg4QUGneMBauJl4q7kOWNV0aARrzMt0uY8rzRinVpottC0yu2Q FwPYVjHQVGBTR0ABD4JR0OjR5wbS5gKBUgNU0lv/bfr354T/GhACUfIBn/+x +VOhFAVfr29RAHWvh6NglZWuX7Osr45p+qXAflcMso1fn19s0UW/VQESWbsq tM/jpALOz6q9uQUQHe7SDzjDXdEcVt9o33rfAIzfnt+w38Lf1N/m3/jfRvTv UUBAmuCRuVNA/0A/Kpgi42ws9CTo8xMcNuhwCxxrD10NAsBJ71vvbe9/7wCR 76Pvte/H6CsR1u8jragmaAH/E/+8mlLoFBlV0FQbQFDoFFlSTvaoJqcnAQao JnT0LT8/Ps0nWj+69gB1P4c54f/z/wUPFw8pDzsPAE0PXw9xD4MPlQ+nD7kP CB8A3Q+ITPMeqE8hF9ahLx9BH4BTH2Ufdx91VqvB1oQRsRRARkyrwavFLb/L H5NGQPAQ2uoSjwgwYGbOUOYD7CTozVPzPzfnGC1ED1T7IflI7zAvQi9ULwBm L3gvii/G6dahrC8jrYJmVNcv6S4ItF0z5WHfCDCIu1Y7gyMwDFHzgmaBZwYA gmZ09Ad/GX6nZzR/uvZPf8pheSnXgnBOYR2xBwBulngBAIAbYEB6nqALQP6R cgGXf6l+tz/JP0DbP+0//z8RT7+stRUr16+QHLINkqdiFFEB1aKIggEALk9A T1JPZE92T4hPmk+sTyC+T/2BzV4SnyEXHgJfFF8AJl84X0pfhh+YH6ofvB+o X/oAlBjsaFNb2jpY00wZJOgzowk26dQNDz3fR/74XwpvHG8ubwBAb1JvZG/G 6WuRhm+YbeimQLFv6S/6igsmGuD5sxQS5BAHME/16KbnpwbopnT0bb8Af74N t5q/uva1v8e543/1fwAHjxmPK489jyNPn4+xj8OPANWP54/5j0ifHZ/IzDOu 6M8CIRcfbJ9+n5Cfop+0n8afoNif6p/8nw6vATRDEuQQElG2JyTo+MMXNujx X2mvQHuvja+fr7Gvw6/VrwJB0YDsr5htvuYXvym/+yr91YBEETX81qAjN77m vecGvuYAdPRD/1X+4+dw/7r2i/+d+QD3vwnPG88tzz/PUc9jz3XPAIfPmc+r z73Pz88e3/PPng8ADOu+DyEX+eFY0VDRTdQcBwZg3///SAFEGYLb9PPOA4Cg 1UwHstJbD20PWNHj2ZH+pgRYe+I4rttE/0BAchzHcRxH/ssQnRJ2HEeRPv9A QI3jOI7jOHUq1RUu3hUaQFCeGQPPA6Ij+eHC75htDybt7wD/4s3xnQXZ/7EK +PyU1Q8OKBoEJw85CA6aCIpIALDQ/swFCkNvbmZpZ/91cmF0aW9uIL9hdCBS MSD7JQr/cm91dGVyIG//c3BmIDEwMAr9IAUzLWlkIDIw+zAuIDgKIGxvZ/8t YWRqYWNlbv9jeS1jaGFuZ/tlcxQwZWRpc3R3cmliBzAgc3TwIO9jIG1lTTBj LXT/eXBlIDEgc3XPYm5ldEU9DDIzMPkwaTdtMHdvcmsg7zE1LjGcMDAgMP0u ozEyNTUgYXIfZWEgMAoEOoUwFDmPMTUwLtA4Lz9BP2X/IGNvbm5lY3TDZWRb P20/CzaIPzM3CJw/rjRI0iGWFEnfMR9DHwB/35oYnN+u38Df0t+yH1yz/2BR QEC3bdu2921bRyLlMUBASauSJD5QLSLlITxVHQL5FoADKW1BEy+YbW1WPi8A UC9iL3Qvhi+YL6ovSo+bF/B9cWqMnQKJEQICgEwKQAkGhhAwUQeFEY0ZYmEI YmH7T35DBYAPaYKXC1yxIKkPRQL9D9IB4mIB9OnSDQDHaztQ2G8FG/hv0gWd FRp+AD5VM39Ff12X1gH3Rcdufn8A6m+if9YBoWG8f9Vv4H9MfwjTAMwhlNV6 2UMEdQU+DDIAFjfRMJWFRj9YPpA/VU50QgBo4XFPg0+VT6dPuU/LT91PUO9P AV8TXwhxMlu1RSLlHyxAQGhmvZEodkiwAryTIVdfuuFo4XFf/q3rlgCcX65f wF/SX+Rf9l8IbxpvACxvPm9Qb2JvdG92n5hvqm8AvG8ajyyPPo8Ef0G/uZBW vwCQf3q/YX+euLyVsb/Dv6p/AOe/swD8v2i/IM8Hj0u/sL+IaM9Pj4WyvWeP CT/LMzEA0jAP198/8T+tjxhLxYYiMP8uMTI5LjEzMgJVRDOsNURP1o4BAAAA AgAAAAIAAAAAAAAAEVVS6fIC6/AR6/AS6/BVE+vwFOvwFevwFuvwVRfr8Bjr 8Bnr8Brr8FUb6/Ac6/Ad6/Ae6/BVH+vwIOvwIevwIuvwASXc8AAApXTr8DTv 9OrxFOD7IVtAAenyJkDj+AIfBPrq8Qvr8EzSDxDDr2oAAEXr8EMtAkQWMwAI ayADQenyDwEUAAAAFgAAAAAAAAAAAAAAAAAAAAEAAAAAAAAAQAAAAFTRDxAK awAAOwAAAFIABUjr8Dzc/wQPFg8oChQAAAAWAAAAAAAAAAAAAAAAAAAAAQAA AAAAAABFAAAAxNEPEHNrAAANAAAAUgAUAAAAXAAAAAAAAAAAAAAAAQAAAAQA AAAAAAAAFQAAAFTUDxDTTAAA3x0AANcAyQAAANy/DxCyagAAEQAAAEQAPwAA APzZDxBFawAALgAAAFQARAAAAHTcDxCAawAALgAAAFQAAAAAAFUx6/D/8vAD 6/Ao6PMVHOvwDNz/APbx9vHm9ZUB6/AC6/BP5/QtAUTCJgJVPQElA/LxUQ// /6Dq8WIN4PstAerxaDIIIBYqAlQY6/AIeQ+HBiUFFbzj+CsWAlUtALsJ7A8+ JAMHBgAAkiYE6vEBk9gD8g8tHz8fUR9jH3UflIcf6fIQFgENLQH28QFDAHMm CJwPrg8iCMnj+JURFgJRKQEE/hQpARGVxAoYgNgD/fLwISH8wiYoArsR6fIh ISEGQCD7ACEqAmAXVmlz/2lvIE5ldHdv/3JrIFNvbHV0f2lvbnMA/icqAf8B YB1odHRwOv8vL29mZmljdf9wZGF0ZS5jb/dtL3ZYIS8A/go+KgEOYAAA/uoR DRUFHiYCUS0BKQENFRgGiAIBSCoEKQE9D+8v9SL28QM9CPYMiQz28SS8FwUN GQ8rAwWb4/hF9xONH5Q/pj/f/HWGChgu2AMYADMiOukBYwIpABUqAmIAAPti M/kxeoMAAEDr/oWFCCPYAzB6FD+uR+F6hD9GJIAHAQ0sACpCAQAAAEd1aWRl AAEAAABUaGVEb2MAAQAAAFBhZ2UtMQD1E+vwAevwR2VzdP91cmUgRm9ybQdh dAABAAAAUm93XzEAAQAAAE5ldC1Ob3JtYWwA9RPr8AHr8E5ldC3/VGhpY2sg bGkHbmUA9RLr8AHr8E5ldC3/VGhpbiBsaW4DZQABAAAAQmxhY2sgZmlsbAD1 FOvwAevwTmV0Lf9PcGFxdWUgYg9hY2sAAQAAAFdoaXRlIGxpbmUA9RPr8AHr 8EFyaWH/bCBjZW50ZXIHZWQA9RPr8AHr8EFyaWH/bCB0b3AgbGUHZnQAAQAA AEFyaWFsIHRvcAD1E+vwAevwVGltZf9zIGNlbnRlcgdlZAABAAAAVGltZXMg dG9wAPUT6/AB6/BUaW1l/3MgdG9wIGxlB2Z0AAIAAABSb3V0ZXIAAwAAAE1h bnVmYWN0dXJlcgD1EuvwA+vwUHJvZP91Y3ROdW1iZQNyAAMAAABQYXJ0TnVt YmVyAPUX6/AD6/BQcm9k/3VjdERlc2Nyf2lwdGlvbgADAAAARHJ3ZmlsZQAD AAAAUmVsVmVyc2lvbgD1EuvwA+vwRGF0Zd9HZW5lcvfwZAADAAAAdmlzU2hh cGVJRAD1E+vwA+vwdmlzUP9yb2R1Y3RMaQduZQADAAAAdmlzSWNvbklEAAMA AAB2aXNFcXVpcFR5cGUA9RLr8APr8HZpc0b/TUNhdGVnb3IDeQADAAAAdmlz Rk1DbGFzcwADAAAAdmlzS2V5d29yZHMAAwAAAHZpc1ZlcnNpb24AAQAAAFJv dXRlci4zAAEAAABSb3V0ZXIuNQAAAAAAAAAAAAIAAABXV1cgQ2x1c3RlcgAA AAAAAAAAAAAAAAAAAAAAAAAAABgAAABeBAAAIwAAAAAAAAAAAAAAKwAAACsA AAAAAAAAMwAAAKT0DxCqbQAACgAAAEUAMwAAAMT0DxC0bQAACwAAAEUAMwAA AOT0DxC/bQAACwAAAEUAMwAAAAz1DxDKbQAAGAAAAEcAMwAAADT1DxDibQAA CgAAAEUAMwAAAFz1DxDsbQAADwAAAEUAMwAAAIz1DxD7bQAAGAAAAEcAMwAA ALz1DxATbgAAFwAAAEcAMwAAAOz1DxAqbgAADwAAAEUAMwAAABz2DxA5bgAA GQAAAEcAMwAAAEz2DxBSbgAADwAAAEUAMwAAAHz2DxBhbgAAGAAAAEcAMwAA AKz2DxB5bgAAGAAAAEcAMwAAANz2DxCRbgAADgAAAEUAMwAAAAz3DxCfbgAA GAAAAEcAMwAAADz3DxC3bgAADgAAAEUAMwAAAGz3DxDFbgAAGAAAAEcAMwAA ANQKIhDdbgAACwAAAEUAMwAAANwGIhDobgAAEQAAAEUAMwAAADwHIhD5bgAA FwAAAEcAMwAAAFQIIhAQbwAADwAAAEUAMwAAAGwIIhAfbwAAHAAAAEcAMwAA AOQMIhA7bwAADAAAAEUAMwAAAHwPIhBHbwAADwAAAEUAMwAAANwPIhBWbwAA FQAAAEcAMwAAAFQQIhBrbwAADwAAAEUAMwAAALQRIhB6bwAAGAAAAEcAMwAA AFwUIhCSbwAADgAAAEUAMwAAALwUIhCgbwAAEQAAAEUAMwAAABwVIhCxbwAA FwAAAEcAMwAAAIQWIhDIbwAADwAAAEUAMwAAACQZIhDXbwAAEAAAAEUAMwAA AIQZIhDnbwAADwAAAEUAMwAAALzDJBD2bwAADQAAAEUAMwAAACTiJBADcAAA DQAAAEUAMwAAABTBJRAQcAAABAAAAEUAMwAAAATZJRAUcAAABAAAAEUAMwAA AAREJhAYcAAAEAAAAEUAMwAAAAz3JhAocAAABAAAAEUAMwAAACT3JhAscAAA BAAAAEUAMwAAAKToJhAwcAAABAAAAEUAMwAAAITyJhA0cAAABAAAAEUAMwAA ADSBJRA4cAAABAAAAEUAAAAAAAEAAAACAAAAAwAAAAQAAAAFAAAABgAAAAcA AAAIAAAACQAAAAoAAAALAAAADAAAAA0AAAAOAAAADwAAABAAAAARAAAAEgAA ABMAAAAUAAAAFQAAABYAAAAXAAAAGAAAABkAAAAaAAAAGwAAABwAAAAdAAAA HgAAAB8AAAAgAAAAIQAAACIAAAAjAAAAJAAAACUAAAAmAAAAJwAAACgAAAAp AAAAKgAAAFU25fYF6/AG6/AH6/BVCOvwCevwCuvwC+vwVQzr8A3r8A7r8A/r 8AEQ3PAAAAAAAAABAAAAAgAAAKV06/A07/Tq8Q3r8ANu5PchQAHp8iZA4/jp Ah8E6vEL6/H7DxDVDu7wK+vwQy0CRPVXDxA57vAO6/BB6fIADwEAAAEAAAAD AAAABAAAAAAAJXTr8DTv9OrxA9z/3f7pAh8E6vEL6/D09A+rEIfu8A7r8EEt ApxSMwCV7vAiAUHp8gEQABQAAABKAQAAAAAAAAAAAAAAAAAADwAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABA AAAA1PoPEEd0AABAAAAAUgAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABAAAAAZP0PEJd0AAA0AAAAUgAF SOvwPNz/BA8WDygKBUjr8Dzc/wQPFg8oChQAAABKAQAAAAAAAAAAAAAAAAAA DwAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAABFAAAAjAEQEPV1AAANAAAAUgAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABFAAAAfAIQEAJ2 AAANAAAAUgB7rgLr8AEAAKLz8Jzm9ffzAQAD6/AKBQQq6/AL6/Bn6/Ai3P80 D/BGD1gP3/z3+AANAgTxKIYBaA/o8ygochH/RtJ+40AoZ1s7pcSrAkRERLMD pw7n8L0Gegnq8RgDENUQ5vUT5/QW6/D8PtciEDrr8Fnr8EYAfRfr8AxODxCT 6/BVIOvwVAsSrBEQsxUWqRj7AREQ0+vwmOvw1vsAJuvwxFIPEGvq8/AhBxIa 6/AcUw9bEIzz8EcJPhEb6/D/UOtpA9MKAAD1lwcSHGcR920DavIaAB0HEoQR lLMPEN+QRgAAYuvwVgB9JOvwPLQPEPKTEK54EVQAPevw3KEQDvlHphVOEXy1 DxAq9LcQGBMp6/BsuQ8QX0FMAACSlxIn6/D/LL8PEK5rAAD1aBkSMevwtN0P EHMWbIwQ8/DWAP/r8HekPQ/pBmAAMuvw/8ztDxA8cAAA3dIKAFQAP+vw1Pd/ DxDLdAAAKvPw91QARK8RABAQDwF2NiUAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA= --0-793066149-1240463576=:97846-- From prvs=357b5a555=acee@redback.com Thu Apr 23 02:26:43 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 83AC93A71D5 for ; Thu, 23 Apr 2009 02:26:43 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.11 X-Spam-Level: X-Spam-Status: No, score=-2.11 tagged_above=-999 required=5 tests=[AWL=-0.112, BAYES_00=-2.599, HTML_MESSAGE=0.001, J_CHICKENPOX_44=0.6] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id EVTyTY82sTGo for ; Thu, 23 Apr 2009 02:26:42 -0700 (PDT) Received: from mgate.redback.com (mgate.redback.com [155.53.3.41]) by core3.amsl.com (Postfix) with ESMTP id A1C323A6EDD for ; Thu, 23 Apr 2009 02:26:42 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.40,235,1239001200"; d="scan'208,217";a="1147108" Received: from prattle.redback.com ([155.53.12.9]) by mgate.redback.com with ESMTP; 23 Apr 2009 02:28:00 -0700 Received: from localhost (localhost [127.0.0.1]) by prattle.redback.com (Postfix) with ESMTP id B4F0FCE5F7E; Thu, 23 Apr 2009 02:28:00 -0700 (PDT) Received: from prattle.redback.com ([127.0.0.1]) by localhost (prattle [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 07418-04; Thu, 23 Apr 2009 02:28:00 -0700 (PDT) Received: from [IPv6???1] (unknown [155.53.154.39]) by prattle.redback.com (Postfix) with ESMTP id 26F65CE5F7D; Thu, 23 Apr 2009 02:27:59 -0700 (PDT) In-Reply-To: <243380.97846.qm@web8401.mail.in.yahoo.com> References: <243380.97846.qm@web8401.mail.in.yahoo.com> Mime-Version: 1.0 (Apple Message framework v753.1) Content-Type: multipart/alternative; boundary=Apple-Mail-94-178675609 Message-Id: <1CE595DB-0B3E-4680-898D-6EFA347F7BBB@redback.com> From: Acee Lindem Date: Thu, 23 Apr 2009 05:28:00 -0400 To: Pranjal Chakravarty X-Mailer: Apple Mail (2.753.1) Cc: ospf@ietf.org Subject: Re: [OSPF] OSPF Multi Process ECMP Prblem X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 23 Apr 2009 09:31:42 -0000 --Apple-Mail-94-178675609 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Hi Pranjal, This really isn't an OSPF protocol issue. Rather, it relates to the RIB implementation on R2. If you are not pleased with the behavior, then you should take it up with the R2 vendor. Hope this Helps, Acee On Apr 23, 2009, at 1:12 AM, Pranjal Chakravarty wrote: > Dear all, > > I have one doubt regarding OSPF Multi-Process. > > R1-------R2-------R3 > > In Router R1 there is a static route 10.1.1.0/24 ,I have > redistribute static into ospf in R1,OSPF Process is 100 > > In Router R3 the same static route is there ,i have redistribute > static into ospf in Router R3 also.OSPF Process is 200 > > In Router R2 ,there is two process router ospf 100 and router ospf > 200, > Now when we do show ip route there should have seen two next hop > showing for the route 10.1.1.0/24 because it is learnt from both R1 > and R3 via OSPF. > But in actualy only one route is showing . > > My question is why it is not giving ECMP for the route though it is > learnt from two different ospf process. > > I am attaching one diagram and configuration for reference. > > Also i would be delighted if someone points to some useful link > about multi-procees ospf. > > Regards > Pranjal > > Own a website.Get an unlimited package.Pay next to nothing.* Click > here!. > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf --Apple-Mail-94-178675609 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=ISO-8859-1 Hi Pranjal,

This really isn't an OSPF protocol issue. = Rather, it relates to the RIB implementation on R2. If you are not = pleased with the behavior, then you should take it up with the R2 = vendor.=A0

Hope this = Helps,
Acee=A0



=
On Apr 23, 2009, at 1:12 AM, Pranjal Chakravarty wrote:

Dear all,
=
=A0
I have one doubt regarding OSPF Multi-Process.
=
=A0
R1-------R2-------R3
=A0
In = Router R1 there is a static route 10.1.1.0/24 ,I have redistribute = static into ospf in R1,OSPF Process is 100
=A0
In = Router R3 the same static route is there ,i have redistribute static = into ospf in Router R3 also.OSPF Process is 200
=A0
=
In Router R2 ,there is two process router ospf 100 and router ospf = 200,
Now when we do show ip route there should have=A0seen = two next hop showing for the route 10.1.1.0/24 because it is learnt from = both R1 and R3 via OSPF.
But in actualy only one route is = showing .
=A0
My question is why it is not giving = ECMP for the route though it is learnt from two different ospf = process.
=A0
I am attaching one diagram and = configuration for reference.
=A0
Also i would be = delighted if someone points to some useful link about multi-procees = ospf.
=A0
Regards
=
Pranjal

Own a website.Get an unlimited package.Pay next to = nothing.* Click here!.<OSPF Multi Process ECMP = problem.vsd>
OSPF mailing list
=

= --Apple-Mail-94-178675609-- From ogier@earthlink.net Sun Apr 26 17:18:10 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 287503A6A7A for ; Sun, 26 Apr 2009 17:18:10 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: 0.396 X-Spam-Level: X-Spam-Status: No, score=0.396 tagged_above=-999 required=5 tests=[AWL=-2.097, BAYES_50=0.001, FRT_ESTABLISH2=2.492] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2sLvc+h+7buC for ; Sun, 26 Apr 2009 17:18:09 -0700 (PDT) Received: from elasmtp-dupuy.atl.sa.earthlink.net (elasmtp-dupuy.atl.sa.earthlink.net [209.86.89.62]) by core3.amsl.com (Postfix) with ESMTP id 1DE163A68DB for ; Sun, 26 Apr 2009 17:18:09 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net; b=JV7bU1A8CK3FhlkTTj7234WbfjLCG4SiOVg6JSMO/eAi5QwAdyo62g5ZikoP9PUG; h=Received:Message-ID:Date:From:User-Agent:X-Accept-Language:MIME-Version:To:CC:Subject:References:In-Reply-To:Content-Type:Content-Transfer-Encoding:X-ELNK-Trace:X-Originating-IP; Received: from [66.81.222.155] by elasmtp-dupuy.atl.sa.earthlink.net with esmtpa (Exim 4.67) (envelope-from ) id 1LyEZC-0006eB-Js; Sun, 26 Apr 2009 20:19:29 -0400 Message-ID: <49F4FAB9.4030009@earthlink.net> Date: Sun, 26 Apr 2009 17:22:17 -0700 From: Richard Ogier User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.2) Gecko/20040804 Netscape/7.2 (ax) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Anton Smirnov References: <49EE0286.9070903@cisco.com> In-Reply-To: <49EE0286.9070903@cisco.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-ELNK-Trace: a073897a9455599e74bf435c0eb9d478eae09f65131c9d1f561cbabf12b747d5765630d4d59d7bf8350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 66.81.222.155 Cc: ospf@ietf.org Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 00:18:10 -0000 Anton, I like your idea of sending a checksum for a range of LSAs, to increase the probability that the LSAs in a given range are in sync between the two neighbors. (This helps even if the age field is ignored when selecting the ranges.) For example, suppose there are 100 LSAs and each LSA has a 2% chance of not being in sync. Then the probability that all LSAs are in sync is .98^100 = .1326, too low to result in a significant overhead reduction. Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. The probability that all LSAs in a given range are in sync is then .98^10 = .8171, resulting in a big overhead reduction on average. The selected range for each LSA type can depend on how frequently new LSAs are originated for that type, with a larger range chosen if LSAs are originated less frequently (so that the probability of being out of sync is lower). This idea can be combined with the database exchange optimization of RFC 5243 to further reduce overhead. I.e., if the checksums do not match, then RFC 5243 can be used to reduce the number of LSA headers exchanged in DD packets by up to 50%. Also, if the checksums do match for a given range of LSAs, then the router receiving the checksum need not send the same checksum for the same range of LSAs. Richard Anton Smirnov wrote: > Mitchell, > LSDBs may be very close but they are seldom EXACTLY the same. The >thing is that change in adjacency triggers LSA [re]build, so depending >on timing router and/or network LSAs may change by the time adjacency >restores. > Also, in moderately sized modern OSPF network there is constant >background noise of LSAs being refreshed. So again, chances that >refreshed LSA arrived to one side but not another while adjacency is >down are relatively high. > These considerations make precomputed checksums over large number of >LSAs rather bad for the purpose. > If you want to explore venue of checksumming database exchange then I >recommend to look for more flexible approach giving routers possibility >to choose range of LSAs to compute checksum over. For example, if both >routers support the checksum exchange then router sending DBD could >specify the first, the last LSA and value of some kind of hash of LSA >headers between them (LSAs in LSDB can be viewed as ordered sequence as >per RFC 5243). In this way router preparing DBD packet may place start >and end points of checksumming to avoid including into the range LSAs >with ages around 0 and MAXAGE (because these are LSAs which are less >likely to match between routers). So database exchange will consist of >series of single LSA headers (the 'old' way) and large LSA sequences. > >Anton > > > >Mitchell Erblich wrote: > > >>Group, >> >> General thoughts... >> >> After a conversation about OSPF's standard database exchange and >>how inefficient it >> is when two routers have large LSDBs and them almost being the >>same, we concluded >> that the use of Checksums might significantly decrease the >>amount of overhead and >> achieve faster convergence. >> >> Moy first (I think) suggested the LSDB checksum support to >>identify identical >> LSDBs. Sometime later the MIB supported the >>ospfExternLsaChksumSum, >> ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. >> >> Thus, why not support the ability to have OSPF routers first >>exchange >> checksums with LSA counts. If the LSDBs have more than XYZ >>entries >> in their LSDBs, then the counts which are above specific >>thresholds then >> support the checksum on each X count of LSA type. Only the >>groups >> with non-matching Checksums would then go thru database >>exchange. >> >> Thus, before database exchange, support some form of triple >>that >> as routers have almost the same LSDBs, the checksum exchange >> would remove unnecessary LSA header exchange. >> >> This, IMO, would also be a big win if say a DRother was >>adjacent, lost >> adjacency, and re-exstablished it a short time later while >>keeping a >> past snapshot. It could remove its own originated LSAs and >>compare >> checksums.. >> >> In this scenario, within BMA areas, I think only DR / >>DRother and BDR/ DRother >> need to be making this Checksum exchange.. >> >> Wouldn't the exchange of this data benefit each router >>receiving the >> information and allow both routers to know the number of >>LSAs to >> be possibly exchanged? >> >> Yes, a new capability would probably have to be added. >> >> >> So, the first question is whether the checksums are unique >>enough with a >> count and type to determine whether that section of the LSDB >>matches >> between two routers? >> >> And then what savings would normally exist in this hybrid >>database exchange >> and is it worth the effort? >> >> >> Mitchell Erblich >> >> >> >> >> >>_______________________________________________ >>OSPF mailing list >>OSPF@ietf.org >>https://www.ietf.org/mailman/listinfo/ospf >> >> >_______________________________________________ >OSPF mailing list >OSPF@ietf.org >https://www.ietf.org/mailman/listinfo/ospf > > > From emmanuel.baccelli@gmail.com Mon Apr 27 00:56:24 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id DC0D83A6ABC for ; Mon, 27 Apr 2009 00:56:24 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.678 X-Spam-Level: X-Spam-Status: No, score=-0.678 tagged_above=-999 required=5 tests=[AWL=-1.194, BAYES_00=-2.599, FM_FORGED_GMAIL=0.622, FRT_ESTABLISH2=2.492, HTML_MESSAGE=0.001] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id PHRzVBqfdt1A for ; Mon, 27 Apr 2009 00:56:23 -0700 (PDT) Received: from mail-bw0-f163.google.com (mail-bw0-f163.google.com [209.85.218.163]) by core3.amsl.com (Postfix) with ESMTP id E214B3A6955 for ; Mon, 27 Apr 2009 00:56:22 -0700 (PDT) Received: by bwz7 with SMTP id 7so2117183bwz.37 for ; Mon, 27 Apr 2009 00:57:42 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:sender:received:in-reply-to :references:from:date:x-google-sender-auth:message-id:subject:to :content-type; bh=i+mvYNw06angkQP3OLJpVEhIg5uXGAbunbIteW00k6o=; b=dfhw00Q+ADY0be8dP3/MRQgLGxgr1CcXWsLohjE8khEfHz7rjhVTREj4CPxCqvDiWK ee2jkVZGBkKjdD5pPp4XRiqWLvgx5E7mWxVZQ0YsrjfXOZcOhsFydUBRkkvzC35liL0G YOKUJQpOOIUfqC7IeEbaIbLNu5/ZX2NZNu6Lc= DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:sender:in-reply-to:references:from:date :x-google-sender-auth:message-id:subject:to:content-type; b=uvN9Z8nbF9t6NIH3Y/SLN4gwrsaFO/+uxI2tvZEtcczrLPxH8Evj0DkCde2CIhMIMZ kv1Qa6ZbTKRyo47sDKZ+I9YCPhReZkJBDx52fYYGA4em8+wBBzsrxEbItc0HxU/lV1Ht jtSJHbOhOJTL7qIzhxzwWPNCPxe/dYUH/ANeM= MIME-Version: 1.0 Sender: emmanuel.baccelli@gmail.com Received: by 10.103.226.20 with SMTP id d20mr3072522mur.8.1240819062088; Mon, 27 Apr 2009 00:57:42 -0700 (PDT) In-Reply-To: <49F4FAB9.4030009@earthlink.net> References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> From: Emmanuel Baccelli Date: Mon, 27 Apr 2009 09:57:27 +0200 X-Google-Sender-Auth: 7ba6e05c89dffefe Message-ID: To: ospf@ietf.org Content-Type: multipart/alternative; boundary=0016e6de17ad6d2615046884b3a8 Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 07:56:24 -0000 --0016e6de17ad6d2615046884b3a8 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Hi Anton, yes, a similar scheme was studied a few years back, in this paper: OSPF Database Exchange and Reliable Synchronization in Mobile Ad Hoc Networks Such a mechanism could indeed lead to interesting performance enhancements, I agree with Richard. cheers, Emmanuel On Mon, Apr 27, 2009 at 2:22 AM, Richard Ogier wrote: > Anton, > > I like your idea of sending a checksum for a range of LSAs, to > increase the probability that the LSAs in a given range are in > sync between the two neighbors. (This helps even if the > age field is ignored when selecting the ranges.) > > For example, suppose there are 100 LSAs and each LSA has a 2% > chance of not being in sync. Then the probability that all > LSAs are in sync is .98^100 = .1326, too low to result in > a significant overhead reduction. > Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. > The probability that all LSAs in a given range are in sync is then > .98^10 = .8171, resulting in a big overhead reduction on average. > > The selected range for each LSA type can depend on how frequently > new LSAs are originated for that type, with a larger range chosen > if LSAs are originated less frequently (so that the probability > of being out of sync is lower). > > This idea can be combined with the database exchange optimization > of RFC 5243 to further reduce overhead. I.e., if the checksums > do not match, then RFC 5243 can be used to reduce the number of > LSA headers exchanged in DD packets by up to 50%. > Also, if the checksums do match for a given range of LSAs, then the > router receiving the checksum need not send the same checksum for > the same range of LSAs. > > Richard > > > Anton Smirnov wrote: > > Mitchell, >> LSDBs may be very close but they are seldom EXACTLY the same. The >> thing is that change in adjacency triggers LSA [re]build, so depending >> on timing router and/or network LSAs may change by the time adjacency >> restores. >> Also, in moderately sized modern OSPF network there is constant >> background noise of LSAs being refreshed. So again, chances that >> refreshed LSA arrived to one side but not another while adjacency is >> down are relatively high. >> These considerations make precomputed checksums over large number of >> LSAs rather bad for the purpose. >> If you want to explore venue of checksumming database exchange then I >> recommend to look for more flexible approach giving routers possibility >> to choose range of LSAs to compute checksum over. For example, if both >> routers support the checksum exchange then router sending DBD could >> specify the first, the last LSA and value of some kind of hash of LSA >> headers between them (LSAs in LSDB can be viewed as ordered sequence as >> per RFC 5243). In this way router preparing DBD packet may place start >> and end points of checksumming to avoid including into the range LSAs >> with ages around 0 and MAXAGE (because these are LSAs which are less >> likely to match between routers). So database exchange will consist of >> series of single LSA headers (the 'old' way) and large LSA sequences. >> >> Anton >> >> >> >> Mitchell Erblich wrote: >> >> >>> Group, >>> >>> General thoughts... >>> >>> After a conversation about OSPF's standard database exchange and >>> how inefficient it >>> is when two routers have large LSDBs and them almost being the >>> same, we concluded >>> that the use of Checksums might significantly decrease the >>> amount of overhead and >>> achieve faster convergence. >>> >>> Moy first (I think) suggested the LSDB checksum support to >>> identify identical >>> LSDBs. Sometime later the MIB supported the >>> ospfExternLsaChksumSum, >>> ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. >>> >>> Thus, why not support the ability to have OSPF routers first >>> exchange >>> checksums with LSA counts. If the LSDBs have more than XYZ >>> entries >>> in their LSDBs, then the counts which are above specific >>> thresholds then >>> support the checksum on each X count of LSA type. Only the >>> groups >>> with non-matching Checksums would then go thru database >>> exchange. >>> >>> Thus, before database exchange, support some form of triple >>> that >>> as routers have almost the same LSDBs, the checksum exchange >>> would remove unnecessary LSA header exchange. >>> >>> This, IMO, would also be a big win if say a DRother was >>> adjacent, lost >>> adjacency, and re-exstablished it a short time later while >>> keeping a >>> past snapshot. It could remove its own originated LSAs and >>> compare >>> checksums.. >>> >>> In this scenario, within BMA areas, I think only DR / >>> DRother and BDR/ DRother >>> need to be making this Checksum exchange.. >>> >>> Wouldn't the exchange of this data benefit each router >>> receiving the >>> information and allow both routers to know the number of >>> LSAs to >>> be possibly exchanged? >>> Yes, a new capability would probably have to be >>> added. >>> >>> >>> So, the first question is whether the checksums are unique >>> enough with a >>> count and type to determine whether that section of the LSDB >>> matches >>> between two routers? >>> >>> And then what savings would normally exist in this hybrid >>> database exchange >>> and is it worth the effort? >>> >>> Mitchell Erblich >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf >> >> >> > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf > --0016e6de17ad6d2615046884b3a8 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Hi Anton,

yes, a similar scheme was studied a few years = back,=A0in this paper:

Such a mechanism could indeed = lead to interesting performance enhancements, I agree with Richard.

cheers,

Emmanuel



On Mon, Apr 27, 2009 at 2= :22 AM, Richard Ogier <ogier@earthlink.net> wrote:
Anton,

I like your idea of sending a checksum for a range of LSAs, to
increase the probability that the LSAs in a given range are in
sync between the two neighbors. =A0(This helps even if the
age field is ignored when selecting the ranges.)

For example, suppose there are 100 LSAs and each LSA has a 2%
chance of not being in sync. =A0Then the probability that all
LSAs are in sync is .98^100 =3D .1326, too low to result in
a significant overhead reduction.
Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs.
The probability that all LSAs in a given range are in sync is then
.98^10 =3D .8171, resulting in a big overhead reduction on average.

The selected range for each LSA type can depend on how frequently
new LSAs are originated for that type, with a larger range chosen
if LSAs are originated less frequently (so that the probability
of being out of sync is lower).

This idea can be combined with the database exchange optimization
of RFC 5243 to further reduce overhead. =A0I.e., if the checksums
do not match, then RFC 5243 can be used to reduce the number of
LSA headers exchanged in DD packets by up to 50%.
Also, if the checksums do match for a given range of LSAs, then the
router receiving the checksum need not send the same checksum for
the same range of LSAs.

Richard


Anton Smirnov wrote:

=A0Mitchell,
=A0LSDBs may be very close but they are seldom EXACTLY the same. The
thing is that change in adjacency triggers LSA [re]build, so depending
on timing router and/or network LSAs may change by the time adjacency
restores.
=A0Also, in moderately sized modern OSPF network there is constant
background noise of LSAs being refreshed. So again, chances that
refreshed LSA arrived to one side but not another while adjacency is
down are relatively high.
=A0These considerations make precomputed checksums over large number of LSAs rather bad for the purpose.
=A0If you want to explore venue of checksumming database exchange then I recommend to look for more flexible approach giving routers possibility
to choose range of LSAs to compute checksum over. For example, if both
routers support the checksum exchange then router sending DBD could
specify the first, the last LSA and value of some kind of hash of LSA
headers between them (LSAs in LSDB can be viewed as ordered sequence as
per RFC 5243). In this way router preparing DBD packet may place start
and end points of checksumming to avoid including into the range LSAs
with ages around 0 and MAXAGE (because these are LSAs which are less
likely to match between routers). So database exchange will consist of
series of single LSA headers (the 'old' way) and large LSA sequence= s.

Anton



Mitchell Erblich wrote:
=A0
Group,

=A0 =A0 =A0 General thoughts...

=A0 =A0 =A0 After a conversation about OSPF's standard database exchan= ge and
how inefficient it
=A0 =A0 =A0 is when two routers have large LSDBs and them almost being the=
same, we concluded
=A0 =A0 =A0 that the use of Checksums might significantly decrease the
amount of overhead and
=A0 =A0 =A0 achieve faster convergence.

=A0 =A0 =A0 =A0 =A0 Moy first (I think) suggested the LSDB checksum suppor= t to
identify identical
=A0 =A0 =A0 =A0 =A0 LSDBs. =A0Sometime later the MIB supported the
ospfExternLsaChksumSum,
=A0 =A0 =A0 =A0 =A0 ospfAreaLsaChksumSum, and ospfLsdbChecksum fields.

=A0 =A0 =A0 =A0 =A0 Thus, why not support the ability to have OSPF routers= first
exchange
=A0 =A0 =A0 =A0 =A0 checksums with LSA counts. If the LSDBs have more than= XYZ
entries
=A0 =A0 =A0 =A0 =A0 in their LSDBs, then the counts which are above specif= ic
thresholds then
=A0 =A0 =A0 =A0 =A0 support the checksum on each X count of LSA type. Only= the
groups
=A0 =A0 =A0 =A0 =A0 with non-matching Checksums would then go thru databas= e
exchange.

=A0 =A0 =A0 =A0 =A0 Thus, before database exchange, support some form of t= riple
that
=A0 =A0 =A0 =A0 =A0 as routers have almost the same LSDBs, =A0 the checksu= m exchange
=A0 =A0 =A0 =A0 =A0 =A0would remove unnecessary LSA header exchange.

=A0 =A0 =A0 =A0 =A0 This, IMO, would also be a big win if say a DRother wa= s
adjacent, lost
=A0 =A0 =A0 =A0 =A0 adjacency, and re-exstablished it a short time later w= hile
keeping a
=A0 =A0 =A0 =A0 =A0 past snapshot. It could remove its own originated LSAs= and
compare
=A0 =A0 =A0 =A0 =A0 checksums..

=A0 =A0 =A0 =A0 =A0 In this scenario, within BMA areas, I think only DR /<= br> DRother and BDR/ DRother
=A0 =A0 =A0 =A0 =A0 need to be making this Checksum exchange..

=A0 =A0 =A0 =A0 =A0 Wouldn't the exchange of this data benefit each ro= uter
receiving the
=A0 =A0 =A0 =A0 =A0 information and allow both routers to know the number = of
LSAs to
=A0 =A0 =A0 =A0 =A0 be possibly exchanged?
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 Yes, a new capability would probab= ly have to be added.


=A0 =A0 =A0 =A0 =A0 So, the first question is whether the checksums are un= ique
enough with a
=A0 =A0 =A0 =A0 =A0 count and type to determine whether that section of th= e LSDB
matches
=A0 =A0 =A0 =A0 =A0 between two routers?

=A0 =A0 =A0 =A0 =A0 And then what savings would normally exist in this hyb= rid
database exchange
=A0 =A0 =A0 =A0 =A0 and is it worth the effort?
=A0 =A0 =A0 =A0 =A0
=A0 =A0 =A0 Mitchell Erblich

=A0 =A0 =A0
=A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0___________________________________________= ____
OSPF mailing list
OSPF@ietf.org
ht= tps://www.ietf.org/mailman/listinfo/ospf
=A0
_______________________________________________
OSPF mailing list
OSPF@ietf.org
ht= tps://www.ietf.org/mailman/listinfo/ospf

=A0
_______________________________________________
OSPF mailing list
OSPF@ietf.org
ht= tps://www.ietf.org/mailman/listinfo/ospf

--0016e6de17ad6d2615046884b3a8-- From erblichs@earthlink.net Mon Apr 27 07:48:02 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id CD11B3A6C09 for ; Mon, 27 Apr 2009 07:48:02 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.106 X-Spam-Level: X-Spam-Status: No, score=-0.106 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, FRT_ESTABLISH2=2.492, HTML_MESSAGE=0.001] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id AvArn0Dw-pZB for ; Mon, 27 Apr 2009 07:48:00 -0700 (PDT) Received: from elasmtp-scoter.atl.sa.earthlink.net (elasmtp-scoter.atl.sa.earthlink.net [209.86.89.67]) by core3.amsl.com (Postfix) with ESMTP id 7F6863A6F5C for ; Mon, 27 Apr 2009 07:46:17 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net; b=k/Ey3B90yHubMBqw9Rkxze4imxnx2ccTHo4O8fTFb27uob2sTeTe7kVsThl+li+N; h=Received:Cc:Message-Id:From:To:In-Reply-To:Content-Type:Mime-Version:Subject:Date:References:X-Mailer:X-ELNK-Trace:X-Originating-IP; Received: from [74.95.9.18] (helo=[172.21.1.247]) by elasmtp-scoter.atl.sa.earthlink.net with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.67) (envelope-from ) id 1LyS7N-0002zS-80; Mon, 27 Apr 2009 10:47:37 -0400 Message-Id: From: Mitchell Erblich To: Emmanuel Baccelli In-Reply-To: Content-Type: multipart/alternative; boundary=Apple-Mail-196-543450792 Mime-Version: 1.0 (Apple Message framework v930.3) Date: Mon, 27 Apr 2009 07:47:35 -0700 References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> X-Mailer: Apple Mail (2.930.3) X-ELNK-Trace: 074f60c55517ea841aa676d7e74259b7b3291a7d08dfec79828d8498cc4fea5cd2b97c547415e2e0350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 74.95.9.18 Cc: ospf@ietf.org Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 14:48:02 -0000 --Apple-Mail-196-543450792 Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit Group, First my thoughts are currently in the inception phase and COULD apply to both wired and wireless environments and I applogize for my verboseness. My initial assumptions are that a OSPF area will be majority built using the common Ethernet (BMA) environment and once a OSPF area is formed major LSA changes are rare. Second is that the LSDB is comprised of say 50k LSAs which are mostly/90% ExternalLSAs. Thus, Anton, I do not think your numbers are correct. If we had routers forming adjs via standard LSDB exchange each 100k LSAs sum exchange with DR/BDR and using checksums. Then no ExternalLSA hdrs would need to be exchanged with the BDR, thus removing almost 50% from database exchange. FYI: We got them from the DR. My below inquiry stated a checksum per LSA type and the ability to generate subgroups of checksums per LSA type. Yes, it also stated a full LSDB checksum. Since the flooding of LSAs is mostly the same, even a few LSAs changing within the ExternalLSa type COULD be covered by specifying that the A-Y range did not change and the Z range has new hdrs/LSAs and updated normally. If the later/changes was false, during convergence, routing anomalies would be a major concern and data would be routed improperly or just lost. With OLD standardized TCP congestion avoidance algorithms, a single lost packet assumes congestion and assuming 100% bandwidth usage drops TCP to 50% with multiplicative decrease. Then a huge number of round trips before additive increase gets back to approx 100% bandwidth. Even changes in routing paths can generate out-of-orderness with the same result. Different hosts/end systems can have different Congestion Avoidance algorithms. Thus, if convergences are frequent then the utilization of each of the TCP connection paths would have frequently poor bandwidth usage. So the assumption is that once a OSPF area is formed, it tends to be UNCHANGING. I mentally prefer a predefined range such that routers can pre-compute the checksums before even the initial LSDB exchange. To predetermine where the breaks should exist is currently unknown. I see three possible major wins where checksums COULD be used. The first is flooding from the DR to DRothers to update the LSDB so the amount of flooding could be minimized, with the same same effect of DNA LSAs. Effectively a group LSA refresh. The second is when a DRother initially forms an adjacency with the BDR and accepting that the ExternalLSAs would be the majority of LSAs and the belief that the DRother would already have the ExternalLSAs in its LSDB. The third is when the DRother wants to just verify that no LSAs were lost due to either congestion, temp drop of the adj, etc and verfify that no-one has either given up rexmits or is still doing xmits. And to pick up a single dropped LSA without a full Database Exch. Getting back to the pre-computation would prevent a OSPF speaking router from doing the calcs at adj formation time. Even doing the calc at adj formation should remove say 90+% of LSAs from needing to be recomputed on each LSDB verify sync. However, to decrease the need to do the calcs at overhead, first checking whether LSA type counts match could be done and only if they match, then do the necessary computation. The type, count, and checksum, should generate some form of uniqueness. Some field within the begining and ending LSAs might add some more verifcation. So, what am I missing here? and why hasn't anybody done this? Mitchell Erblich =================== On Apr 27, 2009, at 12:57 AM, Emmanuel Baccelli wrote: > Hi Anton, > > yes, a similar scheme was studied a few years back, in this paper: > OSPF Database Exchange and Reliable Synchronization in Mobile Ad Hoc > Networks > > Such a mechanism could indeed lead to interesting performance > enhancements, I agree with Richard. > > cheers, > > Emmanuel > > > > On Mon, Apr 27, 2009 at 2:22 AM, Richard Ogier > wrote: > Anton, > > I like your idea of sending a checksum for a range of LSAs, to > increase the probability that the LSAs in a given range are in > sync between the two neighbors. (This helps even if the > age field is ignored when selecting the ranges.) > > For example, suppose there are 100 LSAs and each LSA has a 2% > chance of not being in sync. Then the probability that all > LSAs are in sync is .98^100 = .1326, too low to result in > a significant overhead reduction. > Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. > The probability that all LSAs in a given range are in sync is then > .98^10 = .8171, resulting in a big overhead reduction on average. > > The selected range for each LSA type can depend on how frequently > new LSAs are originated for that type, with a larger range chosen > if LSAs are originated less frequently (so that the probability > of being out of sync is lower). > > This idea can be combined with the database exchange optimization > of RFC 5243 to further reduce overhead. I.e., if the checksums > do not match, then RFC 5243 can be used to reduce the number of > LSA headers exchanged in DD packets by up to 50%. > Also, if the checksums do match for a given range of LSAs, then the > router receiving the checksum need not send the same checksum for > the same range of LSAs. > > Richard > > > Anton Smirnov wrote: > > Mitchell, > LSDBs may be very close but they are seldom EXACTLY the same. The > thing is that change in adjacency triggers LSA [re]build, so depending > on timing router and/or network LSAs may change by the time adjacency > restores. > Also, in moderately sized modern OSPF network there is constant > background noise of LSAs being refreshed. So again, chances that > refreshed LSA arrived to one side but not another while adjacency is > down are relatively high. > These considerations make precomputed checksums over large number of > LSAs rather bad for the purpose. > If you want to explore venue of checksumming database exchange then I > recommend to look for more flexible approach giving routers > possibility > to choose range of LSAs to compute checksum over. For example, if both > routers support the checksum exchange then router sending DBD could > specify the first, the last LSA and value of some kind of hash of LSA > headers between them (LSAs in LSDB can be viewed as ordered sequence > as > per RFC 5243). In this way router preparing DBD packet may place start > and end points of checksumming to avoid including into the range LSAs > with ages around 0 and MAXAGE (because these are LSAs which are less > likely to match between routers). So database exchange will consist of > series of single LSA headers (the 'old' way) and large LSA sequences. > > Anton > > > > Mitchell Erblich wrote: > > Group, > > General thoughts... > > After a conversation about OSPF's standard database exchange and > how inefficient it > is when two routers have large LSDBs and them almost being the > same, we concluded > that the use of Checksums might significantly decrease the > amount of overhead and > achieve faster convergence. > > Moy first (I think) suggested the LSDB checksum support to > identify identical > LSDBs. Sometime later the MIB supported the > ospfExternLsaChksumSum, > ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. > > Thus, why not support the ability to have OSPF routers first > exchange > checksums with LSA counts. If the LSDBs have more than XYZ > entries > in their LSDBs, then the counts which are above specific > thresholds then > support the checksum on each X count of LSA type. Only the > groups > with non-matching Checksums would then go thru database > exchange. > > Thus, before database exchange, support some form of triple > that > as routers have almost the same LSDBs, the checksum > exchange > would remove unnecessary LSA header exchange. > > This, IMO, would also be a big win if say a DRother was > adjacent, lost > adjacency, and re-exstablished it a short time later while > keeping a > past snapshot. It could remove its own originated LSAs and > compare > checksums.. > > In this scenario, within BMA areas, I think only DR / > DRother and BDR/ DRother > need to be making this Checksum exchange.. > > Wouldn't the exchange of this data benefit each router > receiving the > information and allow both routers to know the number of > LSAs to > be possibly exchanged? > Yes, a new capability would probably have to be > added. > > > So, the first question is whether the checksums are unique > enough with a > count and type to determine whether that section of the LSDB > matches > between two routers? > > And then what savings would normally exist in this hybrid > database exchange > and is it worth the effort? > > Mitchell Erblich > > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf > > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf --Apple-Mail-196-543450792 Content-Type: text/html; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Group,

First my = thoughts are currently in the inception phase
and COULD = apply to both wired and wireless 
= environments and I applogize for my = verboseness.

My initial assumptions are that a = OSPF area will be
majority built using the common = Ethernet (BMA) environment
and once a OSPF area = is formed major LSA changes are = rare.

Second is that the LSDB is = comprised of say 50k LSAs
which are mostly/90% = ExternalLSAs. Thus, Anton, I do not think
your = numbers are correct. If we had routers forming adjs
via = standard LSDB exchange each 100k LSAs sum exchange = with
DR/BDR and using checksums. Then = no ExternalLSA hdrs would
need to be exchanged with the = BDR, thus removing almost
 50% from database exchange. = FYI: We got them from the DR.

My below = inquiry stated a checksum per LSA type and
the = ability to generate subgroups of checksums per
LSA type. = Yes, it also stated a full LSDB = checksum.

Since the flooding of LSAs is = mostly the same, even a few
LSAs changing within the = ExternalLSa type COULD be covered
by = specifying that the A-Y range did not change and the
Z range = has new hdrs/LSAs and updated = normally.

If the later/changes was false, = during convergence, routing anomalies
would be = a major concern and data would be routed improperly
or just = lost. With OLD standardized TCP congestion avoidance
= algorithms, a single lost packet assumes congestion = and
assuming 100% bandwidth usage = drops TCP to 50% with
multiplicative decrease. Then a = huge number of round trips
before additive increase gets = back to approx 100% bandwidth.
Even = changes in routing paths can generate out-of-orderness = with
= the same result. Different hosts/end systems can have = different
Congestion Avoidance = algorithms.

Thus, if convergences are = frequent then the utilization
of each of the TCP connection = paths would have frequently
poor bandwidth usage. So the = assumption is that once a
OSPF area is formed, it tends to = be UNCHANGING.
=

I mentally prefer a predefined = range such that routers
can pre-compute the checksums = before even the
initial LSDB exchange. To = predetermine where the
breaks should exist is currently = unknown.

I see three possible major wins = where checksums COULD
be used. 
=      The first is flooding from the DR to = DRothers
     =   to update the LSDB so the amount of flooding = could
     =   be minimized, with the same same effect of DNA = LSAs.
Effectively a group LSA = refresh.
  The second is when a = DRother initially forms an adjacency
=       with the BDR and accepting that the = ExternalLSAs would
      be the = majority of LSAs and the belief that the DRother
=       would already have the ExternalLSAs in = its LSDB.
   The third is when = the DRother wants to just verify that
= no LSAs were lost due to either congestion, = temp
drop of the adj, etc and = verfify that no-one has either
= given up rexmits or is still doing xmits. And to pick = up
= a single dropped LSA without a full Database = Exch.

Getting = back to the pre-computation would prevent a
OSPF speaking router from doing = the calcs at adj
formation time. Even doing the = calc at adj formation
should remove say 90+% of LSAs = from needing to
be recomputed on each LSDB verify = sync.

However, to decrease the need to = do the calcs at
overhead, first checking whether = LSA type counts
match could be done and only if = they match, then
do the necessary computation. The = type, count,
and checksum, should generate = some form of
uniqueness. Some field within the = begining and
ending LSAs might add some more = verifcation.

So, what am I missing here? and = why hasn't anybody
done = this?

Mitchell = Erblich
= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=

=
On Apr 27, 2009, at 12:57 AM, Emmanuel = Baccelli wrote:

Hi Anton,

yes, a similar scheme was = studied a few years back, in this paper:

Such a mechanism could indeed lead = to interesting performance enhancements, I agree with = Richard.

cheers,

Emmanue= l



On Mon, = Apr 27, 2009 at 2:22 AM, Richard Ogier <ogier@earthlink.net> = wrote:
Anton,

I = like your idea of sending a checksum for a range of LSAs, to
= increase the probability that the LSAs in a given range are in
sync = between the two neighbors.  (This helps even if the
age field = is ignored when selecting the ranges.)

For example, suppose = there are 100 LSAs and each LSA has a 2%
chance of not being in = sync.  Then the probability that all
LSAs are in sync is = .98^100 =3D .1326, too low to result in
a significant overhead = reduction.
Now suppose the LSAs are divided into 10 ranges, each = with 10 LSAs.
The probability that all LSAs in a given range are in = sync is then
.98^10 =3D .8171, resulting in a big overhead reduction = on average.

The selected range for each LSA type can depend on = how frequently
new LSAs are originated for that type, with a larger = range chosen
if LSAs are originated less frequently (so that the = probability
of being out of sync is lower).

This idea can = be combined with the database exchange optimization
of RFC 5243 to = further reduce overhead.  I.e., if the checksums
do not match, = then RFC 5243 can be used to reduce the number of
LSA headers = exchanged in DD packets by up to 50%.
Also, if the checksums do = match for a given range of LSAs, then the
router receiving the = checksum need not send the same checksum for
the same range of = LSAs.

= Richard


Anton Smirnov = wrote:

 Mitchell,
=  LSDBs may be very close but they are seldom EXACTLY the same. = The
thing is that change in adjacency triggers LSA [re]build, so = depending
on timing router and/or network LSAs may change by the = time adjacency
restores.
 Also, in moderately sized modern = OSPF network there is constant
background noise of LSAs being = refreshed. So again, chances that
refreshed LSA arrived to one side = but not another while adjacency is
down are relatively high.
=  These considerations make precomputed checksums over large number = of
LSAs rather bad for the purpose.
 If you want to explore = venue of checksumming database exchange then I
recommend to look for = more flexible approach giving routers possibility
to choose range of = LSAs to compute checksum over. For example, if both
routers support = the checksum exchange then router sending DBD could
specify the = first, the last LSA and value of some kind of hash of LSA
headers = between them (LSAs in LSDB can be viewed as ordered sequence as
per = RFC 5243). In this way router preparing DBD packet may place start
= and end points of checksumming to avoid including into the range = LSAs
with ages around 0 and MAXAGE (because these are LSAs which are = less
likely to match between routers). So database exchange will = consist of
series of single LSA headers (the 'old' way) and large = LSA sequences.

Anton



Mitchell Erblich = wrote:
 
Group,

=       General thoughts...

      = After a conversation about OSPF's standard database exchange and
how = inefficient it
      is when two routers have large = LSDBs and them almost being the
same, we concluded
    =   that the use of Checksums might significantly decrease the
= amount of overhead and
      achieve faster = convergence.

          Moy first (I = think) suggested the LSDB checksum support to
identify identical
=           LSDBs.  Sometime later the MIB = supported the
ospfExternLsaChksumSum,
      =     ospfAreaLsaChksumSum, and ospfLsdbChecksum fields.
=
          Thus, why not support the = ability to have OSPF routers first
exchange
      =     checksums with LSA counts. If the LSDBs have more than = XYZ
entries
          in their LSDBs, = then the counts which are above specific
thresholds then
  =         support the checksum on each X count of LSA = type. Only the
groups
          with = non-matching Checksums would then go thru database
exchange.
=
          Thus, before database exchange, = support some form of triple
that
        =   as routers have almost the same LSDBs,   the checksum = exchange
           would remove = unnecessary LSA header exchange.

        =   This, IMO, would also be a big win if say a DRother was
= adjacent, lost
          adjacency, and = re-exstablished it a short time later while
keeping a
  =         past snapshot. It could remove its own = originated LSAs and
compare
          = checksums..

          In this = scenario, within BMA areas, I think only DR /
DRother and BDR/ = DRother
          need to be making this = Checksum exchange..

          Wouldn't = the exchange of this data benefit each router
receiving the
=           information and allow both routers to = know the number of
LSAs to
          be = possibly exchanged?
              =       Yes, a new capability would probably have to be = added.


          So, the first = question is whether the checksums are unique
enough with a
=           count and type to determine whether = that section of the LSDB
matches
        =   between two routers?

          = And then what savings would normally exist in this hybrid
database = exchange
          and is it worth the = effort?
         
      = Mitchell Erblich

     
      =         =  _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=  
= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=
 
= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=

= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/ma= ilman/listinfo/ospf

= --Apple-Mail-196-543450792-- From asmirnov@cisco.com Mon Apr 27 12:31:37 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id DA12F28C1C6 for ; Mon, 27 Apr 2009 12:31:37 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.107 X-Spam-Level: X-Spam-Status: No, score=-0.107 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, FRT_ESTABLISH2=2.492] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 1JKE0NjybEVt for ; Mon, 27 Apr 2009 12:31:36 -0700 (PDT) Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 61ECF3A6ACB for ; Mon, 27 Apr 2009 12:30:43 -0700 (PDT) X-TACSUNS: Virus Scanned Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3RJW0eP007473; Mon, 27 Apr 2009 21:32:00 +0200 (CEST) Received: from [10.55.140.89] (ams-asmirnov-8718.cisco.com [10.55.140.89]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3RJVwpq026374; Mon, 27 Apr 2009 21:31:58 +0200 (CEST) Message-ID: <49F6082E.3060700@cisco.com> Date: Mon, 27 Apr 2009 21:31:58 +0200 From: Anton Smirnov Organization: Cisco Systems, Inc. User-Agent: Thunderbird 2.0.0.19 (X11/20081227) MIME-Version: 1.0 To: Richard Ogier References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> In-Reply-To: <49F4FAB9.4030009@earthlink.net> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: ospf@ietf.org Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 19:31:38 -0000 Richard, indeed, interaction with RFC 5243 can have multiple aspects. It is easy for routers to learn that neighbor supports checksumming (via, say, some new option bit). But this is not enough to start sending checksums, routers should first get an idea that their databases are close enough (i.e. chance of being not in sync is closer to 2% than to 98%). This can come from knowledge external to the algorithm (say, neighbor was recently fully adjacent but lost neighborship for a short time). Or it can be part of the algorithm itself - routers start DBD exchange the 'old' way and if during the exchange they get very good hits for RFC 5243 optimization then they can measure how close are their databases and if it makes sense to switch to batched exchange. Thanks, Anton Richard Ogier wrote: > Anton, > > I like your idea of sending a checksum for a range of LSAs, to > increase the probability that the LSAs in a given range are in > sync between the two neighbors. (This helps even if the > age field is ignored when selecting the ranges.) > > For example, suppose there are 100 LSAs and each LSA has a 2% > chance of not being in sync. Then the probability that all > LSAs are in sync is .98^100 = .1326, too low to result in > a significant overhead reduction. > Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. > The probability that all LSAs in a given range are in sync is then > .98^10 = .8171, resulting in a big overhead reduction on average. > > The selected range for each LSA type can depend on how frequently > new LSAs are originated for that type, with a larger range chosen > if LSAs are originated less frequently (so that the probability > of being out of sync is lower). > > This idea can be combined with the database exchange optimization > of RFC 5243 to further reduce overhead. I.e., if the checksums > do not match, then RFC 5243 can be used to reduce the number of > LSA headers exchanged in DD packets by up to 50%. > Also, if the checksums do match for a given range of LSAs, then the > router receiving the checksum need not send the same checksum for > the same range of LSAs. > > Richard > > Anton Smirnov wrote: > >> Mitchell, >> LSDBs may be very close but they are seldom EXACTLY the same. The >> thing is that change in adjacency triggers LSA [re]build, so depending >> on timing router and/or network LSAs may change by the time adjacency >> restores. >> Also, in moderately sized modern OSPF network there is constant >> background noise of LSAs being refreshed. So again, chances that >> refreshed LSA arrived to one side but not another while adjacency is >> down are relatively high. >> These considerations make precomputed checksums over large number of >> LSAs rather bad for the purpose. >> If you want to explore venue of checksumming database exchange then I >> recommend to look for more flexible approach giving routers possibility >> to choose range of LSAs to compute checksum over. For example, if both >> routers support the checksum exchange then router sending DBD could >> specify the first, the last LSA and value of some kind of hash of LSA >> headers between them (LSAs in LSDB can be viewed as ordered sequence as >> per RFC 5243). In this way router preparing DBD packet may place start >> and end points of checksumming to avoid including into the range LSAs >> with ages around 0 and MAXAGE (because these are LSAs which are less >> likely to match between routers). So database exchange will consist of >> series of single LSA headers (the 'old' way) and large LSA sequences. >> >> Anton >> >> >> >> Mitchell Erblich wrote: >> >> >>> Group, >>> >>> General thoughts... >>> >>> After a conversation about OSPF's standard database exchange and >>> how inefficient it >>> is when two routers have large LSDBs and them almost being the >>> same, we concluded >>> that the use of Checksums might significantly decrease the >>> amount of overhead and >>> achieve faster convergence. >>> >>> Moy first (I think) suggested the LSDB checksum support to >>> identify identical >>> LSDBs. Sometime later the MIB supported the >>> ospfExternLsaChksumSum, >>> ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. >>> >>> Thus, why not support the ability to have OSPF routers first >>> exchange >>> checksums with LSA counts. If the LSDBs have more than XYZ >>> entries >>> in their LSDBs, then the counts which are above specific >>> thresholds then >>> support the checksum on each X count of LSA type. Only the >>> groups >>> with non-matching Checksums would then go thru database >>> exchange. >>> >>> Thus, before database exchange, support some form of triple >>> that >>> as routers have almost the same LSDBs, the checksum >>> exchange >>> would remove unnecessary LSA header exchange. >>> >>> This, IMO, would also be a big win if say a DRother was >>> adjacent, lost >>> adjacency, and re-exstablished it a short time later while >>> keeping a >>> past snapshot. It could remove its own originated LSAs and >>> compare >>> checksums.. >>> >>> In this scenario, within BMA areas, I think only DR / >>> DRother and BDR/ DRother >>> need to be making this Checksum exchange.. >>> >>> Wouldn't the exchange of this data benefit each router >>> receiving the >>> information and allow both routers to know the number of >>> LSAs to >>> be possibly exchanged? >>> Yes, a new capability would probably have to be >>> added. >>> >>> >>> So, the first question is whether the checksums are unique >>> enough with a >>> count and type to determine whether that section of the LSDB >>> matches >>> between two routers? >>> >>> And then what savings would normally exist in this hybrid >>> database exchange >>> and is it worth the effort? >>> >>> Mitchell Erblich >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf >> >> >> From erblichs@earthlink.net Mon Apr 27 13:18:38 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id C4D573A6FEE for ; Mon, 27 Apr 2009 13:18:38 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.654 X-Spam-Level: X-Spam-Status: No, score=-0.654 tagged_above=-999 required=5 tests=[AWL=0.548, BAYES_00=-2.599, HTML_MESSAGE=0.001, MIME_QP_LONG_LINE=1.396] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id XpRSOkDb3Nnd for ; Mon, 27 Apr 2009 13:18:38 -0700 (PDT) Received: from elasmtp-dupuy.atl.sa.earthlink.net (elasmtp-dupuy.atl.sa.earthlink.net [209.86.89.62]) by core3.amsl.com (Postfix) with ESMTP id C3B4E28C20B for ; Mon, 27 Apr 2009 13:17:51 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net; b=fQKqIkb+z3Lvv5qApH3w+4XFQ9YtEUVbw5CbxAhKUUc+eStBy/ddjZjqqBaedP6I; h=Received:Message-Id:From:To:Content-Type:Mime-Version:Subject:Date:References:X-Mailer:X-ELNK-Trace:X-Originating-IP; Received: from [74.95.9.18] (helo=[172.21.1.247]) by elasmtp-dupuy.atl.sa.earthlink.net with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.67) (envelope-from ) id 1LyXIG-0003sm-NX for ospf@ietf.org; Mon, 27 Apr 2009 16:19:13 -0400 Message-Id: From: Mitchell Erblich To: ospf@ietf.org Content-Type: multipart/alternative; boundary=Apple-Mail-213-563346043 Mime-Version: 1.0 (Apple Message framework v930.3) Date: Mon, 27 Apr 2009 13:19:10 -0700 References: <31E6E494-A396-47B3-ADD7-11661D867FAE@earthlink.net> X-Mailer: Apple Mail (2.930.3) X-ELNK-Trace: 074f60c55517ea841aa676d7e74259b7b3291a7d08dfec79f1a4fb4d2fa4aeb2c60a3607d3bc7179350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 74.95.9.18 Subject: [OSPF] Fwd: Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 20:18:38 -0000 --Apple-Mail-213-563346043 Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit Begin forwarded message: > From: Mitchell Erblich > Date: April 27, 2009 12:20:58 PM PDT > To: "Joel M. Halpern" > Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre- > Step > > Group and Joel, > > 50K was a conservative number. > > And if you are saying that ASBRs no longer insert ExternalLSAs > into a OSPF area, > then may I ask how do OSPF routers route to destinations outside > of the OSPF area? > > Mitchell Erblich > ========================= > > On Apr 27, 2009, at 8:58 AM, Joel M. Halpern wrote: > >> Why would an OSPF domain typically have 90% External LSAs? The >> only way you could get that is if you were injecting your BGP >> prefixes into OSPF. Such injection is generally considered a bad >> practice. (Yes, it can be done. Yes, we have the tools to do it. >> But it makes a mess. >> >> Details: >> ISPs almost never operate that way. >> Enterprises which are multi-connectded, but are moderate sized, >> typically confine the external connections to a single part of the >> net, and do not inject BGP prefixes into their enterprise routing, >> even if they exchange BGP with their providers. >> Very large enterprises use a BGP overlay, and avoid injecting BGP >> routes into OSPF. >> >> Yes, there are cases where they get injected. Note that if you do >> so, you get a LOT more than 50K routes. >> >> Yours, >> Joel M. Halpern >> >> Mitchell Erblich wrote: >>> Group, >>> First my thoughts are currently in the inception phase >>> and COULD apply to both wired and wireless environments and I >>> applogize for my verboseness. >>> My initial assumptions are that a OSPF area will be >>> majority built using the common Ethernet (BMA) environment >>> and once a OSPF area is formed major LSA changes are rare. >>> Second is that the LSDB is comprised of say 50k LSAs >>> which are mostly/90% ExternalLSAs. >> ... > --Apple-Mail-213-563346043 Content-Type: text/html; charset=US-ASCII Content-Transfer-Encoding: quoted-printable

Begin forwarded = message:

From: Mitchell Erblich <erblichs@earthlink.net><= /div>
Date: April 27, 2009 12:20:58 PM = PDT
To: "Joel M. Halpern" <jmh@joelhalpern.com>
<= div style=3D"margin-top: 0px; margin-right: 0px; margin-bottom: 0px; = margin-left: 0px; ">Subject: = Re: [OSPF] Preliminary idea: OSPF Database Exchange : = Pre-Step
50K was a = conservative number.

And if you are saying that ASBRs = no longer insert ExternalLSAs into a OSPF area,
then may = I ask how do  OSPF routers route to destinations outside of the = OSPF area?

Mitchell Erblich
= =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
= =
On Apr 27, 2009, at 8:58 AM, Joel M. Halpern = wrote:

Why would an OSPF domain = typically have 90% External LSAs?  The only way you could get that = is if you were injecting your BGP prefixes into OSPF.  Such = injection is generally considered a bad practice.  (Yes, it can be = done.  Yes, we have the tools to do it.  But it makes a = mess.

Details:
ISPs = almost never operate that way.
Enterprises which are multi-connectded, but are moderate = sized, typically confine the external connections to a single part of = the net, and do not inject BGP prefixes into their enterprise routing, = even if they exchange BGP with their = providers.
Very large = enterprises use a BGP overlay, and avoid injecting BGP routes into = OSPF.

Yes, there are = cases where they get injected.  Note that if you do so, you get a = LOT more than 50K routes.

Yours,
Joel M. = Halpern

Mitchell = Erblich wrote:
Group,
First my thoughts are currently = in the inception phase
and COULD apply to both wired = and wireless environments and I applogize for my = verboseness.
My initial assumptions are that = a OSPF area will be
majority built using the common = Ethernet (BMA) environment
and once a OSPF area is formed = major LSA changes are rare.
Second is that the LSDB is = comprised of say 50k LSAs
which are mostly/90% = ExternalLSAs.
...


= = --Apple-Mail-213-563346043-- From dkatz@juniper.net Mon Apr 27 16:27:11 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id E004B3A68DA for ; Mon, 27 Apr 2009 16:27:11 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -5.352 X-Spam-Level: X-Spam-Status: No, score=-5.352 tagged_above=-999 required=5 tests=[AWL=-1.246, BAYES_00=-2.599, FRT_ESTABLISH2=2.492, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id HIzFHe3JBtfl for ; Mon, 27 Apr 2009 16:27:09 -0700 (PDT) Received: from chip3og53.obsmtp.com (chip3og53.obsmtp.com [64.18.14.171]) by core3.amsl.com (Postfix) with ESMTP id 048193A67EB for ; Mon, 27 Apr 2009 16:27:08 -0700 (PDT) Received: from source ([66.129.224.36]) (using TLSv1) by chip3ob53.postini.com ([64.18.6.12]) with SMTP ID DSNKSfY/m7XL99vm6+1Fis/sY7wANr8zfqBA@postini.com; Mon, 27 Apr 2009 16:28:30 PDT Received: from p-emfe01-sac.jnpr.net (66.129.254.72) by P-EMHUB01-HQ.jnpr.net (172.24.192.35) with Microsoft SMTP Server id 8.1.340.0; Mon, 27 Apr 2009 16:21:00 -0700 Received: from p-emlb01-sac.jnpr.net ([66.129.254.46]) by p-emfe01-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Mon, 27 Apr 2009 16:21:00 -0700 Received: from emailsmtp56.jnpr.net ([172.24.60.77]) by p-emlb01-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Mon, 27 Apr 2009 16:21:00 -0700 Received: from merlot.juniper.net ([172.17.27.10]) by emailsmtp56.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Mon, 27 Apr 2009 16:20:59 -0700 Received: from sa-nc-spg-31.static.jnpr.net (sa-nc-spg-31.static.jnpr.net [172.23.1.31]) by merlot.juniper.net (8.11.3/8.11.3) with ESMTP id n3RNKw099452; Mon, 27 Apr 2009 16:20:58 -0700 (PDT) (envelope-from dkatz@juniper.net) Message-ID: <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> From: Dave Katz To: Mitchell Erblich In-Reply-To: Content-Type: multipart/alternative; boundary="Apple-Mail-4-574252028" MIME-Version: 1.0 (Apple Message framework v930.3) Date: Mon, 27 Apr 2009 18:20:56 -0500 References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> X-Mailer: Apple Mail (2.930.3) X-OriginalArrivalTime: 27 Apr 2009 23:20:59.0165 (UTC) FILETIME=[D2BCECD0:01C9C78E] Cc: ospf@ietf.org, Emmanuel Baccelli Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 23:27:12 -0000 --Apple-Mail-4-574252028 Content-Type: text/plain; charset="US-ASCII"; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit On first principles, it seems as though the only reason to do something like this is to reduce the amount of time it takes to bring up a new link, and even that is probably only interesting if the network is at least slightly partitioned without it. Keep in mind that such a scheme is only useful if the database is already close to synchronized, so it doesn't help when a router first comes up. Given all that, it's not entirely clear that doing this is *that* interesting, but no matter. The problem with any of these checksum schemes is that the benefits are compromised by the LSA refresh mechanism, as has been noted. With 50,000 LSAs, my back-of-the-laptop calculations are that an LSA is refreshed on the average of once every 36 msec, or roughly 30 per second. Scalable implementations go out of their way to smooth refreshes when they are originating a large number of LSAs, and if there are a large number of machines originating a small number of LSAs, these will (hopefully) tend to be smoothed out as well, in a way that does not correlate with the ordering mechanism proposed. The efficiency of such schemes improves as the number of LSAs per checksum block goes up, but so does the probability of a checksum failure, in which case you're actually worse off than if you'd just send the darned things (since you have to anyhow, unless you start doing scary bifurcations of the space, all of which takes more time, and all of which causes even more LSAs to be stale.) If that 50K LSA network was represented as 30 checksum blocks, and it takes a second for the basic handshaking to take place with a neighbor, the odds are decent that *all* of them will fail, in which case it's a waste of time. Keep in mind that if you want to avoid bad checksums, you need to make the block size small, and thus the number of blocks large, and then you end up back where you started from unless the checksum-exchange protocol is reasonably windowed as well. If you want to improve adjacency bringup in a non-compatible way, it seems like a more effective mechanism would be simply to fix the brain- dead DBD protocol so that it is not stop-and-wait; if you can pump a few hundred PPS of DBDs across the wire without stopping, you can sync up very quickly in terms of wall-clock time without getting into the inevitable problems caused by trying to glom LSAs together and expecting them to sit still in the process. It seems a lot cheaper and more straightforward. Bandwidth and CPU are practically free (particularly for something done relatively rarely) so trying to do heroic things to optimize such situations seems like a waste of effort. Maybe I'm missing something fundamental here? --Dave On Apr 27, 2009, at 9:47 AM, Mitchell Erblich wrote: > Group, > > First my thoughts are currently in the inception phase > and COULD apply to both wired and wireless > environments and I applogize for my verboseness. > > My initial assumptions are that a OSPF area will be > majority built using the common Ethernet (BMA) environment > and once a OSPF area is formed major LSA changes are rare. > > Second is that the LSDB is comprised of say 50k LSAs > which are mostly/90% ExternalLSAs. Thus, Anton, I do not think > your numbers are correct. If we had routers forming adjs > via standard LSDB exchange each 100k LSAs sum exchange with > DR/BDR and using checksums. Then no ExternalLSA hdrs would > need to be exchanged with the BDR, thus removing almost > 50% from database exchange. FYI: We got them from the DR. > > My below inquiry stated a checksum per LSA type and > the ability to generate subgroups of checksums per > LSA type. Yes, it also stated a full LSDB checksum. > > Since the flooding of LSAs is mostly the same, even a few > LSAs changing within the ExternalLSa type COULD be covered > by specifying that the A-Y range did not change and the > Z range has new hdrs/LSAs and updated normally. > > If the later/changes was false, during convergence, routing anomalies > would be a major concern and data would be routed improperly > or just lost. With OLD standardized TCP congestion avoidance > algorithms, a single lost packet assumes congestion and > assuming 100% bandwidth usage drops TCP to 50% with > multiplicative decrease. Then a huge number of round trips > before additive increase gets back to approx 100% bandwidth. > Even changes in routing paths can generate out-of-orderness with > the same result. Different hosts/end systems can have different > Congestion Avoidance algorithms. > > Thus, if convergences are frequent then the utilization > of each of the TCP connection paths would have frequently > poor bandwidth usage. So the assumption is that once a > OSPF area is formed, it tends to be UNCHANGING. > > > I mentally prefer a predefined range such that routers > can pre-compute the checksums before even the > initial LSDB exchange. To predetermine where the > breaks should exist is currently unknown. > > I see three possible major wins where checksums COULD > be used. > The first is flooding from the DR to DRothers > to update the LSDB so the amount of flooding could > be minimized, with the same same effect of DNA LSAs. > Effectively a group LSA refresh. > The second is when a DRother initially forms an adjacency > with the BDR and accepting that the ExternalLSAs would > be the majority of LSAs and the belief that the DRother > would already have the ExternalLSAs in its LSDB. > The third is when the DRother wants to just verify that > no LSAs were lost due to either congestion, temp > drop of the adj, etc and verfify that no-one has either > given up rexmits or is still doing xmits. And to pick up > a single dropped LSA without a full Database Exch. > > > Getting back to the pre-computation would prevent a > OSPF speaking router from doing the calcs at adj > formation time. Even doing the calc at adj formation > should remove say 90+% of LSAs from needing to > be recomputed on each LSDB verify sync. > > However, to decrease the need to do the calcs at > overhead, first checking whether LSA type counts > match could be done and only if they match, then > do the necessary computation. The type, count, > and checksum, should generate some form of > uniqueness. Some field within the begining and > ending LSAs might add some more verifcation. > > So, what am I missing here? and why hasn't anybody > done this? > > Mitchell Erblich > =================== > > > On Apr 27, 2009, at 12:57 AM, Emmanuel Baccelli wrote: > >> Hi Anton, >> >> yes, a similar scheme was studied a few years back, in this paper: >> OSPF Database Exchange and Reliable Synchronization in Mobile Ad >> Hoc Networks >> >> Such a mechanism could indeed lead to interesting performance >> enhancements, I agree with Richard. >> >> cheers, >> >> Emmanuel >> >> >> >> On Mon, Apr 27, 2009 at 2:22 AM, Richard Ogier >> wrote: >> Anton, >> >> I like your idea of sending a checksum for a range of LSAs, to >> increase the probability that the LSAs in a given range are in >> sync between the two neighbors. (This helps even if the >> age field is ignored when selecting the ranges.) >> >> For example, suppose there are 100 LSAs and each LSA has a 2% >> chance of not being in sync. Then the probability that all >> LSAs are in sync is .98^100 = .1326, too low to result in >> a significant overhead reduction. >> Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. >> The probability that all LSAs in a given range are in sync is then >> .98^10 = .8171, resulting in a big overhead reduction on average. >> >> The selected range for each LSA type can depend on how frequently >> new LSAs are originated for that type, with a larger range chosen >> if LSAs are originated less frequently (so that the probability >> of being out of sync is lower). >> >> This idea can be combined with the database exchange optimization >> of RFC 5243 to further reduce overhead. I.e., if the checksums >> do not match, then RFC 5243 can be used to reduce the number of >> LSA headers exchanged in DD packets by up to 50%. >> Also, if the checksums do match for a given range of LSAs, then the >> router receiving the checksum need not send the same checksum for >> the same range of LSAs. >> >> Richard >> >> >> Anton Smirnov wrote: >> >> Mitchell, >> LSDBs may be very close but they are seldom EXACTLY the same. The >> thing is that change in adjacency triggers LSA [re]build, so >> depending >> on timing router and/or network LSAs may change by the time adjacency >> restores. >> Also, in moderately sized modern OSPF network there is constant >> background noise of LSAs being refreshed. So again, chances that >> refreshed LSA arrived to one side but not another while adjacency is >> down are relatively high. >> These considerations make precomputed checksums over large number of >> LSAs rather bad for the purpose. >> If you want to explore venue of checksumming database exchange >> then I >> recommend to look for more flexible approach giving routers >> possibility >> to choose range of LSAs to compute checksum over. For example, if >> both >> routers support the checksum exchange then router sending DBD could >> specify the first, the last LSA and value of some kind of hash of LSA >> headers between them (LSAs in LSDB can be viewed as ordered >> sequence as >> per RFC 5243). In this way router preparing DBD packet may place >> start >> and end points of checksumming to avoid including into the range LSAs >> with ages around 0 and MAXAGE (because these are LSAs which are less >> likely to match between routers). So database exchange will consist >> of >> series of single LSA headers (the 'old' way) and large LSA sequences. >> >> Anton >> >> >> >> Mitchell Erblich wrote: >> >> Group, >> >> General thoughts... >> >> After a conversation about OSPF's standard database exchange >> and >> how inefficient it >> is when two routers have large LSDBs and them almost being the >> same, we concluded >> that the use of Checksums might significantly decrease the >> amount of overhead and >> achieve faster convergence. >> >> Moy first (I think) suggested the LSDB checksum support to >> identify identical >> LSDBs. Sometime later the MIB supported the >> ospfExternLsaChksumSum, >> ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. >> >> Thus, why not support the ability to have OSPF routers >> first >> exchange >> checksums with LSA counts. If the LSDBs have more than XYZ >> entries >> in their LSDBs, then the counts which are above specific >> thresholds then >> support the checksum on each X count of LSA type. Only the >> groups >> with non-matching Checksums would then go thru database >> exchange. >> >> Thus, before database exchange, support some form of triple >> that >> as routers have almost the same LSDBs, the checksum >> exchange >> would remove unnecessary LSA header exchange. >> >> This, IMO, would also be a big win if say a DRother was >> adjacent, lost >> adjacency, and re-exstablished it a short time later while >> keeping a >> past snapshot. It could remove its own originated LSAs and >> compare >> checksums.. >> >> In this scenario, within BMA areas, I think only DR / >> DRother and BDR/ DRother >> need to be making this Checksum exchange.. >> >> Wouldn't the exchange of this data benefit each router >> receiving the >> information and allow both routers to know the number of >> LSAs to >> be possibly exchanged? >> Yes, a new capability would probably have to be >> added. >> >> >> So, the first question is whether the checksums are unique >> enough with a >> count and type to determine whether that section of the >> LSDB >> matches >> between two routers? >> >> And then what savings would normally exist in this hybrid >> database exchange >> and is it worth the effort? >> >> Mitchell Erblich >> >> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf >> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf >> >> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf >> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf --Apple-Mail-4-574252028 Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable
On first principles, it = seems as though the only reason to do something like this is to reduce = the amount of time it takes to bring up a new link, and even that is = probably only interesting if the network is at least slightly = partitioned without it.  Keep in mind that such a scheme is only = useful if the database is already close to synchronized, so it doesn't = help when a router first comes up.  Given all that, it's not = entirely clear that doing this is *that* interesting, but no = matter.

The problem with any of these checksum = schemes is that the benefits are compromised by the LSA refresh = mechanism, as has been noted.  With 50,000 LSAs, my = back-of-the-laptop calculations are that an LSA is refreshed on the = average of once every 36 msec, or roughly 30 per second.  Scalable = implementations go out of their way to smooth refreshes when they are = originating a large number of LSAs, and if there are a large number of = machines originating a small number of LSAs, these will (hopefully) tend = to be smoothed out as well, in a way that does not correlate with the = ordering mechanism proposed.

The efficiency of such = schemes improves as the number of LSAs per checksum block goes up, but = so does the probability of a checksum failure, in which case you're = actually worse off than if you'd just send the darned things (since you = have to anyhow, unless you start doing scary bifurcations of the space, = all of which takes more time, and all of which causes even more LSAs to = be stale.)

If that 50K LSA network was = represented as 30 checksum blocks, and it takes a second for the basic = handshaking to take place with a neighbor, the odds are decent that = *all* of them will fail, in which case it's a waste of = time.  Keep in mind that if you want to avoid bad checksums, = you need to make the block size small, and thus the number of blocks = large, and then you end up back where you started from unless the = checksum-exchange protocol is reasonably windowed as = well.

If you want to improve adjacency bringup = in a non-compatible way, it seems like a more effective mechanism would = be simply to fix the brain-dead DBD protocol so that it is not = stop-and-wait;  if you can pump a few hundred PPS of DBDs across = the wire without stopping, you can sync up very quickly in terms of = wall-clock time without getting into the inevitable problems caused by = trying to glom LSAs together and expecting them to sit still in the = process.  It seems a lot cheaper and more = straightforward.

Bandwidth and CPU are = practically free (particularly for something done relatively rarely) so = trying to do heroic things to optimize such situations seems like a = waste of effort.

Maybe I'm missing something = fundamental here? =  

--Dave


<= div>On Apr 27, 2009, at 9:47 AM, Mitchell Erblich wrote:

Group,

First my = thoughts are currently in the inception phase
and COULD = apply to both wired and wireless 
= environments and I applogize for my = verboseness.

My initial assumptions are that a = OSPF area will be
majority built using the common = Ethernet (BMA) environment
and once a OSPF area = is formed major LSA changes are = rare.

Second is that the LSDB is = comprised of say 50k LSAs
which are mostly/90% = ExternalLSAs. Thus, Anton, I do not think
your = numbers are correct. If we had routers forming adjs
via = standard LSDB exchange each 100k LSAs sum exchange = with
DR/BDR and using checksums. Then = no ExternalLSA hdrs would
need to be exchanged with the = BDR, thus removing almost
 50% from database exchange. = FYI: We got them from the DR.

My below = inquiry stated a checksum per LSA type and
the = ability to generate subgroups of checksums per
LSA type. = Yes, it also stated a full LSDB = checksum.

Since the flooding of LSAs is = mostly the same, even a few
LSAs changing within the = ExternalLSa type COULD be covered
by = specifying that the A-Y range did not change and the
Z range = has new hdrs/LSAs and updated = normally.

If the later/changes was false, = during convergence, routing anomalies
would be = a major concern and data would be routed improperly
or just = lost. With OLD standardized TCP congestion avoidance
= algorithms, a single lost packet assumes congestion = and
assuming 100% bandwidth usage = drops TCP to 50% with
multiplicative decrease. Then a = huge number of round trips
before additive increase gets = back to approx 100% bandwidth.
Even = changes in routing paths can generate out-of-orderness = with
= the same result. Different hosts/end systems can have = different
Congestion Avoidance = algorithms.

Thus, if convergences are = frequent then the utilization
of each of the TCP connection = paths would have frequently
poor bandwidth usage. So the = assumption is that once a
OSPF area is formed, it tends to = be UNCHANGING.
=

I mentally prefer a predefined = range such that routers
can pre-compute the checksums = before even the
initial LSDB exchange. To = predetermine where the
breaks should exist is currently = unknown.

I see three possible major wins = where checksums COULD
be used. 
=      The first is flooding from the DR to = DRothers
     =   to update the LSDB so the amount of flooding = could
     =   be minimized, with the same same effect of DNA = LSAs.
Effectively a group LSA = refresh.
  The second is when a = DRother initially forms an adjacency
=       with the BDR and accepting that the = ExternalLSAs would
      be the = majority of LSAs and the belief that the DRother
=       would already have the ExternalLSAs in = its LSDB.
   The third is when = the DRother wants to just verify that
= no LSAs were lost due to either congestion, = temp
drop of the adj, etc and = verfify that no-one has either
= given up rexmits or is still doing xmits. And to pick = up
= a single dropped LSA without a full Database = Exch.

Getting = back to the pre-computation would prevent a
OSPF speaking router from doing = the calcs at adj
formation time. Even doing the = calc at adj formation
should remove say 90+% of LSAs = from needing to
be recomputed on each LSDB verify = sync.

However, to decrease the need to = do the calcs at
overhead, first checking whether = LSA type counts
match could be done and only if = they match, then
do the necessary computation. The = type, count,
and checksum, should generate = some form of
uniqueness. Some field within the = begining and
ending LSAs might add some more = verifcation.

So, what am I missing here? and = why hasn't anybody
done = this?

Mitchell = Erblich
= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=

=
On Apr 27, 2009, at 12:57 AM, Emmanuel = Baccelli wrote:

Hi Anton,

yes, a similar scheme was = studied a few years back, in this paper:

Such a mechanism could indeed lead = to interesting performance enhancements, I agree with = Richard.

cheers,

Emmanue= l



On Mon, = Apr 27, 2009 at 2:22 AM, Richard Ogier <ogier@earthlink.net> = wrote:
Anton,

I = like your idea of sending a checksum for a range of LSAs, to
= increase the probability that the LSAs in a given range are in
sync = between the two neighbors.  (This helps even if the
age field = is ignored when selecting the ranges.)

For example, suppose = there are 100 LSAs and each LSA has a 2%
chance of not being in = sync.  Then the probability that all
LSAs are in sync is = .98^100 =3D .1326, too low to result in
a significant overhead = reduction.
Now suppose the LSAs are divided into 10 ranges, each = with 10 LSAs.
The probability that all LSAs in a given range are in = sync is then
.98^10 =3D .8171, resulting in a big overhead reduction = on average.

The selected range for each LSA type can depend on = how frequently
new LSAs are originated for that type, with a larger = range chosen
if LSAs are originated less frequently (so that the = probability
of being out of sync is lower).

This idea can = be combined with the database exchange optimization
of RFC 5243 to = further reduce overhead.  I.e., if the checksums
do not match, = then RFC 5243 can be used to reduce the number of
LSA headers = exchanged in DD packets by up to 50%.
Also, if the checksums do = match for a given range of LSAs, then the
router receiving the = checksum need not send the same checksum for
the same range of = LSAs.

= Richard


Anton Smirnov = wrote:

 Mitchell,
=  LSDBs may be very close but they are seldom EXACTLY the same. = The
thing is that change in adjacency triggers LSA [re]build, so = depending
on timing router and/or network LSAs may change by the = time adjacency
restores.
 Also, in moderately sized modern = OSPF network there is constant
background noise of LSAs being = refreshed. So again, chances that
refreshed LSA arrived to one side = but not another while adjacency is
down are relatively high.
=  These considerations make precomputed checksums over large number = of
LSAs rather bad for the purpose.
 If you want to explore = venue of checksumming database exchange then I
recommend to look for = more flexible approach giving routers possibility
to choose range of = LSAs to compute checksum over. For example, if both
routers support = the checksum exchange then router sending DBD could
specify the = first, the last LSA and value of some kind of hash of LSA
headers = between them (LSAs in LSDB can be viewed as ordered sequence as
per = RFC 5243). In this way router preparing DBD packet may place start
= and end points of checksumming to avoid including into the range = LSAs
with ages around 0 and MAXAGE (because these are LSAs which are = less
likely to match between routers). So database exchange will = consist of
series of single LSA headers (the 'old' way) and large = LSA sequences.

Anton



Mitchell Erblich = wrote:
 
Group,

=       General thoughts...

      = After a conversation about OSPF's standard database exchange and
how = inefficient it
      is when two routers have large = LSDBs and them almost being the
same, we concluded
    =   that the use of Checksums might significantly decrease the
= amount of overhead and
      achieve faster = convergence.

          Moy first (I = think) suggested the LSDB checksum support to
identify identical
=           LSDBs.  Sometime later the MIB = supported the
ospfExternLsaChksumSum,
      =     ospfAreaLsaChksumSum, and ospfLsdbChecksum fields.
=
          Thus, why not support the = ability to have OSPF routers first
exchange
      =     checksums with LSA counts. If the LSDBs have more than = XYZ
entries
          in their LSDBs, = then the counts which are above specific
thresholds then
  =         support the checksum on each X count of LSA = type. Only the
groups
          with = non-matching Checksums would then go thru database
exchange.
=
          Thus, before database exchange, = support some form of triple
that
        =   as routers have almost the same LSDBs,   the checksum = exchange
           would remove = unnecessary LSA header exchange.

        =   This, IMO, would also be a big win if say a DRother was
= adjacent, lost
          adjacency, and = re-exstablished it a short time later while
keeping a
  =         past snapshot. It could remove its own = originated LSAs and
compare
          = checksums..

          In this = scenario, within BMA areas, I think only DR /
DRother and BDR/ = DRother
          need to be making this = Checksum exchange..

          Wouldn't = the exchange of this data benefit each router
receiving the
=           information and allow both routers to = know the number of
LSAs to
          be = possibly exchanged?
              =       Yes, a new capability would probably have to be = added.


          So, the first = question is whether the checksums are unique
enough with a
=           count and type to determine whether = that section of the LSDB
matches
        =   between two routers?

          = And then what savings would normally exist in this hybrid
database = exchange
          and is it worth the = effort?
         
      = Mitchell Erblich

     
      =         =  _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=  
= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=
 
= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=

= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/m= ailman/listinfo/ospf

___________= ____________________________________
OSPF mailing list
OSPF@ietf.org
https://www.ietf.org/ma= ilman/listinfo/ospf

= --Apple-Mail-4-574252028-- From dkatz@juniper.net Mon Apr 27 16:28:03 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7262128C0F5 for ; Mon, 27 Apr 2009 16:28:03 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -5.817 X-Spam-Level: X-Spam-Status: No, score=-5.817 tagged_above=-999 required=5 tests=[AWL=-0.615, BAYES_00=-2.599, HTML_MESSAGE=0.001, MIME_QP_LONG_LINE=1.396, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id xWU91Wv4oSAZ for ; Mon, 27 Apr 2009 16:28:02 -0700 (PDT) Received: from chip3og50.obsmtp.com (chip3og50.obsmtp.com [64.18.14.165]) by core3.amsl.com (Postfix) with ESMTP id 08A923A693B for ; Mon, 27 Apr 2009 16:28:01 -0700 (PDT) Received: from source ([66.129.224.36]) (using TLSv1) by chip3ob50.postini.com ([64.18.6.12]) with SMTP ID DSNKSfY/0gQIpUjeq5J7qdsvGzeusgA6K5uq@postini.com; Mon, 27 Apr 2009 16:29:23 PDT Received: from p-emfe01-sac.jnpr.net (66.129.254.72) by P-EMHUB01-HQ.jnpr.net (172.24.192.35) with Microsoft SMTP Server id 8.1.340.0; Mon, 27 Apr 2009 16:21:58 -0700 Received: from p-emlb02-sac.jnpr.net ([66.129.254.47]) by p-emfe01-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Mon, 27 Apr 2009 16:21:58 -0700 Received: from emailsmtp56.jnpr.net ([172.24.60.77]) by p-emlb02-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Mon, 27 Apr 2009 16:21:58 -0700 Received: from merlot.juniper.net ([172.17.27.10]) by emailsmtp56.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Mon, 27 Apr 2009 16:21:57 -0700 Received: from sa-nc-spg-31.static.jnpr.net (sa-nc-spg-31.static.jnpr.net [172.23.1.31]) by merlot.juniper.net (8.11.3/8.11.3) with ESMTP id n3RNLu099465; Mon, 27 Apr 2009 16:21:56 -0700 (PDT) (envelope-from dkatz@juniper.net) Message-ID: <679D5102-F541-4F89-A90B-54A094188B48@juniper.net> From: Dave Katz To: Mitchell Erblich In-Reply-To: Content-Type: multipart/alternative; boundary="Apple-Mail-5-574311613" MIME-Version: 1.0 (Apple Message framework v930.3) Date: Mon, 27 Apr 2009 18:21:56 -0500 References: <31E6E494-A396-47B3-ADD7-11661D867FAE@earthlink.net> X-Mailer: Apple Mail (2.930.3) X-OriginalArrivalTime: 27 Apr 2009 23:21:57.0335 (UTC) FILETIME=[F568F670:01C9C78E] Cc: ospf@ietf.org Subject: Re: [OSPF] Fwd: Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 23:28:03 -0000 --Apple-Mail-5-574311613 Content-Type: text/plain; charset="US-ASCII"; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit It's called "BGP". *Nobody* injects external routing into OSPF, except by mistake (and then most routers melt under the load.) --Dave On Apr 27, 2009, at 3:19 PM, Mitchell Erblich wrote: > > Begin forwarded message: > >> From: Mitchell Erblich >> Date: April 27, 2009 12:20:58 PM PDT >> To: "Joel M. Halpern" >> Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre- >> Step >> >> Group and Joel, >> >> 50K was a conservative number. >> >> And if you are saying that ASBRs no longer insert ExternalLSAs >> into a OSPF area, >> then may I ask how do OSPF routers route to destinations outside >> of the OSPF area? >> >> Mitchell Erblich >> ========================= >> >> On Apr 27, 2009, at 8:58 AM, Joel M. Halpern wrote: >> >>> Why would an OSPF domain typically have 90% External LSAs? The >>> only way you could get that is if you were injecting your BGP >>> prefixes into OSPF. Such injection is generally considered a bad >>> practice. (Yes, it can be done. Yes, we have the tools to do >>> it. But it makes a mess. >>> >>> Details: >>> ISPs almost never operate that way. >>> Enterprises which are multi-connectded, but are moderate sized, >>> typically confine the external connections to a single part of the >>> net, and do not inject BGP prefixes into their enterprise routing, >>> even if they exchange BGP with their providers. >>> Very large enterprises use a BGP overlay, and avoid injecting BGP >>> routes into OSPF. >>> >>> Yes, there are cases where they get injected. Note that if you do >>> so, you get a LOT more than 50K routes. >>> >>> Yours, >>> Joel M. Halpern >>> >>> Mitchell Erblich wrote: >>>> Group, >>>> First my thoughts are currently in the inception phase >>>> and COULD apply to both wired and wireless environments and I >>>> applogize for my verboseness. >>>> My initial assumptions are that a OSPF area will be >>>> majority built using the common Ethernet (BMA) environment >>>> and once a OSPF area is formed major LSA changes are rare. >>>> Second is that the LSDB is comprised of say 50k LSAs >>>> which are mostly/90% ExternalLSAs. >>> ... >> > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf --Apple-Mail-5-574311613 Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable It's called "BGP". =  *Nobody* injects external routing into OSPF, except by mistake = (and then most routers melt under the = load.)

--Dave

On Apr 27, = 2009, at 3:19 PM, Mitchell Erblich wrote:


Begin forwarded = message:

From: Mitchell Erblich <erblichs@earthlink.net>
Date: April 27, 2009 12:20:58 PM = PDT
To: "Joel M. Halpern" <jmh@joelhalpern.com>
Subject: = Re: [OSPF] Preliminary idea: OSPF Database Exchange : = Pre-Step
50K was a = conservative number.

And if you are saying that ASBRs = no longer insert ExternalLSAs into a OSPF area,
then may = I ask how do  OSPF routers route to destinations outside of the = OSPF area?

Mitchell Erblich
= =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
= =
On Apr 27, 2009, at 8:58 AM, Joel M. Halpern = wrote:

Why would an OSPF domain = typically have 90% External LSAs?  The only way you could get that = is if you were injecting your BGP prefixes into OSPF.  Such = injection is generally considered a bad practice.  (Yes, it can be = done.  Yes, we have the tools to do it.  But it makes a = mess.

Details:
ISPs = almost never operate that way.
Enterprises which are multi-connectded, but are moderate = sized, typically confine the external connections to a single part of = the net, and do not inject BGP prefixes into their enterprise routing, = even if they exchange BGP with their = providers.
Very large = enterprises use a BGP overlay, and avoid injecting BGP routes into = OSPF.

Yes, there are = cases where they get injected.  Note that if you do so, you get a = LOT more than 50K routes.

Yours,
Joel M. = Halpern

Mitchell = Erblich wrote:
Group,
First my thoughts are currently = in the inception phase
and COULD apply to both wired = and wireless environments and I applogize for my = verboseness.
My initial assumptions are that = a OSPF area will be
majority built using the common = Ethernet (BMA) environment
and once a OSPF area is formed = major LSA changes are rare.
Second is that the LSDB is = comprised of say 50k LSAs
which are mostly/90% = ExternalLSAs.
...


_= ______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/ma= ilman/listinfo/ospf

= --Apple-Mail-5-574311613-- From asmirnov@cisco.com Tue Apr 28 03:48:45 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0E6983A6CA1 for ; Tue, 28 Apr 2009 03:48:45 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.107 X-Spam-Level: X-Spam-Status: No, score=-0.107 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, FRT_ESTABLISH2=2.492] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Ox3WsFsXILU0 for ; Tue, 28 Apr 2009 03:48:43 -0700 (PDT) Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id F31003A6C99 for ; Tue, 28 Apr 2009 03:48:42 -0700 (PDT) X-TACSUNS: Virus Scanned Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3SAo0Ke009181; Tue, 28 Apr 2009 12:50:01 +0200 (CEST) Received: from [10.55.140.89] (ams-asmirnov-8718.cisco.com [10.55.140.89]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3SAo0U9012571; Tue, 28 Apr 2009 12:50:00 +0200 (CEST) Message-ID: <49F6DF5D.80409@cisco.com> Date: Tue, 28 Apr 2009 12:50:05 +0200 From: Anton Smirnov Organization: Cisco Systems, Inc. User-Agent: Thunderbird 2.0.0.19 (X11/20081227) MIME-Version: 1.0 To: Dave Katz References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> In-Reply-To: <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: ospf@ietf.org Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 28 Apr 2009 10:48:45 -0000 Hi Dave, I agree with you in most (but not all) statements. > Bandwidth and CPU are practically free Unless you consider MANETs where bandwidth is not free and CPU resources may also be relatively limited. Originally idea of checksumming was born in context of MANETs. And in that environment exchange of database checksums may actually work well. The first Mitchell's post was about network scenario NOT related to MANETs but questioned how well that idea will work. So my post was more to point out that no, idea conceived with MANET in mind will not work in non-MANET environment. At least not until the idea is properly generalized. > If you want to improve adjacency bringup in a non-compatible way, it > seems like a more effective mechanism would be simply to fix the > brain-dead DBD protocol so that it is not stop-and-wait Agree. > if you can > pump a few hundred PPS of DBDs across the wire without stopping, you can > sync up very quickly in terms of wall-clock time without getting into > the inevitable problems caused by trying to glom LSAs together and > expecting them to sit still in the process. It seems a lot cheaper and > more straightforward. That's good only in case of homogeneous network of powerful routers connected by high-bandwidth links. It is likely to give a lot of troubles in star topologies where one router has many neighbors. Fast DBD flooding from spokes will just swamp the hub router. And I have mentioned it is not going to work in case of MANETs. Of course, transport protocols have solved all these issues long time ago. But we don't want to turn DBD into window-based protocol with slow start, congestion avoidance and all the stuff :-) Thanks, Anton Dave Katz wrote: > On first principles, it seems as though the only reason to do something > like this is to reduce the amount of time it takes to bring up a new > link, and even that is probably only interesting if the network is at > least slightly partitioned without it. Keep in mind that such a scheme > is only useful if the database is already close to synchronized, so it > doesn't help when a router first comes up. Given all that, it's not > entirely clear that doing this is *that* interesting, but no matter. > > The problem with any of these checksum schemes is that the benefits are > compromised by the LSA refresh mechanism, as has been noted. With > 50,000 LSAs, my back-of-the-laptop calculations are that an LSA is > refreshed on the average of once every 36 msec, or roughly 30 per > second. Scalable implementations go out of their way to smooth > refreshes when they are originating a large number of LSAs, and if there > are a large number of machines originating a small number of LSAs, these > will (hopefully) tend to be smoothed out as well, in a way that does not > correlate with the ordering mechanism proposed. > > The efficiency of such schemes improves as the number of LSAs per > checksum block goes up, but so does the probability of a checksum > failure, in which case you're actually worse off than if you'd just send > the darned things (since you have to anyhow, unless you start doing > scary bifurcations of the space, all of which takes more time, and all > of which causes even more LSAs to be stale.) > > If that 50K LSA network was represented as 30 checksum blocks, and it > takes a second for the basic handshaking to take place with a neighbor, > the odds are decent that *all* of them will fail, in which case it's a > waste of time. Keep in mind that if you want to avoid bad checksums, > you need to make the block size small, and thus the number of blocks > large, and then you end up back where you started from unless the > checksum-exchange protocol is reasonably windowed as well. > > If you want to improve adjacency bringup in a non-compatible way, it > seems like a more effective mechanism would be simply to fix the > brain-dead DBD protocol so that it is not stop-and-wait; if you can > pump a few hundred PPS of DBDs across the wire without stopping, you can > sync up very quickly in terms of wall-clock time without getting into > the inevitable problems caused by trying to glom LSAs together and > expecting them to sit still in the process. It seems a lot cheaper and > more straightforward. > > Bandwidth and CPU are practically free (particularly for something done > relatively rarely) so trying to do heroic things to optimize such > situations seems like a waste of effort. > > Maybe I'm missing something fundamental here? > > --Dave > > > On Apr 27, 2009, at 9:47 AM, Mitchell Erblich wrote: > >> Group, >> >> First my thoughts are currently in the inception phase >> and COULD apply to both wired and wireless >> environments and I applogize for my verboseness. >> >> My initial assumptions are that a OSPF area will be >> majority built using the common Ethernet (BMA) environment >> and once a OSPF area is formed major LSA changes are rare. >> >> Second is that the LSDB is comprised of say 50k LSAs >> which are mostly/90% ExternalLSAs. Thus, Anton, I do not think >> your numbers are correct. If we had routers forming adjs >> via standard LSDB exchange each 100k LSAs sum exchange with >> DR/BDR and using checksums. Then no ExternalLSA hdrs would >> need to be exchanged with the BDR, thus removing almost >> 50% from database exchange. FYI: We got them from the DR. >> >> My below inquiry stated a checksum per LSA type and >> the ability to generate subgroups of checksums per >> LSA type. Yes, it also stated a full LSDB checksum. >> >> Since the flooding of LSAs is mostly the same, even a few >> LSAs changing within the ExternalLSa type COULD be covered >> by specifying that the A-Y range did not change and the >> Z range has new hdrs/LSAs and updated normally. >> >> If the later/changes was false, during convergence, routing anomalies >> would be a major concern and data would be routed improperly >> or just lost. With OLD standardized TCP congestion avoidance >> algorithms, a single lost packet assumes congestion and >> assuming 100% bandwidth usage drops TCP to 50% with >> multiplicative decrease. Then a huge number of round trips >> before additive increase gets back to approx 100% bandwidth. >> Even changes in routing paths can generate out-of-orderness with >> the same result. Different hosts/end systems can have different >> Congestion Avoidance algorithms. >> >> Thus, if convergences are frequent then the utilization >> of each of the TCP connection paths would have frequently >> poor bandwidth usage. So the assumption is that once a >> OSPF area is formed, it tends to be UNCHANGING. >> >> >> I mentally prefer a predefined range such that routers >> can pre-compute the checksums before even the >> initial LSDB exchange. To predetermine where the >> breaks should exist is currently unknown. >> >> I see three possible major wins where checksums COULD >> be used. >> The first is flooding from the DR to DRothers >> to update the LSDB so the amount of flooding could >> be minimized, with the same same effect of DNA LSAs. >> Effectively a group LSA refresh. >> The second is when a DRother initially forms an adjacency >> with the BDR and accepting that the ExternalLSAs would >> be the majority of LSAs and the belief that the DRother >> would already have the ExternalLSAs in its LSDB. >> The third is when the DRother wants to just verify that >> no LSAs were lost due to either congestion, temp >> drop of the adj, etc and verfify that no-one has either >> given up rexmits or is still doing xmits. And to pick up >> a single dropped LSA without a full Database Exch. >> >> Getting back to the pre-computation would prevent a >> OSPF speaking router from doing the calcs at adj >> formation time. Even doing the calc at adj formation >> should remove say 90+% of LSAs from needing to >> be recomputed on each LSDB verify sync. >> >> However, to decrease the need to do the calcs at >> overhead, first checking whether LSA type counts >> match could be done and only if they match, then >> do the necessary computation. The type, count, >> and checksum, should generate some form of >> uniqueness. Some field within the begining and >> ending LSAs might add some more verifcation. >> >> So, what am I missing here? and why hasn't anybody >> done this? >> >> Mitchell Erblich >> =================== >> >> >> On Apr 27, 2009, at 12:57 AM, Emmanuel Baccelli wrote: >> >>> Hi Anton, >>> >>> yes, a similar scheme was studied a few years back, in this paper: >>> OSPF Database Exchange and Reliable Synchronization in Mobile Ad Hoc >>> Networks >>> >>> Such a mechanism could indeed lead to interesting performance >>> enhancements, I agree with Richard. >>> >>> cheers, >>> >>> Emmanuel >>> >>> >>> >>> On Mon, Apr 27, 2009 at 2:22 AM, Richard Ogier >> > wrote: >>> >>> Anton, >>> >>> I like your idea of sending a checksum for a range of LSAs, to >>> increase the probability that the LSAs in a given range are in >>> sync between the two neighbors. (This helps even if the >>> age field is ignored when selecting the ranges.) >>> >>> For example, suppose there are 100 LSAs and each LSA has a 2% >>> chance of not being in sync. Then the probability that all >>> LSAs are in sync is .98^100 = .1326, too low to result in >>> a significant overhead reduction. >>> Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. >>> The probability that all LSAs in a given range are in sync is then >>> .98^10 = .8171, resulting in a big overhead reduction on average. >>> >>> The selected range for each LSA type can depend on how frequently >>> new LSAs are originated for that type, with a larger range chosen >>> if LSAs are originated less frequently (so that the probability >>> of being out of sync is lower). >>> >>> This idea can be combined with the database exchange optimization >>> of RFC 5243 to further reduce overhead. I.e., if the checksums >>> do not match, then RFC 5243 can be used to reduce the number of >>> LSA headers exchanged in DD packets by up to 50%. >>> Also, if the checksums do match for a given range of LSAs, then the >>> router receiving the checksum need not send the same checksum for >>> the same range of LSAs. >>> >>> Richard >>> >>> >>> Anton Smirnov wrote: >>> >>> Mitchell, >>> LSDBs may be very close but they are seldom EXACTLY the >>> same. The >>> thing is that change in adjacency triggers LSA [re]build, so >>> depending >>> on timing router and/or network LSAs may change by the time >>> adjacency >>> restores. >>> Also, in moderately sized modern OSPF network there is constant >>> background noise of LSAs being refreshed. So again, chances that >>> refreshed LSA arrived to one side but not another while >>> adjacency is >>> down are relatively high. >>> These considerations make precomputed checksums over large >>> number of >>> LSAs rather bad for the purpose. >>> If you want to explore venue of checksumming database >>> exchange then I >>> recommend to look for more flexible approach giving routers >>> possibility >>> to choose range of LSAs to compute checksum over. For >>> example, if both >>> routers support the checksum exchange then router sending DBD >>> could >>> specify the first, the last LSA and value of some kind of >>> hash of LSA >>> headers between them (LSAs in LSDB can be viewed as ordered >>> sequence as >>> per RFC 5243). In this way router preparing DBD packet may >>> place start >>> and end points of checksumming to avoid including into the >>> range LSAs >>> with ages around 0 and MAXAGE (because these are LSAs which >>> are less >>> likely to match between routers). So database exchange will >>> consist of >>> series of single LSA headers (the 'old' way) and large LSA >>> sequences. >>> >>> Anton >>> >>> >>> >>> Mitchell Erblich wrote: >>> >>> >>> Group, >>> >>> General thoughts... >>> >>> After a conversation about OSPF's standard database >>> exchange and >>> how inefficient it >>> is when two routers have large LSDBs and them >>> almost being the >>> same, we concluded >>> that the use of Checksums might significantly >>> decrease the >>> amount of overhead and >>> achieve faster convergence. >>> >>> Moy first (I think) suggested the LSDB checksum >>> support to >>> identify identical >>> LSDBs. Sometime later the MIB supported the >>> ospfExternLsaChksumSum, >>> ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. >>> >>> Thus, why not support the ability to have OSPF >>> routers first >>> exchange >>> checksums with LSA counts. If the LSDBs have >>> more than XYZ >>> entries >>> in their LSDBs, then the counts which are above >>> specific >>> thresholds then >>> support the checksum on each X count of LSA >>> type. Only the >>> groups >>> with non-matching Checksums would then go thru >>> database >>> exchange. >>> >>> Thus, before database exchange, support some >>> form of triple >>> that >>> as routers have almost the same LSDBs, the >>> checksum exchange >>> would remove unnecessary LSA header exchange. >>> >>> This, IMO, would also be a big win if say a >>> DRother was >>> adjacent, lost >>> adjacency, and re-exstablished it a short time >>> later while >>> keeping a >>> past snapshot. It could remove its own >>> originated LSAs and >>> compare >>> checksums.. >>> >>> In this scenario, within BMA areas, I think >>> only DR / >>> DRother and BDR/ DRother >>> need to be making this Checksum exchange.. >>> >>> Wouldn't the exchange of this data benefit each >>> router >>> receiving the >>> information and allow both routers to know the >>> number of >>> LSAs to >>> be possibly exchanged? >>> Yes, a new capability would probably >>> have to be added. >>> >>> >>> So, the first question is whether the checksums >>> are unique >>> enough with a >>> count and type to determine whether that >>> section of the LSDB >>> matches >>> between two routers? >>> >>> And then what savings would normally exist in >>> this hybrid >>> database exchange >>> and is it worth the effort? >>> >>> Mitchell Erblich >>> >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf > > > ------------------------------------------------------------------------ > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf From dkatz@juniper.net Tue Apr 28 17:03:46 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0E8843A6781 for ; Tue, 28 Apr 2009 17:03:46 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -5.231 X-Spam-Level: X-Spam-Status: No, score=-5.231 tagged_above=-999 required=5 tests=[AWL=-1.124, BAYES_00=-2.599, FRT_ESTABLISH2=2.492, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 3raPsY5A0N6a for ; Tue, 28 Apr 2009 17:03:44 -0700 (PDT) Received: from chip3og59.obsmtp.com (chip3og59.obsmtp.com [64.18.14.183]) by core3.amsl.com (Postfix) with ESMTP id 4DC853A6DB5 for ; Tue, 28 Apr 2009 17:01:02 -0700 (PDT) Received: from source ([66.129.224.36]) (using TLSv1) by chip3ob59.postini.com ([64.18.6.12]) with SMTP ID DSNKSfeZD5Yr34vjliPNCC6LrIAdtOHIbQ1E@postini.com; Tue, 28 Apr 2009 17:02:45 PDT Received: from p-emfe01-sac.jnpr.net (66.129.254.72) by P-EMHUB01-HQ.jnpr.net (172.24.192.35) with Microsoft SMTP Server id 8.1.340.0; Tue, 28 Apr 2009 16:50:37 -0700 Received: from p-emlb02-sac.jnpr.net ([66.129.254.47]) by p-emfe01-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Tue, 28 Apr 2009 16:50:37 -0700 Received: from emailsmtp55.jnpr.net ([172.24.18.132]) by p-emlb02-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Tue, 28 Apr 2009 16:50:37 -0700 Received: from merlot.juniper.net ([172.17.27.10]) by emailsmtp55.jnpr.net with Microsoft SMTPSVC(6.0.3790.1830); Tue, 28 Apr 2009 16:50:36 -0700 Received: from sa-nc-spg-52.static.jnpr.net (sa-nc-spg-52.static.jnpr.net [172.23.1.52]) by merlot.juniper.net (8.11.3/8.11.3) with ESMTP id n3SNoZ044676; Tue, 28 Apr 2009 16:50:36 -0700 (PDT) (envelope-from dkatz@juniper.net) Message-ID: From: Dave Katz To: Anton Smirnov In-Reply-To: <49F6DF5D.80409@cisco.com> Content-Type: text/plain; charset="US-ASCII"; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit MIME-Version: 1.0 (Apple Message framework v930.3) Date: Tue, 28 Apr 2009 18:50:34 -0500 References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> <49F6DF5D.80409@cisco.com> X-Mailer: Apple Mail (2.930.3) X-OriginalArrivalTime: 28 Apr 2009 23:50:36.0611 (UTC) FILETIME=[20976930:01C9C85C] Cc: ospf@ietf.org Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 29 Apr 2009 00:03:46 -0000 On Apr 28, 2009, at 5:50 AM, Anton Smirnov wrote: > Hi Dave, > I agree with you in most (but not all) statements. > >> Bandwidth and CPU are practically free > > Unless you consider MANETs where bandwidth is not free and CPU > resources may also be relatively limited. > Originally idea of checksumming was born in context of MANETs. And > in > that environment exchange of database checksums may actually work > well. > The first Mitchell's post was about network scenario NOT related to > MANETs but questioned how well that idea will work. So my post was > more > to point out that no, idea conceived with MANET in mind will not > work in > non-MANET environment. At least not until the idea is properly > generalized. I guess I'd be impressed with a network that had 50,000 LSAs but very limited bandwidth and CPU. I have a feeling that such a network would have bigger problems than the time it takes to bring up an additional link. > > >> If you want to improve adjacency bringup in a non-compatible way, it >> seems like a more effective mechanism would be simply to fix the >> brain-dead DBD protocol so that it is not stop-and-wait > > Agree. > > >> if you can >> pump a few hundred PPS of DBDs across the wire without stopping, >> you can >> sync up very quickly in terms of wall-clock time without getting into >> the inevitable problems caused by trying to glom LSAs together and >> expecting them to sit still in the process. It seems a lot cheaper >> and >> more straightforward. > > That's good only in case of homogeneous network of powerful routers > connected by high-bandwidth links. > It is likely to give a lot of troubles in star topologies where one > router has many neighbors. Fast DBD flooding from spokes will just > swamp > the hub router. > And I have mentioned it is not going to work in case of MANETs. But this is no different than the situation with LSUpd packets today, and needs to be solved in the same way (hack.) We used to deal with a much worse situation ten or fifteen years ago, which were sizable full meshes. If it were easy, we wouldn't make any money doing it. ;-) --Dave > > > Of course, transport protocols have solved all these issues long > time > ago. But we don't want to turn DBD into window-based protocol with > slow > start, congestion avoidance and all the stuff :-) > > Thanks, > > Anton > > > > Dave Katz wrote: >> On first principles, it seems as though the only reason to do >> something >> like this is to reduce the amount of time it takes to bring up a new >> link, and even that is probably only interesting if the network is at >> least slightly partitioned without it. Keep in mind that such a >> scheme >> is only useful if the database is already close to synchronized, so >> it >> doesn't help when a router first comes up. Given all that, it's not >> entirely clear that doing this is *that* interesting, but no matter. >> >> The problem with any of these checksum schemes is that the benefits >> are >> compromised by the LSA refresh mechanism, as has been noted. With >> 50,000 LSAs, my back-of-the-laptop calculations are that an LSA is >> refreshed on the average of once every 36 msec, or roughly 30 per >> second. Scalable implementations go out of their way to smooth >> refreshes when they are originating a large number of LSAs, and if >> there >> are a large number of machines originating a small number of LSAs, >> these >> will (hopefully) tend to be smoothed out as well, in a way that >> does not >> correlate with the ordering mechanism proposed. >> >> The efficiency of such schemes improves as the number of LSAs per >> checksum block goes up, but so does the probability of a checksum >> failure, in which case you're actually worse off than if you'd just >> send >> the darned things (since you have to anyhow, unless you start doing >> scary bifurcations of the space, all of which takes more time, and >> all >> of which causes even more LSAs to be stale.) >> >> If that 50K LSA network was represented as 30 checksum blocks, and it >> takes a second for the basic handshaking to take place with a >> neighbor, >> the odds are decent that *all* of them will fail, in which case >> it's a >> waste of time. Keep in mind that if you want to avoid bad checksums, >> you need to make the block size small, and thus the number of blocks >> large, and then you end up back where you started from unless the >> checksum-exchange protocol is reasonably windowed as well. >> >> If you want to improve adjacency bringup in a non-compatible way, it >> seems like a more effective mechanism would be simply to fix the >> brain-dead DBD protocol so that it is not stop-and-wait; if you can >> pump a few hundred PPS of DBDs across the wire without stopping, >> you can >> sync up very quickly in terms of wall-clock time without getting into >> the inevitable problems caused by trying to glom LSAs together and >> expecting them to sit still in the process. It seems a lot cheaper >> and >> more straightforward. >> >> Bandwidth and CPU are practically free (particularly for something >> done >> relatively rarely) so trying to do heroic things to optimize such >> situations seems like a waste of effort. >> >> Maybe I'm missing something fundamental here? >> >> --Dave >> >> >> On Apr 27, 2009, at 9:47 AM, Mitchell Erblich wrote: >> >>> Group, >>> >>> First my thoughts are currently in the inception phase >>> and COULD apply to both wired and wireless >>> environments and I applogize for my verboseness. >>> >>> My initial assumptions are that a OSPF area will be >>> majority built using the common Ethernet (BMA) environment >>> and once a OSPF area is formed major LSA changes are rare. >>> >>> Second is that the LSDB is comprised of say 50k LSAs >>> which are mostly/90% ExternalLSAs. Thus, Anton, I do not think >>> your numbers are correct. If we had routers forming adjs >>> via standard LSDB exchange each 100k LSAs sum exchange with >>> DR/BDR and using checksums. Then no ExternalLSA hdrs would >>> need to be exchanged with the BDR, thus removing almost >>> 50% from database exchange. FYI: We got them from the DR. >>> >>> My below inquiry stated a checksum per LSA type and >>> the ability to generate subgroups of checksums per >>> LSA type. Yes, it also stated a full LSDB checksum. >>> >>> Since the flooding of LSAs is mostly the same, even a few >>> LSAs changing within the ExternalLSa type COULD be covered >>> by specifying that the A-Y range did not change and the >>> Z range has new hdrs/LSAs and updated normally. >>> >>> If the later/changes was false, during convergence, routing >>> anomalies >>> would be a major concern and data would be routed improperly >>> or just lost. With OLD standardized TCP congestion avoidance >>> algorithms, a single lost packet assumes congestion and >>> assuming 100% bandwidth usage drops TCP to 50% with >>> multiplicative decrease. Then a huge number of round trips >>> before additive increase gets back to approx 100% bandwidth. >>> Even changes in routing paths can generate out-of-orderness with >>> the same result. Different hosts/end systems can have different >>> Congestion Avoidance algorithms. >>> >>> Thus, if convergences are frequent then the utilization >>> of each of the TCP connection paths would have frequently >>> poor bandwidth usage. So the assumption is that once a >>> OSPF area is formed, it tends to be UNCHANGING. >>> >>> >>> I mentally prefer a predefined range such that routers >>> can pre-compute the checksums before even the >>> initial LSDB exchange. To predetermine where the >>> breaks should exist is currently unknown. >>> >>> I see three possible major wins where checksums COULD >>> be used. >>> The first is flooding from the DR to DRothers >>> to update the LSDB so the amount of flooding could >>> be minimized, with the same same effect of DNA LSAs. >>> Effectively a group LSA refresh. >>> The second is when a DRother initially forms an adjacency >>> with the BDR and accepting that the ExternalLSAs would >>> be the majority of LSAs and the belief that the DRother >>> would already have the ExternalLSAs in its LSDB. >>> The third is when the DRother wants to just verify that >>> no LSAs were lost due to either congestion, temp >>> drop of the adj, etc and verfify that no-one has either >>> given up rexmits or is still doing xmits. And to pick up >>> a single dropped LSA without a full Database Exch. >>> >>> Getting back to the pre-computation would prevent a >>> OSPF speaking router from doing the calcs at adj >>> formation time. Even doing the calc at adj formation >>> should remove say 90+% of LSAs from needing to >>> be recomputed on each LSDB verify sync. >>> >>> However, to decrease the need to do the calcs at >>> overhead, first checking whether LSA type counts >>> match could be done and only if they match, then >>> do the necessary computation. The type, count, >>> and checksum, should generate some form of >>> uniqueness. Some field within the begining and >>> ending LSAs might add some more verifcation. >>> >>> So, what am I missing here? and why hasn't anybody >>> done this? >>> >>> Mitchell Erblich >>> =================== >>> >>> >>> On Apr 27, 2009, at 12:57 AM, Emmanuel Baccelli wrote: >>> >>>> Hi Anton, >>>> >>>> yes, a similar scheme was studied a few years back, in this paper: >>>> OSPF Database Exchange and Reliable Synchronization in Mobile Ad >>>> Hoc >>>> Networks >>> > >>>> >>>> Such a mechanism could indeed lead to interesting performance >>>> enhancements, I agree with Richard. >>>> >>>> cheers, >>>> >>>> Emmanuel >>>> >>>> >>>> >>>> On Mon, Apr 27, 2009 at 2:22 AM, Richard Ogier >>> > wrote: >>>> >>>> Anton, >>>> >>>> I like your idea of sending a checksum for a range of LSAs, to >>>> increase the probability that the LSAs in a given range are in >>>> sync between the two neighbors. (This helps even if the >>>> age field is ignored when selecting the ranges.) >>>> >>>> For example, suppose there are 100 LSAs and each LSA has a 2% >>>> chance of not being in sync. Then the probability that all >>>> LSAs are in sync is .98^100 = .1326, too low to result in >>>> a significant overhead reduction. >>>> Now suppose the LSAs are divided into 10 ranges, each with 10 >>>> LSAs. >>>> The probability that all LSAs in a given range are in sync is >>>> then >>>> .98^10 = .8171, resulting in a big overhead reduction on >>>> average. >>>> >>>> The selected range for each LSA type can depend on how >>>> frequently >>>> new LSAs are originated for that type, with a larger range >>>> chosen >>>> if LSAs are originated less frequently (so that the probability >>>> of being out of sync is lower). >>>> >>>> This idea can be combined with the database exchange >>>> optimization >>>> of RFC 5243 to further reduce overhead. I.e., if the checksums >>>> do not match, then RFC 5243 can be used to reduce the number of >>>> LSA headers exchanged in DD packets by up to 50%. >>>> Also, if the checksums do match for a given range of LSAs, >>>> then the >>>> router receiving the checksum need not send the same checksum >>>> for >>>> the same range of LSAs. >>>> >>>> Richard >>>> >>>> >>>> Anton Smirnov wrote: >>>> >>>> Mitchell, >>>> LSDBs may be very close but they are seldom EXACTLY the >>>> same. The >>>> thing is that change in adjacency triggers LSA [re]build, so >>>> depending >>>> on timing router and/or network LSAs may change by the time >>>> adjacency >>>> restores. >>>> Also, in moderately sized modern OSPF network there is >>>> constant >>>> background noise of LSAs being refreshed. So again, >>>> chances that >>>> refreshed LSA arrived to one side but not another while >>>> adjacency is >>>> down are relatively high. >>>> These considerations make precomputed checksums over large >>>> number of >>>> LSAs rather bad for the purpose. >>>> If you want to explore venue of checksumming database >>>> exchange then I >>>> recommend to look for more flexible approach giving routers >>>> possibility >>>> to choose range of LSAs to compute checksum over. For >>>> example, if both >>>> routers support the checksum exchange then router sending >>>> DBD >>>> could >>>> specify the first, the last LSA and value of some kind of >>>> hash of LSA >>>> headers between them (LSAs in LSDB can be viewed as ordered >>>> sequence as >>>> per RFC 5243). In this way router preparing DBD packet may >>>> place start >>>> and end points of checksumming to avoid including into the >>>> range LSAs >>>> with ages around 0 and MAXAGE (because these are LSAs which >>>> are less >>>> likely to match between routers). So database exchange will >>>> consist of >>>> series of single LSA headers (the 'old' way) and large LSA >>>> sequences. >>>> >>>> Anton >>>> >>>> >>>> >>>> Mitchell Erblich wrote: >>>> >>>> >>>> Group, >>>> >>>> General thoughts... >>>> >>>> After a conversation about OSPF's standard >>>> database >>>> exchange and >>>> how inefficient it >>>> is when two routers have large LSDBs and them >>>> almost being the >>>> same, we concluded >>>> that the use of Checksums might significantly >>>> decrease the >>>> amount of overhead and >>>> achieve faster convergence. >>>> >>>> Moy first (I think) suggested the LSDB >>>> checksum >>>> support to >>>> identify identical >>>> LSDBs. Sometime later the MIB supported the >>>> ospfExternLsaChksumSum, >>>> ospfAreaLsaChksumSum, and ospfLsdbChecksum >>>> fields. >>>> >>>> Thus, why not support the ability to have OSPF >>>> routers first >>>> exchange >>>> checksums with LSA counts. If the LSDBs have >>>> more than XYZ >>>> entries >>>> in their LSDBs, then the counts which are >>>> above >>>> specific >>>> thresholds then >>>> support the checksum on each X count of LSA >>>> type. Only the >>>> groups >>>> with non-matching Checksums would then go thru >>>> database >>>> exchange. >>>> >>>> Thus, before database exchange, support some >>>> form of triple >>>> that >>>> as routers have almost the same LSDBs, the >>>> checksum exchange >>>> would remove unnecessary LSA header exchange. >>>> >>>> This, IMO, would also be a big win if say a >>>> DRother was >>>> adjacent, lost >>>> adjacency, and re-exstablished it a short time >>>> later while >>>> keeping a >>>> past snapshot. It could remove its own >>>> originated LSAs and >>>> compare >>>> checksums.. >>>> >>>> In this scenario, within BMA areas, I think >>>> only DR / >>>> DRother and BDR/ DRother >>>> need to be making this Checksum exchange.. >>>> >>>> Wouldn't the exchange of this data benefit >>>> each >>>> router >>>> receiving the >>>> information and allow both routers to know the >>>> number of >>>> LSAs to >>>> be possibly exchanged? >>>> Yes, a new capability would probably >>>> have to be added. >>>> >>>> >>>> So, the first question is whether the >>>> checksums >>>> are unique >>>> enough with a >>>> count and type to determine whether that >>>> section of the LSDB >>>> matches >>>> between two routers? >>>> >>>> And then what savings would normally exist in >>>> this hybrid >>>> database exchange >>>> and is it worth the effort? >>>> >>>> Mitchell Erblich >>>> >>>> >>>> >>>> _______________________________________________ >>>> OSPF mailing list >>>> OSPF@ietf.org >>>> https://www.ietf.org/mailman/listinfo/ospf >>>> >>>> >>>> _______________________________________________ >>>> OSPF mailing list >>>> OSPF@ietf.org >>>> https://www.ietf.org/mailman/listinfo/ospf >>>> >>>> >>>> >>>> _______________________________________________ >>>> OSPF mailing list >>>> OSPF@ietf.org >>>> https://www.ietf.org/mailman/listinfo/ospf >>>> >>>> >>>> _______________________________________________ >>>> OSPF mailing list >>>> OSPF@ietf.org >>>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >> >> >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf > From erblichs@earthlink.net Wed Apr 29 04:23:20 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 522CE3A6D0D for ; Wed, 29 Apr 2009 04:23:20 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.555 X-Spam-Level: X-Spam-Status: No, score=-0.555 tagged_above=-999 required=5 tests=[AWL=-1.049, BAYES_00=-2.599, FRT_ESTABLISH2=2.492, HTML_MESSAGE=0.001, J_CHICKENPOX_42=0.6] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id GFse8VQozPmV for ; Wed, 29 Apr 2009 04:23:17 -0700 (PDT) Received: from elasmtp-curtail.atl.sa.earthlink.net (elasmtp-curtail.atl.sa.earthlink.net [209.86.89.64]) by core3.amsl.com (Postfix) with ESMTP id C247B3A6CAD for ; Wed, 29 Apr 2009 04:23:16 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net; b=htFlNBQI6NZn5zd434l+NaTXneSOEV0JA3UxZBuCK/f9BswYE3x0oKMRT1UaIdev; h=Received:Cc:Message-Id:From:To:In-Reply-To:Content-Type:Mime-Version:Subject:Date:References:X-Mailer:X-ELNK-Trace:X-Originating-IP; Received: from [98.234.127.54] (helo=[10.0.1.9]) by elasmtp-curtail.atl.sa.earthlink.net with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.67) (envelope-from ) id 1Lz7u0-00033M-AX; Wed, 29 Apr 2009 07:24:36 -0400 Message-Id: From: Mitchell Erblich To: Dave Katz In-Reply-To: <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> Content-Type: multipart/alternative; boundary=Apple-Mail-240-704069833 Mime-Version: 1.0 (Apple Message framework v930.3) Date: Wed, 29 Apr 2009 04:24:34 -0700 References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> X-Mailer: Apple Mail (2.930.3) X-ELNK-Trace: 074f60c55517ea841aa676d7e74259b7b3291a7d08dfec793d980401994a495266e49f0c3eb82ed9350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 98.234.127.54 Cc: ospf@ietf.org, Emmanuel Baccelli Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 29 Apr 2009 11:23:20 -0000 --Apple-Mail-240-704069833 Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit Group, Sorry, verbose.. Inline.. Mitchell Erblich =============== On Apr 27, 2009, at 4:20 PM, Dave Katz wrote: > On first principles, it seems as though the only reason to do > something like this is to reduce the amount of time it takes to > bring up a new link, and even that is probably only interesting if > the network is at least slightly partitioned without it. Keep in > mind that such a scheme is only useful if the database is already > close to synchronized, so it doesn't help when a router first comes > up. Given all that, it's not entirely clear that doing this is > *that* interesting, but no matter. Actually, the first thought was how to verify that two LSDBs are identical at ANY time after Database Exchange. And if not first what are the differences wrt LSA type and count. And have the router periodicly report / summarize any seen diffs. Then how to resynch without waiting for the LSA update or renewing (tear down and reforming the adj) the adj However, if you can then resync, you should be able to do refreshes from the DR and BDR (common Eth). However, the first interaction between two OSPF speakers is with Database Exchange and decreasing the workload on the 2nd adj due to common LSAs should allow the formation of 2x simulateneous adj formations than standard. > > The problem with any of these checksum schemes is that the benefits > are compromised by the LSA refresh mechanism, as has been noted. > With 50,000 LSAs, my back-of-the-laptop calculations are that an LSA > is refreshed on the average of once every 36 msec, or roughly 30 per > second. Scalable implementations go out of their way to smooth > refreshes when they are originating a large number of LSAs, and if > there are a large number of machines originating a small number of > LSAs, these will (hopefully) tend to be smoothed out as well, in a > way that does not correlate with the ordering mechanism proposed. First, 50k was just a number picked that would identify a common large number of LSAs of a specific type(Externals) (also the major of LSAs in the LSDB) that should all match and not have any new DRother add to this LSA type pool. The simplified assumption is that the updates are just age refreshes. I am not sure that this/checksums would be a better approach then say DNA LSAs. The chksum approach fits better IMO in this case when a router stops rexmits of the LSA and has taken down the adj. The DRother either would have to renew the adj or implement a search for the lost LSA. In the common case of Eth, with your flooding example, in my limited experience, the OSPF routers tend to initially come up within the area at the same time and thus iniitally the LSAs are mostly flooded within the same quarter hour, if not within the first 5 minutes. Thus, peaks of LSAs per second would be on the order of 200 LSAs per sec. Since the chksum is even simpler than a LSA hdr, I would expect more than a 100 to 10kx LSA processing of just LSA age refreshes should be able to be accomplished with the same level of work. Somewhere in these numbers is a meltdown by a router with standardized OSPF, where I question whether the chksum approach would have the same failure behaviour. > > The efficiency of such schemes improves as the number of LSAs per > checksum block goes up, but so does the probability of a checksum > failure, in which case you're actually worse off than if you'd just > send the darned things (since you have to anyhow, unless you start > doing scary bifurcations of the space, all of which takes more time, > and all of which causes even more LSAs to be stale.) The assumption is that a checksum is specified for the entire LSDB, then a chksum for all the LSAs of a specified type, then a breakdown of each type. If we have Eth, we have a 1500byte packet. We have no LSA lookups if we can can match each table line. If we add additional values like LSA count, LSA type, etc, to each group checksum, then the same chksums that map to different sets of LSAs should be minimized. Remember, we need a double failure/mapping because not just the group must map identically but the chksum over the entire LSA type. FYI, yes, I know about the 64k byte offset of IP, but I don't want reassembly overhead so a limit of 1.5k. What the additional groups get us, is the ability to locate request entire LSA groups without needing all of the LSAs. The smaller the group, then the smaller number of extra LSAs that we really don't need but with more table lookups. YES, the complication is the CORRECT number of LSAs needed in a block to minimize a duplicate mapping, where all items in the table line match, but the set of LSAs don't match within the group. > > If that 50K LSA network was represented as 30 checksum blocks, and > it takes a second for the basic handshaking to take place with a > neighbor, the odds are decent that *all* of them will fail, in which > case it's a waste of time. Keep in mind that if you want to avoid > bad checksums, you need to make the block size small, and thus the > number of blocks large, and then you end up back where you started > from unless the checksum-exchange protocol is reasonably windowed as > well. Actually, how long does it take to process 1 1500 byte packet and do say 100 table lookups. The key here is that not the entire LSDB summary needs to be exchanged per packet. We could say that every 15 secs the next set of groups is xmit'ed and every x minutes you cover all the LSAs via the groupings. So, I would expect less than 10ms to processs each packet. > > If you want to improve adjacency bringup in a non-compatible way, it > seems like a more effective mechanism would be simply to fix the > brain-dead DBD protocol so that it is not stop-and-wait; if you can > pump a few hundred PPS of DBDs across the wire without stopping, you > can sync up very quickly in terms of wall-clock time without getting > into the inevitable problems caused by trying to glom LSAs together > and expecting them to sit still in the process. It seems a lot > cheaper and more straightforward. The wireless win is the minimum number of bytes that COULD then be sent over the connection and the ability to have dropouts and verifications that the LSDB is in sync and if not, why not. With a protocol like OSPF that has aged, I would think that simpler approaches to scale them in a transparent way would make more sense. With chksums, I would think that a DRother forming an adj with a BDR after the DR, should be able to do this within just a few 100 packets. If the fastpath allows a common read-mostly table lookup and can remove 90% of LSA hdr exchanges, then the ability for the BDR to support 100+ simultaneous adj formations should sum to your 100s of packets per sec rate. With the ability to not send say 100s packets before a response to a DRother during database exchange prevents the DR and/or BDR to FLOOD/overload the DRother with LSAs. TCP's slow start is a multiplicative increase to identify not just un-used capacity but also buffer capacity. If a flood of 100 packets is jjust sent, and they are fragmented, they really need to be reasembled, etc. This is assuming no tail drop or RED, where a dropped fragment requires the sending of the entire packet/segment. The problem is not all routers would have the same pps capacity and if they did and say 5 DRothers implement your flooding to a single DR, wouldn't that be a problem? With a ACK based protocol, the two communicating end points can add small latencies as needed and prevent one end point to overwhelm the other end point. Lastly, the amount of CPU cycles needed to process 100 pps tends to be greater than xmiting the 100 pps rate. Within an area, the assumption is that the link/media connection is probably at 1ms or less and thus the round-trip-time should not be a concern. If the DRother and DR/or/BDR processes the packet within say 1ms each, then the sum of packets per connection should easily reach your 100s of packets per sec. Yes, with a ACK like protocol, having the the ability to poll or generate just 1 interrupt for x number of packets can not be done with only 1 set of concurrent connections. If you had say 100 simultaneous adj formations, yes you can then implement the above. > > Bandwidth and CPU are practically free (particularly for something > done relatively rarely) so trying to do heroic things to optimize > such situations seems like a waste of effort. The above SHOULD use a fraction of the bandwidth and CPU. The gains of doing so is the ability to verify that all the LSDBs are in synch within the area every sec. The decrease of bandwidth is a decrease in latency when differences exist and the ability to correct them quickly. IF flooding uses 1% of a CPU, we then have 99% to recompute routing changes. With 1Gb and faster Eth media, having short convergent cycles is imperative. With this additional CPU cycles and low cost memory, we have the abiltiy to compute backup routes. And to be able to separate the LSA updates into age updates versus content updates. The later need to be processed as a high priority and moving the former into Chksum verification, should be able to take LSA content changes and reconverge the area within a subsec timeframe. However, alot of end systems (and routers) are or becoming mobile and their bandwidth is in the kilobytes/sec capabilities, have expected dropouts, and we still need to route to them. Thus adding downward scalability shouldn't be a negative. > > Maybe I'm missing something fundamental here? > > --Dave > > > On Apr 27, 2009, at 9:47 AM, Mitchell Erblich wrote: > >> Group, >> >> First my thoughts are currently in the inception phase >> and COULD apply to both wired and wireless >> environments and I applogize for my verboseness. >> >> My initial assumptions are that a OSPF area will be >> majority built using the common Ethernet (BMA) environment >> and once a OSPF area is formed major LSA changes are rare. >> >> Second is that the LSDB is comprised of say 50k LSAs >> which are mostly/90% ExternalLSAs. Thus, Anton, I do not think >> your numbers are correct. If we had routers forming adjs >> via standard LSDB exchange each 100k LSAs sum exchange with >> DR/BDR and using checksums. Then no ExternalLSA hdrs would >> need to be exchanged with the BDR, thus removing almost >> 50% from database exchange. FYI: We got them from the DR. >> >> My below inquiry stated a checksum per LSA type and >> the ability to generate subgroups of checksums per >> LSA type. Yes, it also stated a full LSDB checksum. >> >> Since the flooding of LSAs is mostly the same, even a few >> LSAs changing within the ExternalLSa type COULD be covered >> by specifying that the A-Y range did not change and the >> Z range has new hdrs/LSAs and updated normally. >> >> If the later/changes was false, during convergence, routing >> anomalies >> would be a major concern and data would be routed improperly >> or just lost. With OLD standardized TCP congestion avoidance >> algorithms, a single lost packet assumes congestion and >> assuming 100% bandwidth usage drops TCP to 50% with >> multiplicative decrease. Then a huge number of round trips >> before additive increase gets back to approx 100% bandwidth. >> Even changes in routing paths can generate out-of-orderness with >> the same result. Different hosts/end systems can have different >> Congestion Avoidance algorithms. >> >> Thus, if convergences are frequent then the utilization >> of each of the TCP connection paths would have frequently >> poor bandwidth usage. So the assumption is that once a >> OSPF area is formed, it tends to be UNCHANGING. >> >> >> I mentally prefer a predefined range such that routers >> can pre-compute the checksums before even the >> initial LSDB exchange. To predetermine where the >> breaks should exist is currently unknown. >> >> I see three possible major wins where checksums COULD >> be used. >> The first is flooding from the DR to DRothers >> to update the LSDB so the amount of flooding could >> be minimized, with the same same effect of DNA LSAs. >> Effectively a group LSA refresh. >> The second is when a DRother initially forms an adjacency >> with the BDR and accepting that the ExternalLSAs would >> be the majority of LSAs and the belief that the DRother >> would already have the ExternalLSAs in its LSDB. >> The third is when the DRother wants to just verify that >> no LSAs were lost due to either congestion, temp >> drop of the adj, etc and verfify that no-one has either >> given up rexmits or is still doing xmits. And to pick up >> a single dropped LSA without a full Database Exch. >> >> >> Getting back to the pre-computation would prevent a >> OSPF speaking router from doing the calcs at adj >> formation time. Even doing the calc at adj formation >> should remove say 90+% of LSAs from needing to >> be recomputed on each LSDB verify sync. >> >> However, to decrease the need to do the calcs at >> overhead, first checking whether LSA type counts >> match could be done and only if they match, then >> do the necessary computation. The type, count, >> and checksum, should generate some form of >> uniqueness. Some field within the begining and >> ending LSAs might add some more verifcation. >> >> So, what am I missing here? and why hasn't anybody >> done this? >> >> Mitchell Erblich >> =================== >> >> >> On Apr 27, 2009, at 12:57 AM, Emmanuel Baccelli wrote: >> >>> Hi Anton, >>> >>> yes, a similar scheme was studied a few years back, in this paper: >>> OSPF Database Exchange and Reliable Synchronization in Mobile Ad >>> Hoc Networks >>> >>> Such a mechanism could indeed lead to interesting performance >>> enhancements, I agree with Richard. >>> >>> cheers, >>> >>> Emmanuel >>> >>> >>> >>> On Mon, Apr 27, 2009 at 2:22 AM, Richard Ogier >>> wrote: >>> Anton, >>> >>> I like your idea of sending a checksum for a range of LSAs, to >>> increase the probability that the LSAs in a given range are in >>> sync between the two neighbors. (This helps even if the >>> age field is ignored when selecting the ranges.) >>> >>> For example, suppose there are 100 LSAs and each LSA has a 2% >>> chance of not being in sync. Then the probability that all >>> LSAs are in sync is .98^100 = .1326, too low to result in >>> a significant overhead reduction. >>> Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. >>> The probability that all LSAs in a given range are in sync is then >>> .98^10 = .8171, resulting in a big overhead reduction on average. >>> >>> The selected range for each LSA type can depend on how frequently >>> new LSAs are originated for that type, with a larger range chosen >>> if LSAs are originated less frequently (so that the probability >>> of being out of sync is lower). >>> >>> This idea can be combined with the database exchange optimization >>> of RFC 5243 to further reduce overhead. I.e., if the checksums >>> do not match, then RFC 5243 can be used to reduce the number of >>> LSA headers exchanged in DD packets by up to 50%. >>> Also, if the checksums do match for a given range of LSAs, then the >>> router receiving the checksum need not send the same checksum for >>> the same range of LSAs. >>> >>> Richard >>> >>> >>> Anton Smirnov wrote: >>> >>> Mitchell, >>> LSDBs may be very close but they are seldom EXACTLY the same. The >>> thing is that change in adjacency triggers LSA [re]build, so >>> depending >>> on timing router and/or network LSAs may change by the time >>> adjacency >>> restores. >>> Also, in moderately sized modern OSPF network there is constant >>> background noise of LSAs being refreshed. So again, chances that >>> refreshed LSA arrived to one side but not another while adjacency is >>> down are relatively high. >>> These considerations make precomputed checksums over large number >>> of >>> LSAs rather bad for the purpose. >>> If you want to explore venue of checksumming database exchange >>> then I >>> recommend to look for more flexible approach giving routers >>> possibility >>> to choose range of LSAs to compute checksum over. For example, if >>> both >>> routers support the checksum exchange then router sending DBD could >>> specify the first, the last LSA and value of some kind of hash of >>> LSA >>> headers between them (LSAs in LSDB can be viewed as ordered >>> sequence as >>> per RFC 5243). In this way router preparing DBD packet may place >>> start >>> and end points of checksumming to avoid including into the range >>> LSAs >>> with ages around 0 and MAXAGE (because these are LSAs which are less >>> likely to match between routers). So database exchange will >>> consist of >>> series of single LSA headers (the 'old' way) and large LSA >>> sequences. >>> >>> Anton >>> >>> >>> >>> Mitchell Erblich wrote: >>> >>> Group, >>> >>> General thoughts... >>> >>> After a conversation about OSPF's standard database exchange >>> and >>> how inefficient it >>> is when two routers have large LSDBs and them almost being the >>> same, we concluded >>> that the use of Checksums might significantly decrease the >>> amount of overhead and >>> achieve faster convergence. >>> >>> Moy first (I think) suggested the LSDB checksum support to >>> identify identical >>> LSDBs. Sometime later the MIB supported the >>> ospfExternLsaChksumSum, >>> ospfAreaLsaChksumSum, and ospfLsdbChecksum fields. >>> >>> Thus, why not support the ability to have OSPF routers >>> first >>> exchange >>> checksums with LSA counts. If the LSDBs have more than XYZ >>> entries >>> in their LSDBs, then the counts which are above specific >>> thresholds then >>> support the checksum on each X count of LSA type. Only the >>> groups >>> with non-matching Checksums would then go thru database >>> exchange. >>> >>> Thus, before database exchange, support some form of >>> triple >>> that >>> as routers have almost the same LSDBs, the checksum >>> exchange >>> would remove unnecessary LSA header exchange. >>> >>> This, IMO, would also be a big win if say a DRother was >>> adjacent, lost >>> adjacency, and re-exstablished it a short time later while >>> keeping a >>> past snapshot. It could remove its own originated LSAs and >>> compare >>> checksums.. >>> >>> In this scenario, within BMA areas, I think only DR / >>> DRother and BDR/ DRother >>> need to be making this Checksum exchange.. >>> >>> Wouldn't the exchange of this data benefit each router >>> receiving the >>> information and allow both routers to know the number of >>> LSAs to >>> be possibly exchanged? >>> Yes, a new capability would probably have to >>> be added. >>> >>> >>> So, the first question is whether the checksums are unique >>> enough with a >>> count and type to determine whether that section of the >>> LSDB >>> matches >>> between two routers? >>> >>> And then what savings would normally exist in this hybrid >>> database exchange >>> and is it worth the effort? >>> >>> Mitchell Erblich >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >> >> _______________________________________________ >> OSPF mailing list >> OSPF@ietf.org >> https://www.ietf.org/mailman/listinfo/ospf > --Apple-Mail-240-704069833 Content-Type: text/html; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Sorry, = verbose..

= Inline..

Mitchell Erblich
= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
On Apr = 27, 2009, at 4:20 PM, Dave Katz wrote:

On first principles, it = seems as though the only reason to do something like this is to reduce = the amount of time it takes to bring up a new link, and even that is = probably only interesting if the network is at least slightly = partitioned without it.  Keep in mind that such a scheme is only = useful if the database is already close to synchronized, so it doesn't = help when a router first comes up.  Given all that, it's not = entirely clear that doing this is *that* interesting, but no = matter.

Actually, the first = thought was how to verify that two LSDBs are identical at ANY time after = Database Exchange.
And if not first what are the differences = wrt LSA type and count. And have the router periodicly report / = summarize
any seen diffs.

Then = how to resynch without waiting for the LSA update or renewing (tear down = and reforming the adj) the adj

However, if you = can then resync, you should be able to do refreshes from the DR and BDR = (common Eth).

However, the first interaction = between two OSPF speakers is with Database Exchange and decreasing = the
workload on the 2nd adj due to common LSAs should allow = the formation of 2x simulateneous adj formations
than = standard.



The problem with = any of these checksum schemes is that the benefits are compromised by = the LSA refresh mechanism, as has been noted.  With 50,000 LSAs, my = back-of-the-laptop calculations are that an LSA is refreshed on the = average of once every 36 msec, or roughly 30 per second.  Scalable = implementations go out of their way to smooth refreshes when they are = originating a large number of LSAs, and if there are a large number of = machines originating a small number of LSAs, these will (hopefully) tend = to be smoothed out as well, in a way that does not correlate with the = ordering mechanism = proposed.

First, 50k was just a = number picked that would identify a common large number of LSAs of a = specific type(Externals)
 (also the major of LSAs in the = LSDB) that should all match and not have any new DRother add to = this LSA type pool.

The simplified assumption = is that the updates are just age refreshes.

I am not = sure that this/checksums would be a better approach then say DNA LSAs. = The chksum approach fits better
IMO in this case when a router = stops rexmits of the LSA and has taken down the adj. The DRother either = would
have to renew the adj or implement a search for the lost = LSA.

In the common case of Eth, with your = flooding example, in my limited experience, the OSPF routers tend = to 
initially come up within the area at the same time = and thus iniitally the LSAs are mostly flooded within the
same = quarter hour, if not within the first 5 minutes. Thus, peaks of LSAs per = second would be on the order
of 200 LSAs per = sec.

Since the chksum is even simpler than a = LSA hdr, I would expect more than a 100 to 10kx LSA processing = of
just LSA age refreshes should be able to be accomplished = with the same level of work.

Somewhere in these = numbers is a meltdown by a router with standardized OSPF, where I = question whether
the chksum approach would have the same = failure behaviour.


The = efficiency of such schemes improves as the number of LSAs per checksum = block goes up, but so does the probability of a checksum failure, in = which case you're actually worse off than if you'd just send the darned = things (since you have to anyhow, unless you start doing scary = bifurcations of the space, all of which takes more time, and all of = which causes even more LSAs to be = stale.)

The assumption is that a = checksum is specified for the entire LSDB, then a chksum for all the = LSAs of a specified
type, then a breakdown of each type. If we = have Eth, we have a 1500byte packet. We have no LSA lookups = if
we can can match each table line. If we add additional = values like LSA count, LSA type, etc, to each group
checksum, = then the same chksums that map to different sets of LSAs should be = minimized. Remember, we
need a double failure/mapping because = not just the group must map identically but the chksum over = the
entire LSA type.

FYI, yes, I know = about the 64k byte offset of IP, but I don't want reassembly overhead so = a limit of 1.5k.

What the additional groups get = us, is the ability to locate request entire LSA groups without = needing
all of the LSAs. The smaller the group, then  the = smaller number of extra LSAs that we really don't
need but = with more table lookups.

YES, the complication = is the CORRECT number of LSAs needed in a block to minimize a duplicate = mapping,
where all items in the table line match, but the set = of LSAs don't match within the = group.



If = that 50K LSA network was represented as 30 checksum blocks, and it takes = a second for the basic handshaking to take place with a neighbor, the = odds are decent that *all* of them will fail, in which case it's a waste = of time.  Keep in mind that if you want to avoid bad = checksums, you need to make the block size small, and thus the number of = blocks large, and then you end up back where you started from unless the = checksum-exchange protocol is reasonably windowed as = well.

Actually, how long does it = take to process 1 1500 byte packet and do say 100 table lookups. The key = here is that not
the entire LSDB summary needs to be exchanged = per packet. We could say that every 15 secs the next set of = groups
is xmit'ed and every x minutes you cover all the LSAs = via the groupings. So, I would expect less than 10ms to = processs
each packet.


If = you want to improve adjacency bringup in a non-compatible way, it seems = like a more effective mechanism would be simply to fix the brain-dead = DBD protocol so that it is not stop-and-wait;  if you can pump a = few hundred PPS of DBDs across the wire without stopping, you can sync = up very quickly in terms of wall-clock time without getting into the = inevitable problems caused by trying to glom LSAs together and expecting = them to sit still in the process.  It seems a lot cheaper and more = straightforward.

The = wireless win is the minimum number of bytes that COULD then be sent over = the connection and the ability
to have dropouts and = verifications that the LSDB is in sync and if not, why = not.

With a protocol like OSPF that has aged, I = would think that simpler approaches to 
scale them in a = transparent way would make more = sense.

With chksums, I would think that a = DRother forming an adj with a BDR after the DR, should be able = to
do this within just a few 100 packets. If the fastpath = allows a common read-mostly table lookup and can
remove 90% of = LSA hdr exchanges, then the ability for the BDR to support 100+ = simultaneous adj formations
should sum to your 100s of packets = per sec rate.

With the ability to not send say = 100s packets before a response to a DRother 
during = database exchange prevents the DR and/or BDR to = FLOOD/overload
 the DRother with LSAs. TCP's slow start = is a multiplicative increase to identify
not just un-used = capacity but also buffer capacity. If a flood of 100 packets = is
jjust sent, and they are fragmented, they really need to be = reasembled, etc.
This is assuming no tail drop or RED, where a = dropped fragment requires
the sending of the entire = packet/segment. The problem is not all routers would
have the = same pps capacity and if they did and say 5 DRothers implement = your
flooding to a single DR, wouldn't that be a problem? With = a ACK based protocol,
the two communicating end points can add = small latencies as needed and
prevent one end point to = overwhelm the other end point. Lastly, the amount of
CPU = cycles needed to process 100 pps tends to be greater than xmiting = the
100 pps rate.

Within an area, the = assumption is
that the link/media connection is probably at = 1ms or less and thus the round-trip-time should not
be a = concern. If the DRother and DR/or/BDR processes the packet within say = 1ms each, then
the sum of packets per connection should easily = reach your 100s of packets per sec.

Yes, with a = ACK like protocol, having the the ability to poll or generate just 1 = interrupt
for x number of packets can not be done with only 1 = set of concurrent connections. If
you had say 100 simultaneous = adj formations, yes you can then implement the = above.



Bandwidth = and CPU are practically free (particularly for something done relatively = rarely) so trying to do heroic things to optimize such situations seems = like a waste of effort.

The above = SHOULD use a fraction of the bandwidth and CPU. The gains of doing so is = the ability
to verify that all the LSDBs are in synch within = the area every sec. The decrease of bandwidth
is a decrease in = latency when differences exist and the ability to correct them quickly. = IF flooding
uses 1% of a CPU, we then have 99% to recompute = routing changes. With 1Gb and faster Eth
media, having short = convergent cycles is imperative. With this additional CPU cycles and low = cost
memory, we have the abiltiy to compute backup routes. And = to be able to separate the LSA
updates into age updates versus = content updates. The later need to be processed as a = high
priority and moving the former into Chksum verification, = should be able to take LSA content
changes and reconverge the = area within a subsec timeframe.

However, alot = of end systems (and routers) are or becoming mobile and their bandwidth = is in the 
kilobytes/sec capabilities, have expected = dropouts, and we still need to route to them. Thus = adding
downward scalability shouldn't be a = negative.


Maybe I'm missing something = fundamental here? =  

--Dave


<= div>On Apr 27, 2009, at 9:47 AM, Mitchell Erblich wrote:

Group,

First my = thoughts are currently in the inception phase
and COULD = apply to both wired and wireless 
= environments and I applogize for my = verboseness.

My initial assumptions are that a = OSPF area will be
majority built using the common = Ethernet (BMA) environment
and once a OSPF area = is formed major LSA changes are = rare.

Second is that the LSDB is = comprised of say 50k LSAs
which are mostly/90% = ExternalLSAs. Thus, Anton, I do not think
your = numbers are correct. If we had routers forming adjs
via = standard LSDB exchange each 100k LSAs sum exchange = with
DR/BDR and using checksums. Then = no ExternalLSA hdrs would
need to be exchanged with the = BDR, thus removing almost
 50% from database exchange. = FYI: We got them from the DR.

My below = inquiry stated a checksum per LSA type and
the = ability to generate subgroups of checksums per
LSA type. = Yes, it also stated a full LSDB = checksum.

Since the flooding of LSAs is = mostly the same, even a few
LSAs changing within the = ExternalLSa type COULD be covered
by = specifying that the A-Y range did not change and the
Z range = has new hdrs/LSAs and updated = normally.

If the later/changes was false, = during convergence, routing anomalies
would be = a major concern and data would be routed improperly
or just = lost. With OLD standardized TCP congestion avoidance
= algorithms, a single lost packet assumes congestion = and
assuming 100% bandwidth usage = drops TCP to 50% with
multiplicative decrease. Then a = huge number of round trips
before additive increase gets = back to approx 100% bandwidth.
Even = changes in routing paths can generate out-of-orderness = with
= the same result. Different hosts/end systems can have = different
Congestion Avoidance = algorithms.

Thus, if convergences are = frequent then the utilization
of each of the TCP connection = paths would have frequently
poor bandwidth usage. So the = assumption is that once a
OSPF area is formed, it tends to = be UNCHANGING.
=

I mentally prefer a predefined = range such that routers
can pre-compute the checksums = before even the
initial LSDB exchange. To = predetermine where the
breaks should exist is currently = unknown.

I see three possible major wins = where checksums COULD
be used. 
=      The first is flooding from the DR to = DRothers
     =   to update the LSDB so the amount of flooding = could
     =   be minimized, with the same same effect of DNA = LSAs.
Effectively a group LSA = refresh.
  The second is when a = DRother initially forms an adjacency
=       with the BDR and accepting that the = ExternalLSAs would
      be the = majority of LSAs and the belief that the DRother
=       would already have the ExternalLSAs in = its LSDB.
   The third is when = the DRother wants to just verify that
= no LSAs were lost due to either congestion, = temp
drop of the adj, etc and = verfify that no-one has either
= given up rexmits or is still doing xmits. And to pick = up
= a single dropped LSA without a full Database = Exch.

Getting = back to the pre-computation would prevent a
OSPF speaking router from doing = the calcs at adj
formation time. Even doing the = calc at adj formation
should remove say 90+% of LSAs = from needing to
be recomputed on each LSDB verify = sync.

However, to decrease the need to = do the calcs at
overhead, first checking whether = LSA type counts
match could be done and only if = they match, then
do the necessary computation. The = type, count,
and checksum, should generate = some form of
uniqueness. Some field within the = begining and
ending LSAs might add some more = verifcation.

So, what am I missing here? and = why hasn't anybody
done = this?

Mitchell = Erblich
= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
=

=
On Apr 27, 2009, at 12:57 AM, Emmanuel = Baccelli wrote:

Hi Anton,

yes, a similar scheme was = studied a few years back, in this paper:

Such a mechanism could indeed lead = to interesting performance enhancements, I agree with = Richard.

cheers,

Emmanue= l



On Mon, = Apr 27, 2009 at 2:22 AM, Richard Ogier <ogier@earthlink.net> = wrote:
Anton,

I = like your idea of sending a checksum for a range of LSAs, to
= increase the probability that the LSAs in a given range are in
sync = between the two neighbors.  (This helps even if the
age field = is ignored when selecting the ranges.)

For example, suppose = there are 100 LSAs and each LSA has a 2%
chance of not being in = sync.  Then the probability that all
LSAs are in sync is = .98^100 =3D .1326, too low to result in
a significant overhead = reduction.
Now suppose the LSAs are divided into 10 ranges, each = with 10 LSAs.
The probability that all LSAs in a given range are in = sync is then
.98^10 =3D .8171, resulting in a big overhead reduction = on average.

The selected range for each LSA type can depend on = how frequently
new LSAs are originated for that type, with a larger = range chosen
if LSAs are originated less frequently (so that the = probability
of being out of sync is lower).

This idea can = be combined with the database exchange optimization
of RFC 5243 to = further reduce overhead.  I.e., if the checksums
do not match, = then RFC 5243 can be used to reduce the number of
LSA headers = exchanged in DD packets by up to 50%.
Also, if the checksums do = match for a given range of LSAs, then the
router receiving the = checksum need not send the same checksum for
the same range of = LSAs.

= Richard


Anton Smirnov = wrote:

 Mitchell,
=  LSDBs may be very close but they are seldom EXACTLY the same. = The
thing is that change in adjacency triggers LSA [re]build, so = depending
on timing router and/or network LSAs may change by the = time adjacency
restores.
 Also, in moderately sized modern = OSPF network there is constant
background noise of LSAs being = refreshed. So again, chances that
refreshed LSA arrived to one side = but not another while adjacency is
down are relatively high.
=  These considerations make precomputed checksums over large number = of
LSAs rather bad for the purpose.
 If you want to explore = venue of checksumming database exchange then I
recommend to look for = more flexible approach giving routers possibility
to choose range of = LSAs to compute checksum over. For example, if both
routers support = the checksum exchange then router sending DBD could
specify the = first, the last LSA and value of some kind of hash of LSA
headers = between them (LSAs in LSDB can be viewed as ordered sequence as
per = RFC 5243). In this way router preparing DBD packet may place start
= and end points of checksumming to avoid including into the range = LSAs
with ages around 0 and MAXAGE (because these are LSAs which are = less
likely to match between routers). So database exchange will = consist of
series of single LSA headers (the 'old' way) and large = LSA sequences.

Anton



Mitchell Erblich = wrote:
 
Group,

=       General thoughts...

      = After a conversation about OSPF's standard database exchange and
how = inefficient it
      is when two routers have large = LSDBs and them almost being the
same, we concluded
    =   that the use of Checksums might significantly decrease the
= amount of overhead and
      achieve faster = convergence.

          Moy first (I = think) suggested the LSDB checksum support to
identify identical
=           LSDBs.  Sometime later the MIB = supported the
ospfExternLsaChksumSum,
      =     ospfAreaLsaChksumSum, and ospfLsdbChecksum fields.
=
          Thus, why not support the = ability to have OSPF routers first
exchange
      =     checksums with LSA counts. If the LSDBs have more than = XYZ
entries
          in their LSDBs, = then the counts which are above specific
thresholds then
  =         support the checksum on each X count of LSA = type. Only the
groups
          with = non-matching Checksums would then go thru database
exchange.
=
          Thus, before database exchange, = support some form of triple
that
        =   as routers have almost the same LSDBs,   the checksum = exchange
           would remove = unnecessary LSA header exchange.

        =   This, IMO, would also be a big win if say a DRother was
= adjacent, lost
          adjacency, and = re-exstablished it a short time later while
keeping a
  =         past snapshot. It could remove its own = originated LSAs and
compare
          = checksums..

          In this = scenario, within BMA areas, I think only DR /
DRother and BDR/ = DRother
          need to be making this = Checksum exchange..

          Wouldn't = the exchange of this data benefit each router
receiving the
=           information and allow both routers to = know the number of
LSAs to
          be = possibly exchanged?
              =       Yes, a new capability would probably have to be = added.


          So, the first = question is whether the checksums are unique
enough with a
=           count and type to determine whether = that section of the LSDB
matches
        =   between two routers?

          = And then what savings would normally exist in this hybrid
database = exchange
          and is it worth the = effort?
         
      = Mitchell Erblich

     
      =         =  _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=  
= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=
 
= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/mailman/listinfo/ospf
=

= _______________________________________________
OSPF mailing = list
OSPF@ietf.org
https://www.ietf.org/m= ailman/listinfo/ospf

___________= ____________________________________
OSPF mailing list
OSPF@ietf.org
https://www.ietf.org/m= ailman/listinfo/ospf


= --Apple-Mail-240-704069833-- From dkatz@juniper.net Wed Apr 29 08:53:33 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 60BB13A715F for ; Wed, 29 Apr 2009 08:53:33 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.111 X-Spam-Level: X-Spam-Status: No, score=-6.111 tagged_above=-999 required=5 tests=[AWL=-0.112, BAYES_00=-2.599, J_CHICKENPOX_42=0.6, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id rVcB823HaEDE for ; Wed, 29 Apr 2009 08:53:31 -0700 (PDT) Received: from chip3og57.obsmtp.com (chip3og57.obsmtp.com [64.18.14.179]) by core3.amsl.com (Postfix) with ESMTP id C81A728C2D0 for ; Wed, 29 Apr 2009 08:52:25 -0700 (PDT) Received: from source ([66.129.224.36]) (using TLSv1) by chip3ob57.postini.com ([64.18.6.12]) with SMTP ID DSNKSfh4Cbto86pZifJ3H0I7lbLoQzx3FTAB@postini.com; Wed, 29 Apr 2009 08:53:48 PDT Received: from p-emfe01-sac.jnpr.net (66.129.254.72) by P-EMHUB01-HQ.jnpr.net (172.24.192.35) with Microsoft SMTP Server id 8.1.340.0; Wed, 29 Apr 2009 08:34:25 -0700 Received: from p-emlb01-sac.jnpr.net ([66.129.254.46]) by p-emfe01-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Wed, 29 Apr 2009 08:34:24 -0700 Received: from emailsmtp55.jnpr.net ([172.24.18.132]) by p-emlb01-sac.jnpr.net with Microsoft SMTPSVC(6.0.3790.3959); Wed, 29 Apr 2009 08:34:24 -0700 Received: from merlot.juniper.net ([172.17.27.10]) by emailsmtp55.jnpr.net with Microsoft SMTPSVC(6.0.3790.1830); Wed, 29 Apr 2009 08:34:23 -0700 Received: from sa-nc-spg-32.static.jnpr.net (sa-nc-spg-32.static.jnpr.net [172.23.1.32]) by merlot.juniper.net (8.11.3/8.11.3) with ESMTP id n3TFYN074558; Wed, 29 Apr 2009 08:34:23 -0700 (PDT) (envelope-from dkatz@juniper.net) Message-ID: <63CE22DA-FB2B-433E-BF3C-1F1EBE3B5D18@juniper.net> From: Dave Katz To: Mitchell Erblich In-Reply-To: Content-Type: text/plain; charset="US-ASCII"; format=flowed; delsp=yes Content-Transfer-Encoding: 7bit MIME-Version: 1.0 (Apple Message framework v930.3) Date: Wed, 29 Apr 2009 10:34:22 -0500 References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> X-Mailer: Apple Mail (2.930.3) X-OriginalArrivalTime: 29 Apr 2009 15:34:23.0876 (UTC) FILETIME=[F9125440:01C9C8DF] Cc: ospf@ietf.org, Emmanuel Baccelli Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 29 Apr 2009 15:53:33 -0000 On Apr 29, 2009, at 6:24 AM, Mitchell Erblich wrote: > Group, > > Sorry, verbose.. > > Inline.. > > Mitchell Erblich > =============== > On Apr 27, 2009, at 4:20 PM, Dave Katz wrote: > >> On first principles, it seems as though the only reason to do >> something like this is to reduce the amount of time it takes to >> bring up a new link, and even that is probably only interesting if >> the network is at least slightly partitioned without it. Keep in >> mind that such a scheme is only useful if the database is already >> close to synchronized, so it doesn't help when a router first comes >> up. Given all that, it's not entirely clear that doing this is >> *that* interesting, but no matter. > > Actually, the first thought was how to verify that two LSDBs are > identical at ANY time after Database Exchange. > And if not first what are the differences wrt LSA type and count. > And have the router periodicly report / summarize > any seen diffs. This already happens by magic over a 60 minute period, and is unnecessary in any case unless the implementation is broken (in which case, adding additional complexity is not likely to improve it.) > > Then how to resynch without waiting for the LSA update or renewing > (tear down and reforming the adj) the adj A generalized refresh mechanism would make things like hot standby much easier (it's very tricky to do without it) but seems to be orthogonal to what is being presented here, as you could easily use the existing mechanisms to leverage a refresh-without-teardown scheme. > > However, if you can then resync, you should be able to do refreshes > from the DR and BDR (common Eth). I don't understand what you think is special about the DR or BDR as opposed to any other adjacent router. DR-ness is just an adjacency pruning hack, nothing more. > > However, the first interaction between two OSPF speakers is with > Database Exchange and decreasing the > workload on the 2nd adj due to common LSAs should allow the > formation of 2x simulateneous adj formations > than standard. > >> >> The problem with any of these checksum schemes is that the benefits >> are compromised by the LSA refresh mechanism, as has been noted. >> With 50,000 LSAs, my back-of-the-laptop calculations are that an >> LSA is refreshed on the average of once every 36 msec, or roughly >> 30 per second. Scalable implementations go out of their way to >> smooth refreshes when they are originating a large number of LSAs, >> and if there are a large number of machines originating a small >> number of LSAs, these will (hopefully) tend to be smoothed out as >> well, in a way that does not correlate with the ordering mechanism >> proposed. > > First, 50k was just a number picked that would identify a common > large number of LSAs of a specific type(Externals) > (also the major of LSAs in the LSDB) that should all match and not > have any new DRother add to this LSA type pool. > > The simplified assumption is that the updates are just age refreshes. Refreshed LSAs are no different than LSAs with significant changes, from the standpoint of flooding--even if the contents haven't changed, they still need to be flooded and the identities of those to be flooded need to be ascertained. The fact of the matter is that the LSDB entropy grows with the size, and very large databases are very entropic, and thus resistant to any sort of blocky verification. > > I am not sure that this/checksums would be a better approach then > say DNA LSAs. The chksum approach fits better > IMO in this case when a router stops rexmits of the LSA and has > taken down the adj. The DRother either would > have to renew the adj or implement a search for the lost LSA. I'm lost here. What "lost LSA" are we searching for? Why would a DROther not learn it in any case (from any other box with which it is adjacent)? > > In the common case of Eth, with your flooding example, in my limited > experience, the OSPF routers tend to > initially come up within the area at the same time and thus > iniitally the LSAs are mostly flooded within the > same quarter hour, if not within the first 5 minutes. Thus, peaks of > LSAs per second would be on the order > of 200 LSAs per sec. That's not my experience, but even if it were, our implementation rapidly covers this up so that there are no LSA peaks. Refreshes are smeared smoothly across the refresh period precisely to avoid peaks, but of course this causes worst-case LSDB entropy. Keep in mind that this flooding load is per-adjacency as well. > > Since the chksum is even simpler than a LSA hdr, I would expect more > than a 100 to 10kx LSA processing of > just LSA age refreshes should be able to be accomplished with the > same level of work. But LSA processing is simply not that costly. > > Somewhere in these numbers is a meltdown by a router with > standardized OSPF, where I question whether > the chksum approach would have the same failure behaviour. OSPF meltdowns in my experience are not caused by a lack of CPU; they are caused by poor queueing and memory management behavior and stupid implementations that forget to send Hellos when they are busy. > >> >> The efficiency of such schemes improves as the number of LSAs per >> checksum block goes up, but so does the probability of a checksum >> failure, in which case you're actually worse off than if you'd just >> send the darned things (since you have to anyhow, unless you start >> doing scary bifurcations of the space, all of which takes more >> time, and all of which causes even more LSAs to be stale.) > > The assumption is that a checksum is specified for the entire LSDB, > then a chksum for all the LSAs of a specified > type, then a breakdown of each type. If we have Eth, we have a > 1500byte packet. We have no LSA lookups if > we can can match each table line. If we add additional values like > LSA count, LSA type, etc, to each group > checksum, then the same chksums that map to different sets of LSAs > should be minimized. Remember, we > need a double failure/mapping because not just the group must map > identically but the chksum over the > entire LSA type. > > FYI, yes, I know about the 64k byte offset of IP, but I don't want > reassembly overhead so a limit of 1.5k. > > What the additional groups get us, is the ability to locate request > entire LSA groups without needing > all of the LSAs. The smaller the group, then the smaller number of > extra LSAs that we really don't > need but with more table lookups. > > YES, the complication is the CORRECT number of LSAs needed in a > block to minimize a duplicate mapping, > where all items in the table line match, but the set of LSAs don't > match within the group. I don't really follow all this, but I'll point out once again that because of refresh smoothing, the probability that *any* blocking scheme will work is very low when there are large numbers of LSAs. > > >> >> If that 50K LSA network was represented as 30 checksum blocks, and >> it takes a second for the basic handshaking to take place with a >> neighbor, the odds are decent that *all* of them will fail, in >> which case it's a waste of time. Keep in mind that if you want to >> avoid bad checksums, you need to make the block size small, and >> thus the number of blocks large, and then you end up back where you >> started from unless the checksum-exchange protocol is reasonably >> windowed as well. > > Actually, how long does it take to process 1 1500 byte packet and do > say 100 table lookups. The key here is that not > the entire LSDB summary needs to be exchanged per packet. We could > say that every 15 secs the next set of groups > is xmit'ed and every x minutes you cover all the LSAs via the > groupings. So, I would expect less than 10ms to processs > each packet. You're completely missing the point, I think. Or maybe I am. In 15 seconds in a 50K LSA network (at least with our routers) there will be 400 changed LSAs, randomized across however you divide things into blocks. If there are less than O(400) blocks, the probability of any of them being unchanged is poor. > >> >> If you want to improve adjacency bringup in a non-compatible way, >> it seems like a more effective mechanism would be simply to fix the >> brain-dead DBD protocol so that it is not stop-and-wait; if you >> can pump a few hundred PPS of DBDs across the wire without >> stopping, you can sync up very quickly in terms of wall-clock time >> without getting into the inevitable problems caused by trying to >> glom LSAs together and expecting them to sit still in the process. >> It seems a lot cheaper and more straightforward. > > The wireless win is the minimum number of bytes that COULD then be > sent over the connection and the ability > to have dropouts and verifications that the LSDB is in sync and if > not, why not. > > With a protocol like OSPF that has aged, I would think that simpler > approaches to > scale them in a transparent way would make more sense. > > With chksums, I would think that a DRother forming an adj with a BDR > after the DR, should be able to > do this within just a few 100 packets. If the fastpath allows a > common read-mostly table lookup and can > remove 90% of LSA hdr exchanges, then the ability for the BDR to > support 100+ simultaneous adj formations > should sum to your 100s of packets per sec rate. > > With the ability to not send say 100s packets before a response to a > DRother > during database exchange prevents the DR and/or BDR to FLOOD/overload > the DRother with LSAs. TCP's slow start is a multiplicative > increase to identify > not just un-used capacity but also buffer capacity. If a flood of > 100 packets is > jjust sent, and they are fragmented, they really need to be > reasembled, etc. > This is assuming no tail drop or RED, where a dropped fragment > requires > the sending of the entire packet/segment. The problem is not all > routers would > have the same pps capacity and if they did and say 5 DRothers > implement your > flooding to a single DR, wouldn't that be a problem? With a ACK > based protocol, > the two communicating end points can add small latencies as needed and > prevent one end point to overwhelm the other end point. Lastly, the > amount of > CPU cycles needed to process 100 pps tends to be greater than > xmiting the > 100 pps rate. > > Within an area, the assumption is > that the link/media connection is probably at 1ms or less and thus > the round-trip-time should not > be a concern. The RTT on most networks I've seen are 10s or 100s of msec (they are geographically diverse.) If you're using slow media (which I thought was one of the points of the exercise) the latency is also very high. A 1500 byte packet on a 56K link has over 200 msec of latency even before the speed of light comes into play. > If the DRother and DR/or/BDR processes the packet within say 1ms > each, then > the sum of packets per connection should easily reach your 100s of > packets per sec. > > Yes, with a ACK like protocol, having the the ability to poll or > generate just 1 interrupt > for x number of packets can not be done with only 1 set of > concurrent connections. If > you had say 100 simultaneous adj formations, yes you can then > implement the above. > > >> >> Bandwidth and CPU are practically free (particularly for something >> done relatively rarely) so trying to do heroic things to optimize >> such situations seems like a waste of effort. > > The above SHOULD use a fraction of the bandwidth and CPU. The gains > of doing so is the ability > to verify that all the LSDBs are in synch within the area every sec. > The decrease of bandwidth > is a decrease in latency when differences exist and the ability to > correct them quickly. IF flooding > uses 1% of a CPU, we then have 99% to recompute routing changes. > With 1Gb and faster Eth > media, having short convergent cycles is imperative. With this > additional CPU cycles and low cost > memory, we have the abiltiy to compute backup routes. And to be able > to separate the LSA > updates into age updates versus content updates. The later need to > be processed as a high > priority and moving the former into Chksum verification, should be > able to take LSA content > changes and reconverge the area within a subsec timeframe. > > However, alot of end systems (and routers) are or becoming mobile > and their bandwidth is in the > kilobytes/sec capabilities, have expected dropouts, and we still > need to route to them. Thus adding > downward scalability shouldn't be a negative. > IMHO, you have a solution looking for a problem here, and the solution seems to be uninteresting. There's lots of talk about CPU optimization, and while it may be true that something could take 50% less CPU, if that thing is only taking a nominal amount of CPU to begin with (and is not the long pole in the tent) that optimization may be uninteresting. OSPF as a whole doesn't scale well for large networks with small, unreliable bandwidth, and any scheme to improve initial database synchronization is only a patch on a much larger problem. Rather than coming up with a one-off fix for part of the protocol (which may not even be viable), it seems like the right thing to do is to start on first principles. What problem are you trying to solve, on the large scale? Is OSPF (or link-state routing in general) a good match for solving this problem? What other kinds of things need to be fixed in that problem space? But that's just my opinion... --Dave From ogier@earthlink.net Thu Apr 30 12:27:04 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 0B14D3A6B96 for ; Thu, 30 Apr 2009 12:27:04 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.945 X-Spam-Level: X-Spam-Status: No, score=-1.945 tagged_above=-999 required=5 tests=[AWL=0.654, BAYES_00=-2.599] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 6HG0vomJiJPg for ; Thu, 30 Apr 2009 12:27:03 -0700 (PDT) Received: from elasmtp-kukur.atl.sa.earthlink.net (elasmtp-kukur.atl.sa.earthlink.net [209.86.89.65]) by core3.amsl.com (Postfix) with ESMTP id 54F4A3A67DB for ; Thu, 30 Apr 2009 12:27:03 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=dk20050327; d=earthlink.net; b=jJMvISAsXQ44LzF4QO70tEbjn6za+MnbC7MHoe4nb+WmbKU7nanAKG4CfIpXegpK; h=Received:Message-ID:Date:From:User-Agent:X-Accept-Language:MIME-Version:To:Subject:Content-Type:Content-Transfer-Encoding:X-ELNK-Trace:X-Originating-IP; Received: from [66.81.220.90] by elasmtp-kukur.atl.sa.earthlink.net with esmtpa (Exim 4.67) (envelope-from ) id 1Lzbvl-0002pg-JJ for ospf@ietf.org; Thu, 30 Apr 2009 15:28:26 -0400 Message-ID: <49F9FC9C.7070509@earthlink.net> Date: Thu, 30 Apr 2009 12:31:40 -0700 From: Richard Ogier User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.2) Gecko/20040804 Netscape/7.2 (ax) X-Accept-Language: en-us, en MIME-Version: 1.0 To: ospf@ietf.org Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-ELNK-Trace: a073897a9455599e74bf435c0eb9d478594a5b3a834a310e40658b7f116ba216021ad78892ff807e350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c X-Originating-IP: 66.81.220.90 Subject: [OSPF] draft-ietf-ospf-lls X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 30 Apr 2009 19:27:04 -0000 The LLS draft is in state IESG Evaluation :: Revised ID Needed. https://datatracker.ietf.org/idtracker/draft-ietf-ospf-lls/ When can we expect the draft to be revised? The OSPF-MDR draft (draft-ietf-ospf-manet-mdr) is held up in the RFC Editor Queue waiting for the LLS draft to be approved (since the LLS draft is a normative reference). I am starting to wish I had incorrectly referenced RFC 4813 (LLS for IPv4), as the OSPF-MPR draft did. Richard From akr@cisco.com Thu Apr 30 13:03:15 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 561233A6885 for ; Thu, 30 Apr 2009 13:03:15 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -6.599 X-Spam-Level: X-Spam-Status: No, score=-6.599 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, RCVD_IN_DNSWL_MED=-4] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id VcXq68n93UxY for ; Thu, 30 Apr 2009 13:03:14 -0700 (PDT) Received: from sj-iport-6.cisco.com (sj-iport-6.cisco.com [171.71.176.117]) by core3.amsl.com (Postfix) with ESMTP id 78A663A6B66 for ; Thu, 30 Apr 2009 13:03:14 -0700 (PDT) X-IronPort-AV: E=Sophos;i="4.40,274,1238976000"; d="scan'208";a="296082119" Received: from sj-dkim-2.cisco.com ([171.71.179.186]) by sj-iport-6.cisco.com with ESMTP; 30 Apr 2009 20:04:30 +0000 Received: from sj-core-1.cisco.com (sj-core-1.cisco.com [171.71.177.237]) by sj-dkim-2.cisco.com (8.12.11/8.12.11) with ESMTP id n3UK4UBh008061; Thu, 30 Apr 2009 13:04:30 -0700 Received: from [171.70.225.207] (dhcp-171-70-225-207.cisco.com [171.70.225.207]) by sj-core-1.cisco.com (8.13.8/8.13.8) with ESMTP id n3UK4UnI013600; Thu, 30 Apr 2009 20:04:30 GMT Message-ID: <49FA044E.3050200@cisco.com> Date: Thu, 30 Apr 2009 13:04:30 -0700 From: Abhay Roy User-Agent: Thunderbird 2.0.0.21 (Windows/20090302) MIME-Version: 1.0 To: Richard Ogier References: <49F9FC9C.7070509@earthlink.net> In-Reply-To: <49F9FC9C.7070509@earthlink.net> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; l=831; t=1241121870; x=1241985870; c=relaxed/simple; s=sjdkim2002; h=Content-Type:From:Subject:Content-Transfer-Encoding:MIME-Version; d=cisco.com; i=akr@cisco.com; z=From:=20Abhay=20Roy=20 |Subject:=20Re=3A=20[OSPF]=20draft-ietf-ospf-lls |Sender:=20; bh=X3FJLtzfSIPdmg5k277X4m7rMHbl8Z2U2MIXRKQ+BS0=; b=hijfV9ti2tep1GeVv/4rQ6qFPKzcOItU8hIE7BMl8qW9ir5jGvcD49Ui6g pQ+17Z3bXhvmNZVMvNm7cC502DRfwo2H5t7aLU7zf5R5TulCGDFEkWW02qol NnqLZ33T7N; Authentication-Results: sj-dkim-2; header.From=akr@cisco.com; dkim=pass ( sig from cisco.com/sjdkim2002 verified; ); Cc: ospf@ietf.org Subject: Re: [OSPF] draft-ietf-ospf-lls X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 30 Apr 2009 20:03:15 -0000 Hi Richard, There are still a few discuss issues to be cleared. Let me ping them and see what else is needed in the next revision. Regards, -Abhay On 4/30/2009 12:31 PM, Richard Ogier wrote: > The LLS draft is in state IESG Evaluation :: Revised ID Needed. > https://datatracker.ietf.org/idtracker/draft-ietf-ospf-lls/ > > When can we expect the draft to be revised? The OSPF-MDR draft > (draft-ietf-ospf-manet-mdr) is held up in the RFC Editor Queue > waiting for the LLS draft to be approved (since the LLS draft is > a normative reference). I am starting to wish I had incorrectly > referenced RFC 4813 (LLS for IPv4), as the OSPF-MPR draft did. > > Richard > > > _______________________________________________ > OSPF mailing list > OSPF@ietf.org > https://www.ietf.org/mailman/listinfo/ospf From asmirnov@cisco.com Thu Apr 30 15:11:40 2009 Return-Path: X-Original-To: ospf@core3.amsl.com Delivered-To: ospf@core3.amsl.com Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 4B4373A6853 for ; Thu, 30 Apr 2009 15:11:40 -0700 (PDT) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.107 X-Spam-Level: X-Spam-Status: No, score=-0.107 tagged_above=-999 required=5 tests=[BAYES_00=-2.599, FRT_ESTABLISH2=2.492] Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id OiSx4pNfQt05 for ; Thu, 30 Apr 2009 15:11:38 -0700 (PDT) Received: from av-tac-bru.cisco.com (weird-brew.cisco.com [144.254.15.118]) by core3.amsl.com (Postfix) with ESMTP id 0CC9D3A6A18 for ; Thu, 30 Apr 2009 15:10:47 -0700 (PDT) X-TACSUNS: Virus Scanned Received: from strange-brew.cisco.com (localhost.cisco.com [127.0.0.1]) by av-tac-bru.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3UMC8st011879; Fri, 1 May 2009 00:12:09 +0200 (CEST) Received: from [10.55.140.89] (ams-asmirnov-8718.cisco.com [10.55.140.89]) by strange-brew.cisco.com (8.13.8+Sun/8.13.8) with ESMTP id n3UMC50H014585; Fri, 1 May 2009 00:12:06 +0200 (CEST) Message-ID: <49FA2235.7080507@cisco.com> Date: Fri, 01 May 2009 00:12:05 +0200 From: Anton Smirnov Organization: Cisco Systems, Inc. User-Agent: Thunderbird 2.0.0.19 (X11/20081227) MIME-Version: 1.0 To: Dave Katz References: <49EE0286.9070903@cisco.com> <49F4FAB9.4030009@earthlink.net> <0E0F4684-5F14-4E13-B573-DC1F32C31317@juniper.net> <49F6DF5D.80409@cisco.com> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: ospf@ietf.org Subject: Re: [OSPF] Preliminary idea: OSPF Database Exchange : Pre-Step X-BeenThere: ospf@ietf.org X-Mailman-Version: 2.1.9 Precedence: list List-Id: The Official IETF OSPG WG Mailing List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 30 Apr 2009 22:11:40 -0000 Hi Dave, > I guess I'd be impressed with a network that had 50,000 LSAs but very > limited bandwidth and CPU. I have a feeling that such a network would > have bigger problems than the time it takes to bring up an additional link. In the context of this discussion example with 50000 LSAs does not interest me at all, MANET or not. Not because it is a difficult case. Vice versa, it is unrealistically too optimal case. >>> If that 50K LSA network was represented as 30 checksum blocks, and it >>> takes a second for the basic handshaking to take place with a neighbor, >>> the odds are decent that *all* of them will fail, in which case it's a >>> waste of time. Keep in mind that if you want to avoid bad checksums, >>> you need to make the block size small, and thus the number of blocks >>> large, and then you end up back where you started from unless the >>> checksum-exchange protocol is reasonably windowed as well. Not at all. Refreshed LSAs are easy to exclude from checksumming on both sides exactly because they are known to be young. Let me do MY math :-) Suppose LSDB of routers were in sync N seconds ago. In this case LSA ages for older LSAs should be within N +- InfTransDelay. Lets take a value k with sufficiently big margin, say 15. Router sending DBD places start and end points of checksumming range so that the range does not cover any LSAs with age of N +- k (but may cover LSAs with age less than N-k; those are 'new' refreshed LSAs). After that checksum is calculated over the range but LSAs with age less than N-k are not included into the checksum. Router receiving DBD will have N and start/end points. It too calculates the checksum over the range excluding from it LSAs with age less than N. Value of 2k is << MaxAge (1-2%), hence it is very easy to choose range with good chances of eligible LSAs in the range being in sync. As for the value of N - it can be known from prior events or it can even be estimated dynamically. If a router receives DBD describing LSA with age n and finds this LSA in its LSDB then it is not unreasonable to expect that n seconds ago LSDBs were in sync. Hence N can be set to n. There are some scenarios where this assumption fails, for example if adjacency was terminated while LOADING. But estimation is good enough to try using this for sending older LSAs as checksum ranges. You may wonder what's my point :-) My point is that technically idea of checksumming can be made working. And it will work the better the more LSAs are in LSDB. The downside of it is that it is not backward compatible. So the question is if this scenario needs to be solved even if it can be solved. Or if it needs to be solved by checksumming, by complicating DBD packet serialization or by some other means. Thanks, Anton Dave Katz wrote: > > On Apr 28, 2009, at 5:50 AM, Anton Smirnov wrote: > >> Hi Dave, >> I agree with you in most (but not all) statements. >> >>> Bandwidth and CPU are practically free >> >> Unless you consider MANETs where bandwidth is not free and CPU >> resources may also be relatively limited. >> Originally idea of checksumming was born in context of MANETs. And in >> that environment exchange of database checksums may actually work well. >> The first Mitchell's post was about network scenario NOT related to >> MANETs but questioned how well that idea will work. So my post was more >> to point out that no, idea conceived with MANET in mind will not work in >> non-MANET environment. At least not until the idea is properly >> generalized. > > I guess I'd be impressed with a network that had 50,000 LSAs but very > limited bandwidth and CPU. I have a feeling that such a network would > have bigger problems than the time it takes to bring up an additional link. > >> >> >>> If you want to improve adjacency bringup in a non-compatible way, it >>> seems like a more effective mechanism would be simply to fix the >>> brain-dead DBD protocol so that it is not stop-and-wait >> >> Agree. >> >> >>> if you can >>> pump a few hundred PPS of DBDs across the wire without stopping, you can >>> sync up very quickly in terms of wall-clock time without getting into >>> the inevitable problems caused by trying to glom LSAs together and >>> expecting them to sit still in the process. It seems a lot cheaper and >>> more straightforward. >> >> That's good only in case of homogeneous network of powerful routers >> connected by high-bandwidth links. >> It is likely to give a lot of troubles in star topologies where one >> router has many neighbors. Fast DBD flooding from spokes will just swamp >> the hub router. >> And I have mentioned it is not going to work in case of MANETs. > > But this is no different than the situation with LSUpd packets today, > and needs to be solved in the same way (hack.) We used to deal with a > much worse situation ten or fifteen years ago, which were sizable full > meshes. If it were easy, we wouldn't make any money doing it. ;-) > > --Dave > >> >> >> Of course, transport protocols have solved all these issues long time >> ago. But we don't want to turn DBD into window-based protocol with slow >> start, congestion avoidance and all the stuff :-) >> >> Thanks, >> >> Anton >> >> >> >> Dave Katz wrote: >>> On first principles, it seems as though the only reason to do something >>> like this is to reduce the amount of time it takes to bring up a new >>> link, and even that is probably only interesting if the network is at >>> least slightly partitioned without it. Keep in mind that such a scheme >>> is only useful if the database is already close to synchronized, so it >>> doesn't help when a router first comes up. Given all that, it's not >>> entirely clear that doing this is *that* interesting, but no matter. >>> >>> The problem with any of these checksum schemes is that the benefits are >>> compromised by the LSA refresh mechanism, as has been noted. With >>> 50,000 LSAs, my back-of-the-laptop calculations are that an LSA is >>> refreshed on the average of once every 36 msec, or roughly 30 per >>> second. Scalable implementations go out of their way to smooth >>> refreshes when they are originating a large number of LSAs, and if there >>> are a large number of machines originating a small number of LSAs, these >>> will (hopefully) tend to be smoothed out as well, in a way that does not >>> correlate with the ordering mechanism proposed. >>> >>> The efficiency of such schemes improves as the number of LSAs per >>> checksum block goes up, but so does the probability of a checksum >>> failure, in which case you're actually worse off than if you'd just send >>> the darned things (since you have to anyhow, unless you start doing >>> scary bifurcations of the space, all of which takes more time, and all >>> of which causes even more LSAs to be stale.) >>> >>> If that 50K LSA network was represented as 30 checksum blocks, and it >>> takes a second for the basic handshaking to take place with a neighbor, >>> the odds are decent that *all* of them will fail, in which case it's a >>> waste of time. Keep in mind that if you want to avoid bad checksums, >>> you need to make the block size small, and thus the number of blocks >>> large, and then you end up back where you started from unless the >>> checksum-exchange protocol is reasonably windowed as well. >>> >>> If you want to improve adjacency bringup in a non-compatible way, it >>> seems like a more effective mechanism would be simply to fix the >>> brain-dead DBD protocol so that it is not stop-and-wait; if you can >>> pump a few hundred PPS of DBDs across the wire without stopping, you can >>> sync up very quickly in terms of wall-clock time without getting into >>> the inevitable problems caused by trying to glom LSAs together and >>> expecting them to sit still in the process. It seems a lot cheaper and >>> more straightforward. >>> >>> Bandwidth and CPU are practically free (particularly for something done >>> relatively rarely) so trying to do heroic things to optimize such >>> situations seems like a waste of effort. >>> >>> Maybe I'm missing something fundamental here? >>> >>> --Dave >>> >>> >>> On Apr 27, 2009, at 9:47 AM, Mitchell Erblich wrote: >>> >>>> Group, >>>> >>>> First my thoughts are currently in the inception phase >>>> and COULD apply to both wired and wireless >>>> environments and I applogize for my verboseness. >>>> >>>> My initial assumptions are that a OSPF area will be >>>> majority built using the common Ethernet (BMA) environment >>>> and once a OSPF area is formed major LSA changes are rare. >>>> >>>> Second is that the LSDB is comprised of say 50k LSAs >>>> which are mostly/90% ExternalLSAs. Thus, Anton, I do not think >>>> your numbers are correct. If we had routers forming adjs >>>> via standard LSDB exchange each 100k LSAs sum exchange with >>>> DR/BDR and using checksums. Then no ExternalLSA hdrs would >>>> need to be exchanged with the BDR, thus removing almost >>>> 50% from database exchange. FYI: We got them from the DR. >>>> >>>> My below inquiry stated a checksum per LSA type and >>>> the ability to generate subgroups of checksums per >>>> LSA type. Yes, it also stated a full LSDB checksum. >>>> >>>> Since the flooding of LSAs is mostly the same, even a few >>>> LSAs changing within the ExternalLSa type COULD be covered >>>> by specifying that the A-Y range did not change and the >>>> Z range has new hdrs/LSAs and updated normally. >>>> >>>> If the later/changes was false, during convergence, routing anomalies >>>> would be a major concern and data would be routed improperly >>>> or just lost. With OLD standardized TCP congestion avoidance >>>> algorithms, a single lost packet assumes congestion and >>>> assuming 100% bandwidth usage drops TCP to 50% with >>>> multiplicative decrease. Then a huge number of round trips >>>> before additive increase gets back to approx 100% bandwidth. >>>> Even changes in routing paths can generate out-of-orderness with >>>> the same result. Different hosts/end systems can have different >>>> Congestion Avoidance algorithms. >>>> >>>> Thus, if convergences are frequent then the utilization >>>> of each of the TCP connection paths would have frequently >>>> poor bandwidth usage. So the assumption is that once a >>>> OSPF area is formed, it tends to be UNCHANGING. >>>> >>>> >>>> I mentally prefer a predefined range such that routers >>>> can pre-compute the checksums before even the >>>> initial LSDB exchange. To predetermine where the >>>> breaks should exist is currently unknown. >>>> >>>> I see three possible major wins where checksums COULD >>>> be used. >>>> The first is flooding from the DR to DRothers >>>> to update the LSDB so the amount of flooding could >>>> be minimized, with the same same effect of DNA LSAs. >>>> Effectively a group LSA refresh. >>>> The second is when a DRother initially forms an adjacency >>>> with the BDR and accepting that the ExternalLSAs would >>>> be the majority of LSAs and the belief that the DRother >>>> would already have the ExternalLSAs in its LSDB. >>>> The third is when the DRother wants to just verify that >>>> no LSAs were lost due to either congestion, temp >>>> drop of the adj, etc and verfify that no-one has either >>>> given up rexmits or is still doing xmits. And to pick up >>>> a single dropped LSA without a full Database Exch. >>>> >>>> Getting back to the pre-computation would prevent a >>>> OSPF speaking router from doing the calcs at adj >>>> formation time. Even doing the calc at adj formation >>>> should remove say 90+% of LSAs from needing to >>>> be recomputed on each LSDB verify sync. >>>> >>>> However, to decrease the need to do the calcs at >>>> overhead, first checking whether LSA type counts >>>> match could be done and only if they match, then >>>> do the necessary computation. The type, count, >>>> and checksum, should generate some form of >>>> uniqueness. Some field within the begining and >>>> ending LSAs might add some more verifcation. >>>> >>>> So, what am I missing here? and why hasn't anybody >>>> done this? >>>> >>>> Mitchell Erblich >>>> =================== >>>> >>>> >>>> On Apr 27, 2009, at 12:57 AM, Emmanuel Baccelli wrote: >>>> >>>>> Hi Anton, >>>>> >>>>> yes, a similar scheme was studied a few years back, in this paper: >>>>> OSPF Database Exchange and Reliable Synchronization in Mobile Ad Hoc >>>>> Networks >>>>> >>>>> Such a mechanism could indeed lead to interesting performance >>>>> enhancements, I agree with Richard. >>>>> >>>>> cheers, >>>>> >>>>> Emmanuel >>>>> >>>>> >>>>> >>>>> On Mon, Apr 27, 2009 at 2:22 AM, Richard Ogier >>>> > wrote: >>>>> >>>>> Anton, >>>>> >>>>> I like your idea of sending a checksum for a range of LSAs, to >>>>> increase the probability that the LSAs in a given range are in >>>>> sync between the two neighbors. (This helps even if the >>>>> age field is ignored when selecting the ranges.) >>>>> >>>>> For example, suppose there are 100 LSAs and each LSA has a 2% >>>>> chance of not being in sync. Then the probability that all >>>>> LSAs are in sync is .98^100 = .1326, too low to result in >>>>> a significant overhead reduction. >>>>> Now suppose the LSAs are divided into 10 ranges, each with 10 LSAs. >>>>> The probability that all LSAs in a given range are in sync is then >>>>> .98^10 = .8171, resulting in a big overhead reduction on average. >>>>> >>>>> The selected range for each LSA type can depend on how frequently >>>>> new LSAs are originated for that type, with a larger range chosen >>>>> if LSAs are originated less frequently (so that the probability >>>>> of being out of sync is lower). >>>>> >>>>> This idea can be combined with the database exchange optimization >>>>> of RFC 5243 to further reduce overhead. I.e., if the checksums >>>>> do not match, then RFC 5243 can be used to reduce the number of >>>>> LSA headers exchanged in DD packets by up to 50%. >>>>> Also, if the checksums do match for a given range of LSAs, then the >>>>> router receiving the checksum need not send the same checksum for >>>>> the same range of LSAs. >>>>> >>>>> Richard >>>>> >>>>> >>>>> Anton Smirnov wrote: >>>>> >>>>> Mitchell, >>>>> LSDBs may be very close but they are seldom EXACTLY the >>>>> same. The >>>>> thing is that change in adjacency triggers LSA [re]build, so >>>>> depending >>>>> on timing router and/or network LSAs may change by the time >>>>> adjacency >>>>> restores. >>>>> Also, in moderately sized modern OSPF network there is >>>>> constant >>>>> background noise of LSAs being refreshed. So again, chances >>>>> that >>>>> refreshed LSA arrived to one side but not another while >>>>> adjacency is >>>>> down are relatively high. >>>>> These considerations make precomputed checksums over large >>>>> number of >>>>> LSAs rather bad for the purpose. >>>>> If you want to explore venue of checksumming database >>>>> exchange then I >>>>> recommend to look for more flexible approach giving routers >>>>> possibility >>>>> to choose range of LSAs to compute checksum over. For >>>>> example, if both >>>>> routers support the checksum exchange then router sending DBD >>>>> could >>>>> specify the first, the last LSA and value of some kind of >>>>> hash of LSA >>>>> headers between them (LSAs in LSDB can be viewed as ordered >>>>> sequence as >>>>> per RFC 5243). In this way router preparing DBD packet may >>>>> place start >>>>> and end points of checksumming to avoid including into the >>>>> range LSAs >>>>> with ages around 0 and MAXAGE (because these are LSAs which >>>>> are less >>>>> likely to match between routers). So database exchange will >>>>> consist of >>>>> series of single LSA headers (the 'old' way) and large LSA >>>>> sequences. >>>>> >>>>> Anton >>>>> >>>>> >>>>> >>>>> Mitchell Erblich wrote: >>>>> >>>>> >>>>> Group, >>>>> >>>>> General thoughts... >>>>> >>>>> After a conversation about OSPF's standard database >>>>> exchange and >>>>> how inefficient it >>>>> is when two routers have large LSDBs and them >>>>> almost being the >>>>> same, we concluded >>>>> that the use of Checksums might significantly >>>>> decrease the >>>>> amount of overhead and >>>>> achieve faster convergence. >>>>> >>>>> Moy first (I think) suggested the LSDB checksum >>>>> support to >>>>> identify identical >>>>> LSDBs. Sometime later the MIB supported the >>>>> ospfExternLsaChksumSum, >>>>> ospfAreaLsaChksumSum, and ospfLsdbChecksum >>>>> fields. >>>>> >>>>> Thus, why not support the ability to have OSPF >>>>> routers first >>>>> exchange >>>>> checksums with LSA counts. If the LSDBs have >>>>> more than XYZ >>>>> entries >>>>> in their LSDBs, then the counts which are above >>>>> specific >>>>> thresholds then >>>>> support the checksum on each X count of LSA >>>>> type. Only the >>>>> groups >>>>> with non-matching Checksums would then go thru >>>>> database >>>>> exchange. >>>>> >>>>> Thus, before database exchange, support some >>>>> form of triple >>>>> that >>>>> as routers have almost the same LSDBs, the >>>>> checksum exchange >>>>> would remove unnecessary LSA header exchange. >>>>> >>>>> This, IMO, would also be a big win if say a >>>>> DRother was >>>>> adjacent, lost >>>>> adjacency, and re-exstablished it a short time >>>>> later while >>>>> keeping a >>>>> past snapshot. It could remove its own >>>>> originated LSAs and >>>>> compare >>>>> checksums.. >>>>> >>>>> In this scenario, within BMA areas, I think >>>>> only DR / >>>>> DRother and BDR/ DRother >>>>> need to be making this Checksum exchange.. >>>>> >>>>> Wouldn't the exchange of this data benefit each >>>>> router >>>>> receiving the >>>>> information and allow both routers to know the >>>>> number of >>>>> LSAs to >>>>> be possibly exchanged? >>>>> Yes, a new capability would probably >>>>> have to be added. >>>>> >>>>> >>>>> So, the first question is whether the checksums >>>>> are unique >>>>> enough with a >>>>> count and type to determine whether that >>>>> section of the LSDB >>>>> matches >>>>> between two routers? >>>>> >>>>> And then what savings would normally exist in >>>>> this hybrid >>>>> database exchange >>>>> and is it worth the effort? >>>>> >>>>> Mitchell Erblich >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> OSPF mailing list >>>>> OSPF@ietf.org >>>>> https://www.ietf.org/mailman/listinfo/ospf >>>>> >>>>> >>>>> _______________________________________________ >>>>> OSPF mailing list >>>>> OSPF@ietf.org >>>>> https://www.ietf.org/mailman/listinfo/ospf >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> OSPF mailing list >>>>> OSPF@ietf.org >>>>> https://www.ietf.org/mailman/listinfo/ospf >>>>> >>>>> >>>>> _______________________________________________ >>>>> OSPF mailing list >>>>> OSPF@ietf.org >>>>> https://www.ietf.org/mailman/listinfo/ospf >>>> >>>> _______________________________________________ >>>> OSPF mailing list >>>> OSPF@ietf.org >>>> https://www.ietf.org/mailman/listinfo/ospf >>> >>> >>> ------------------------------------------------------------------------ >>> >>> _______________________________________________ >>> OSPF mailing list >>> OSPF@ietf.org >>> https://www.ietf.org/mailman/listinfo/ospf >>