]> Synchronizing Location-to-Service Translation (LoST) Protocol based Service Boundaries and Mapping Elements Columbia University
Department of Computer Science 450 Computer Science Building New York NY 10027 US +1 212 939 7004 hgs+ecrit@cs.columbia.edu http://www.cs.columbia.edu
Nokia Siemens Networks
Linnoitustie 6 Espoo 02600 Finland +358 (50) 4871445 Hannes.Tschofenig@gmx.net http://www.tschofenig.priv.at
RAI ECRIT Internet-Draft Location The Location-to-Service Translation (LoST) protocol is an XML-based protocol for mapping service identifiers and geodetic or civic location information to service URIs and service boundaries. In particular, it can be used to determine the location-appropriate Public Safety Answering Point (PSAP) for emergency services. The main data structure, the <mapping> element, used for encapsulating information about service boundaries is defined in the LoST protocol specification and circumscribes the region within which all locations map to the same service Uniform Resource Identifier (URI) or set of URIs for a given service. This document defines an XML protocol to exchange these mappings between two nodes. This mechanism can be used for bulk exchange of <mapping> elements between two entities. As such, this document can also be used without the LoST protocol.
The LoST (Location-to-Service Translation) protocol maps service identifiers and geodetic or civic location information to service URIs. The main data structure, the <mapping> element, used for encapsulating information about service boundaries is defined in the LoST protocol specification and circumscribes the region within which all locations map to the same service Uniform Resource Identifier (URI) or set of URIs for a given service. This document enables a bulk exchange of <mapping> elements between two entities (the LoST Sync client and the LoST Sync server). The LoST Sync mechanism can, for example, be used in the LoST architecture, as specified in the . There, LoST servers act in different roles that cooperate to provide an ubiquitous, globally scalable and resilient mapping service. In the LoST mapping architecture, servers can peer, i.e., have an on-going data exchange relationship. Peering relationships are set up manually, based on local policies. A server can peer with any number of other servers. Forest guides peer with other forest guides; resolvers peer with forest guides and other resolvers (in the same cluster); authoritative mapping servers peer with forest guides and other authoritative servers, either in the same cluster or above or below them in the tree. Authoritative mapping servers push coverage regions "up" the tree, i.e., from child nodes to parent nodes. The child informs the parent of the geospatial or civic region that it covers for a specific service. This document defines two types of exchanges and those are best described by the exchange between two nodes as shown in and . The protocol exchange always runs between a LoST Sync client and a LoST Sync server even through the roles are reversed for the two available exchanges and logically the two nodes might often be peers rather than in a client-server relationship. Node A in the example of and has mappings that Node B is going to retrieve. The <getMappingsRequest> and <getMappingsResponse> exchange allows a LoST Sync client to request mappings from a LoST Sync server.
| |----------------------------->| | | | | |<-----------------------------| | | | | | | ]]>
The <pushMappingsRequest> and <pushMappingsResponse> exchange allows a LoST Sync client to push mappings to LoST Sync server. The assumption is being made that Node A and B have previously been configured in a way that they push mappings in such a fashion and that Node A maintains state about the mappings that have to be pushed to Node B. No subscribe mechanism is defined in this document that would allow Node B to tell Node A about what mappings it is interested.
| |----------------------------->| | | | | |<-----------------------------| | | | | | | ]]>
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 . This document reuses terminology introduced by the mapping architecture document. Throughout this document we use the term LoST Sync client and LoST Sync server to denote the protocol end points of the exchange. The protocol is referred as LoST Sync within the text.
A LoST Sync client has two ways to retrieve mapping elements from a LoST Sync server node. A mechanisms that is suitable when no mappings are available on the client side is to submit an empty <getMappingsRequest> message, as shown in . The intent by the LoST Sync client thereby is to retrieve all mappings from the LoST Sync server. Note that the request does not propagate further to other nodes. A client that has already obtained mappings in previous exchanges may want to check whether these mappings have been updated in the meanwhile. The policy when to poll for updated mapping information is outside the scope of this document. The <getMappingsRequest> message with one or multiple <exists> child element(s) allows to reduce the number of returned mappings to those that have been updated and also to those that are missing. In response to the <getMappingsRequest> message the client waits for the <getMappingsResponse> message. In case of a successful response the client stores the received mappings and determines which mappings to replace.
When a LoST Sync server receives an empty <getMappingsRequest> message then all locally available mappings MUST be returned (assuming that the client has been properly authenticated and authorized). When a LoST Sync server receives a <getMappingsRequest> message with one or multiple <exists> child element(s) then it MUST consult with the local mapping database to determine whether any of the mappings of the client is stale and whether there are mappings locally that the client does not yet have. The former can be determined by finding mappings corresponding to the 'source' and 'sourceID' attribut where a mapping with a more recent lastUpdated date exists. Processing a <getMappingsRequest> message MAY lead to a successful response in the form of a <getMappingsResponse> or an <errors> message. Only the <badRequest>, <forbidden>, <internalError>, <serverTimeout> errors, defined in , are utilized by this specification. Neither the <redirect> nor the <warnings> messages are reused by this message.
The first examples show the simplest <getMappingsRequest> message.
]]>
An further example request is shown in , the corresponding response in . In this example a LoST node requests a specific mapping for source="authoritative.bar.example" and sourceId="7e3f40b098c711dbb6060800200c9a66" that is fresher than "2006-11-01T01:00:00Z".
]]>
The response is shown in . A more recent mapping was available with the identification of source="authoritative.bar.example" and sourceId="7e3f40b098c711dbb6060800200c9a66". Only one mapping that matched source="authoritative.foo.example" was found and returned.
Leonia Police Department urn:service:sos.police US NJ Leonia 07605 sip:police@leonianj2.example.org 911 New York City Police Department urn:service:sos.police 37.775 -122.4194 37.555 -122.4194 37.555 -122.4264 37.775 -122.4264 37.775 -122.4194 sip:nypd@example.com xmpp:nypd@example.com 911 ]]>
When a LoST Sync node obtains new information that is of interest to its peers, it MAY push the new mappings to its peers. Configuration settings at both peers decide whether this functionality is used. New mappings may arrive through non-LoST means, such as a manual addition to the local mappings database, or through the interaction with other LoST nodes. Mappings may also be deleted and this may trigger events. A sending node keeps track with which recipient it has exchanged mapping elements with. As discussed in Section 5.1 of , mapping elements are identified by the 'source', 'sourceID' and 'lastUpdated' attributes. A mapping is considered the same if these three attributes match. It is RECOMMENDED not to push the same information to the same peer more than once. A LoST Sync client MUST send a <pushMappings> request containing one or more <mapping> elements. To delete a mapping, the content of the mapping is left empty. The node can delete the mapping from its internal mapping database, but has to remember which peers it has distributed this update to. The 'expires' attribute is required, but ignored. If an attempt is made to delete a non-existent mapping, the request is silently ignored.
When a LoST Sync Server receives a <pushMappingsRequest> message then a newly received mapping M' MUST replace an existing mapping M if all of the following conditions hold: M'.source equals M.source M'.sourceID' equals M.sourceID M'.lastUpdated greater than M.lastUpdated If the received mapping M' does not update any existing mapping M then it MUST be added to the local cache as an independent mapping. If a <pushMappingsRequest> message with an empty <mapping> element is received then a corresponding mapping has to be determined based on the 'source', 'sourceID' and 'lastUpdated' attributes. If a mapping has been found then it MUST be deleted. If no mapping can be identified then an <errors> response MUST be returned that contains the <notDeleted> child element. The <notDeleted> element MAY carry a <message> element and MUST contain the <mapping> element(s) that caused the error. The response to a <pushMappingsRequest> request is a <pushMappingsResponse> message. With this specification, a successful response message returns no additional elements, whereas an <errors> response is returned in the response message, if the request failed. Only the <badRequest>, <forbidden>, <internalError> or <serverTimeout> errors defined in Section 13.1 of , are used. The <redirect> and <warnings> messages are not used for this query/response. If the set of nodes that are synchronizing their data does not form a tree, it is possible that the same information arrives through several other nodes. This is unavoidable, but generally only imposes a modest overhead. (It would be possible to create a spanning tree in the same fashion as IP multicast, but the complexity does not seem warranted, given the relatively low volume of data.)
An example is shown in . Image a LoST node that obtained two new mappings identified as follows: source="authoritative.example" sourceId="7e3f40b098c711dbb6060800200c9a66" lastUpdated="2008-11-26T01:00:00Z" source="authoritative.example" sourceId="7e3f40b098c711dbb606011111111111" lastUpdated="2008-11-01T01:00:00Z" These two mappings have to be added to the peer's mapping database. Additionally, the following mapping has to be deleted: source="nj.us.example" sourceId="123" lastUpdated="2008-11-01T01:00:00Z"
Leonia Police Department urn:service:sos.police US NJ Leonia 07605 sip:police@leonianj.example.org 911 New York City Police Department urn:service:sos.police 37.775 -122.4194 37.555 -122.4194 37.555 -122.4264 37.775 -122.4264 37.775 -122.4194 sip:nypd@example.com xmpp:nypd@example.com 911 ]]>
In response, the peer performs the necessary operation and updates its mapping database. In particular, it will check whether the other peer is authorized to perform the update and whether the elements and attributes contain values that it understands. In our example, a positive response is returned as shown in .
]]>
In case that a mapping could not be deleted as requested the following error response might be returned instead.
]]>
LoST Sync needs an underlying protocol transport mechanism to carry requests and responses. This document defines the use of LoST Sync over HTTP and LoST over HTTP-over-TLS. Client and server developers are reminded that full support of RFC 2616 HTTP facilities is expected. If LoST Sync clients or servers re-implement HTTP, rather than using available servers or client code as a base, careful attention must be paid to full interoperability. Other transport mechanisms are left to future documents. The available transport mechanisms are determined through the use of the LoST U-NAPTR application. In protocols that support content type indication, LoST Sync uses the media type application/lostsync+xml. When using HTTP and HTTP-over-TLS , LoST Sync messages use the HTTP POST method. The HTTP request MUST use the Cache-Control response directive "no-cache" to HTTP-level caching even by caches that have been configured to return stale responses to client requests. All LoST Sync responses, including those indicating a LoST warning or error, are carried in 2xx responses, typically 200 (OK). Other 2xx responses, in particular 203 (Non-authoritative information) may be returned by HTTP caches that disregard the caching instructions. 3xx, 4xx and 5xx HTTP response codes indicates that the HTTP request itself failed or was redirected; these responses do not contain any LoST Sync XML elements. The HTTP URL is derived from the LoST Sync server name via U-NAPTR application.
Location-to-Service Translation (LoST) Synchronization Protocol ]]>
The LoST security considerations are discussed in . The operations described in this document involve mutually-trusting LoST nodes. These nodes need to authenticate each other, using mechanisms such as HTTP Digest, HTTP Basic over TLS or TLS client and server certificates.
This specification requests the registration of a new MIME type according to the procedures of RFC 4288 and guidelines in RFC 3023 . application lostsync+xml none charset Indicates the character encoding of enclosed XML. Uses XML, which can employ 8-bit characters, depending on the character encoding used. See RFC 3023 , Section 3.2. This content type is designed to carry LoST Syncronization protocol payloads. None RFCXXXX [NOTE TO IANA/RFC-EDITOR: Please replace XXXX with the RFC number of this specification.] Emergency and Location-based Systems None .lostsyncxml 'TEXT' Hannes Tschofenig, Hannes.Tschofenig@nsn.com LIMITED USE This specification is a work item of the IETF ECRIT working group, with mailing list address <ecrit@ietf.org>. The IESG <iesg@ietf.org>
urn:ietf:params:xml:schema:lostsync1 IETF ECRIT Working Group, Hannes Tschofenig (Hannes.Tschofenig@nsn.com). The Relax NG schema to be registered is contained in . Its first line is
and its last line is
urn:ietf:params:xml:ns:lostsync1 IETF ECRIT Working Group, Hannes Tschofenig (Hannes.Tschofenig@nsn.com).
LoST Synchronization Namespace

Namespace for LoST server synchronization

urn:ietf:params:xml:ns:lost1:sync

See RFCXXXX [NOTE TO IANA/RFC-EDITOR: Please replace XXXX with the RFC number of this specification.].

END ]]>
Robins George, Cullen Jennings, Karl Heinz Wolf, Richard Barnes and Andrew Newton provided helpful input. Jari Urpalainen assisted with the Relax NG schema. We would also like to thank our PROTO shepherd Roger Marshall for his help with the document.
&rfc2119; &rfc2617; &rfc5222; &rfc5246; &RFC3023; &RFC4288; &RFC2616; &RFC2818; &I-D.ietf-ecrit-mapping-arch;