Efficient RDAP Referrals

Introduction Many Registration Data Access Protocol (RDAP, described in , , , and others) resources contain referrals to related RDAP resources. For example, in the domain space, an RDAP record for a domain name received from the registry operator may include a referral to the RDAP record for the same object provided by the sponsoring registrar, while in the IP address space, an RDAP record for an address allocation may include referrals to enclosing or sibling prefixes. In both cases, RDAP service users are often equally if not more interested in these related RDAP resources than the resource provided by the TLD registry or RIR. While RDAP supports redirection of RDAP requests using HTTP redirections (which use a 3xx HTTP status and the "Location" header field, see Section 15.4 of ), it is not possible for RDAP servers to know a priori whether a client requesting an RDAP record is doing so because it wants to retrieve a related RDAP record, or its own, so it can only respond by providing the full RDAP response. The client must then parse that response in order to extract the relevant URL from the "links" property of the object. This results in the wasteful expenditure of time, compute resources and bandwidth on the part of both the client and server. This document describes how an RDAP server can use "Link" HTTP header fields in responses to HEAD and GET requests to provide RDAP clients with the URL of related RDAP records, without the need for a signalling mechanism for the client to tell the server that it is only interested in retrieving those URLs.

RDAP Link Objects RDAP link objects, described in Section 4.2 of , establish unidirectional relationships between an RDAP resource and other web resources, which may also be RDAP resources. The "rel" property indicates the nature of the relationship, and its possible values are described in . If a link object has a "type" property which contains the value "application/rdap+json", then clients can assume that the linked resource is also an RDAP resource. In the domain name space, this allows clients to discover the URL of the sponsoring registrar's RDAP record for a given domain name, if the "rel" property has the value "related", while in the IP address space, the "up" and "down" values allow RDAP clients to navigate the hierarchy of address space allocations.

HTTP "Link" Header Field "Link" header fields, described in Section 3 of , provide a means for describing a relationship between two resources, generally between the requested resource and some other resource. The "Link" header field is semantically equivalent to the <link> element in HTML, and multiple "Link" headers may be present in the header of an HTTP response. "Link" header fields may contain most of the parameters that are also present in Link objects in RDAP responses (See Section 4.2 of ). So for example, an RDAP link object which has the following JSON representation: may be represented as follows: ; rel="self"; hreflang="en,ch"; title="title"; media="screen"; type="application/json" ]]>

Registrar RDAP "Link" Header Following on from the above, the following RDAP link object (which represents the RDAP URL of the sponsoring registrar of a resource): may be represented as follows: ; title="URL of Sponsoring Registrar's RDAP Record"; rel="related"; type="application/rdap+json" ]]>

RDAP Responses In response to GET and HEAD RDAP requests, RDAP servers which implement this specification MUST include a "Link" header field for each link object which refers to an RDAP resource that is present in the "links" array of the object in question. The server MAY also include "Link" headers for link objects which refer to other types of resource. In all cases, the link attributes MUST be the same in both places.

RDAP HEAD requests The HTTP HEAD method can be used for obtaining metadata about a resource without transferring that resource (see Section 4.3.2 of ). An RDAP client which only wishes to obtain the URLs of related RDAP resources can issue a HEAD request for an RDAP resource and check the response for the presence of an appropriate "Link" header field. If the link is absent, it may then fall back to performing a GET request. An RDAP client interested in both the server's record and related records can use the traditional method of performing a GET request and extracting the link objects from the response. To improve performance, RDAP clients MAY inspect the header of a response, extract the link headers, and issue requests for the related record in parallel while the request to the server is still in flight. As an example, the cURL library provides the CURLOPT_HEADERFUNCTION configuration option to provide a callback that is invoked as soon as it has received header data.

RDAP Conformance Servers which implement this specification MUST include the string "link_headers" in the "rdapConformance" array in all RDAP responses.

IANA Considerations IANA is requested to register the following value in the RDAP Extensions Registry:

Extension identifier:: registrar_link_header
Registry operator:: any
Published specification:: this document
Contact:: IETF <iesg@ietf.org>
Intended usage:: this extension indicates that the server provides the URL of the registrar's RDAP record in a "Link" header in responses to RDAP queries.