| TOC |
|
This Internet-Draft is submitted to IETF in full conformance with the provisions of BCP 78 and BCP 79. This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as “work in progress.”
The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt.
The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html.
This Internet-Draft will expire on September 6, 2009.
Copyright (c) 2009 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents in effect on the date of publication of this document (http://trustee.ietf.org/license-info). Please review these documents carefully, as they describe your rights and restrictions with respect to this document.
The objective of this document is to specify the Best Current Practice (BCP) adopted by a service provider or other organization providing multimedia communication services such as Voice over Internet Protocol(VoIP) in order to locate another service provider to peer with in the context of Session PEERing for Multimedia INTerconnect. This document attempts to fill the gaps in information from RFC 3261, RFC 3263, and other documents, in order to more assist service providers in more easily performing SIP peering.
[EDITORIAL NOTE: XREF ERROR GENERATED IN ABSTRACT, XREFS REMOVED]
1.
Requirements Language
2.
Introduction
3.
Session Peering Setup
3.1.
Target Determination
3.2.
NAPTR Lookup
3.3.
SRV Lookup
3.4.
Using SRV Results
4.
High Availability
4.1.
SBE1 Fails to Reach SBE2
4.2.
Using SRV Results
5.
Caching/TTL
5.1.
Caching
5.2.
TTL
6.
Security Considerations
7.
IANA Considerations
8.
Acknowledgements
9.
References
9.1.
Normative References
9.2.
Informative References
Appendix A.
Document Change Log
Appendix B.
Open Issues
§
Authors' Addresses
| TOC |
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 (Bradner, S., “Key words for use in RFCs to Indicate Requirement Levels,” March 1997.) [RFC2119].
| TOC |
A service provider needs to identify the ingress Session Initiation Protocol (SIP) (RFC 3261 (Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, A., Peterson, J., Sparks, R., Handley, M., and E. Schooler, “SIP: Session Initiation Protocol,” June 2002.) [RFC3261]) server of a peering network before it can signal and route SIP-based real-time communications sessions. This function of locating the ingress SIP server of a peering network is typically performed by the egress SIP server of the service provider originating the SIP session. Also, the ingress server in the peering network needs to locate the originating service provider's egress server in situations where the peering connection is terminated after receiving a SIP request or if the egress SIP server of originating service provider fails. The SIP servers at the originating and peering sides use DNS procedures, using both SRV [RFC3261] (Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, A., Peterson, J., Sparks, R., Handley, M., and E. Schooler, “SIP: Session Initiation Protocol,” June 2002.) and NAPTR (Mealling, M., “Dynamic Delegation Discovery System (DDDS) Part Four: The Uniform Resource Identifiers (URI),” October 2002.) [RFC3404] records, in order to locate each other. It should be noted that
| TOC |
SIP systems are represented by user agents (UA). The diagram below shows the case of Direct Peering where a user agent (UA1), hosted by a service provider (SP1), initiates a SIP session to a User Agent (UA2), hosted by another service provider (SP2). The egress SIP server of SP1 is a SIP Signaling Path Border Element (SBE) as defined in Section 3 of [SPEERMINT‑Terminology] (Malas, D. and D. Meyer, “SPEERMINT Terminology,” November 2008.), called SBE1, that interfaces with session peering service provider SP2. The SIP session initiated by UA1 is received by this network element, SBE1. The resource to which the SIP request needs to be routed by SBE1 is identified by a SIP or SIPS URI. For example, this could be the SIP URI of UA2 found in the Request-URI of the SIP request received by SBE1. Alternatively, for example, this could also be the next hop from SBE1 found in the topmost Route header of SIP request. In order to determine the resource to route the request to, SP1 MAY make use of ENUM [RFC3761] (Faltstrom, P. and M. Mealling, “The E.164 to Uniform Resource Identifiers (URI) Dynamic Delegation Discovery System (DDDS) Application (ENUM),” April 2004.) lookup services or some other internal lookup to determine the SIP URI of the resource. Such an ENUM lookup service may use e164.arpa or one or more other private ENUM zones. This lookup MAY be performed by SBE1 or another network element of SP1.
............................ ............................. . +------+ . . +------+ . . | | . . | | . . | SBE 1|--------------| SBE 2| . . / | | . . | | \ . . / +------+ . . +------+ \ . . +------+ / || . . || \ +------+ . . | | / || . . || \ | | . . | UA 1 | || . . || | UA 2 | . . | | || . . || | | . . +------+ || . . || +------+ . . +-------+ . . +-------+ . . | | . . | | . . | DNS 1 | . . | DNS 2 | . . | | . . | | . . +-------+ . . +-------+ . . . . . . SP 1 . . SP 2 . ............................ .............................
| Figure 1 |
In order to route the SIP request to this resource in SP2, SBE1 needs to determine the ingress SIP signaling path border element for SP2, called SBE2, by resolving the SIP or SIPS URI using DNS. SBE1 makes use of the NAPTR and DNS SRV mechanism defined in [RFC3263] (Rosenberg, J. and H. Schulzrinne, “Session Initiation Protocol (SIP): Locating SIP Servers,” June 2002.) to determine the IP address, port, and transport protocol for peering with the SP2 ingress SIP proxy server (i.e. SBE2). SBE1 and SBE2 which are involved in the session peering, support a set of protocols and have list of preferences for these protocols. UDP, TCP and TLS MUST be supported by these proxies [EDITORIAL NOTE: QUESTION FOR WORKING GROUP ON WHAT SHOULD BE A MUST HERE - JUST TLS OR ALL?].
[EDITORIAL NOTE: ALEX SUGGESTS ADDING A REFERENCE TO TLS IN FIRST USE ABOVE]
As a best current practice, SBE1 and SBE2 SHOULD be deployed in a highly scalable and highly available manner, such as a cluster of servers. These servers are of different prioritization and weight, to ensure capacity-based load balancing. [EDITORIAL NOTE: CONSIDER REWORDING THIS LAST SENTENCE]
The figure below shows the case of Indirect Peering where SBE2 is the ingress SIP server of a transit service provider. The mechanism to locate SBE2 is the same as described for Direct Peering scenario.
............................ .............. . . . +------+ . . +------+ . . . . | | . . | | . . . . | SBE 1|---------| SBE 2| . . . . / | | . . | | . . . . / +------+ . . +------+ . . . . +------+ / || . . || . . +------+ . . | | / || . . || . . | | . . | UA 1 | || . . || . . | UA 2 | . . | | || . . || . . | | . . +------+ || . . || . . +------+ . . +-------+ . . +-------+ . . . . | | . . | | . . . . | DNS 1 | . . | DNS 2 | . . . . | | . . | | . . . . +-------+ . . +-------+ . . . . . . . . . . SP 1 . . Transit SP . . SP 2 . ............................ .............. ...............
| Figure 2 |
The figure below shows a high level SIP call flow setting up a Direct Peering session between SP1 and SP2. In this call flow a VoIP session is established between a caller, Bob (sip:bob@example1.com), in SP1 and callee, Alice(sip:alice@example2.com), in SP2 using SIP INVITE request. All SIP signaling MUST go through both SBE1 and SBE2, as these are the ingress and egress points in SP1 and SP2 networks, respectively.
Bob(UA 1) SBE 1 DNS 1 DNS 2 SBE 2 Alice(UA 2)
| | | | | |
|INVITE| | | | |
|----->| | | | |
| NAPTR Query | | |
| |----->| | | |
| NAPTR Response | | |
| |<-----| | | |
| SRV Query | | |
| |----->| | | |
| SRV Response | | |
| |<-----| | | |
| A Query | | |
| |----->| | | |
| A Response | | |
| |<-----| | | |
| | INVITE | |
| |------------------->| |
| | | | |INVITE|
| | | | |----->|
| | | | |200 OK|
| | | | |<-----|
| | 200 OK | |
| |<-------------------| |
|200 OK| | | | |
|<-----| | | | |
| ACK | | | | |
|----->| | | | |
| | ACK | |
| |------------------->| |
| | | | | ACK |
| | | | |----->|
| 2-Way Media |
|<================================>|
| | | | | |
| | | | | |
| Figure 3 |
[EDITORIAL NOTE: WG Q - DO WE NEED A CALL FLOW FOR Indirect Peering?]
[EDITORIAL NOTE: DO WE NEED A SUB-SECTION NUMBER PRIOR TO NEXT SENTENCE? ALSO, DO WE NEED TO EXPAND IT WITH MORE TEXT?]
The target, to which the request is sent, is determined by SBE1 as follows:
| TOC |
The target resource is identified with a SIP or SIPS URI. This is the URI in the Route header, if present, or the URI from the request URI of the SIP request received by SBE1. The host value of the hostport component of the URI is the TARGET [EDITORIAL NOTE: VERIFY / CLARIFY AS NEEDED]. This TARGET is the domain to be contacted. The NAPTR/SRV/A resource record lookup as described in the following section should be skipped if the transport/port/IP address is already specified for the target URI.
| TOC |
Next, SBE1 determines the transport protocol of the TARGET, SBE2, by performing a NAPTR query for the TARGET. NAPTR processing as described in [RFC2915] (Mealling, M. and R. Daniel, “The Naming Authority Pointer (NAPTR) DNS Resource Record,” September 2000.) will result in the discovery of the most preferred transport protocol [EDITORIAL NOTE: CONSIDER REWORDING / VALIDATE PREVIOUS 3 WORDS] of a server instance of SBE2 and SRV records.
Considering our example call flow above [EDITORIAL NOTE: INCL LOCAL REFERENCE HERE], SBE1 wishes to resolve sip:alice@example.com and performs a NAPTR query for that TARGET domain example.com, and the following NAPTR records are returned:
; order pref flags service regexp replacement IN NAPTR 50 50 "s" "SIPS+D2T" "" _sips._tcp.example.com IN NAPTR 90 50 "s" "SIP+D2T" "" _sip._tcp.example.com IN NAPTR 100 50 "s" "SIP+D2U" "" _sip._udp.example.com
| Figure 4 |
DNS MUST return at least three records - one with "SIP+D2T", one with "SIP+D2U" and one with "SIPS+D2T" service type for the case of Direct and Indirect Peering (section 4.3 in [SPEERMINT‑Terminology] (Malas, D. and D. Meyer, “SPEERMINT Terminology,” November 2008.)). For Indirect Peering (section 4.4 in [SPEERMINT‑Terminology] (Malas, D. and D. Meyer, “SPEERMINT Terminology,” November 2008.)) since domain validation as specified in Section 26.3.2.2 of [RFC3261] (Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, A., Peterson, J., Sparks, R., Handley, M., and E. Schooler, “SIP: Session Initiation Protocol,” June 2002.) for TLS at layer 5 will not work, SIPS over TLS cannot be used. [EDITORIAL NOTE: J.ELWELL CONCERN RAISED - IF THIS IS CASE, FEELS WE DO NOT HAVE A SECURE SOLUTION. WG FEEDBACK?]
| TOC |
Depending on what transport protocols SBE1 supports, SBE1 selects one from the preference list of NAPTR results and performs the SRV lookup to obtain a list of available server instances for SBE2. TLS SHOULD be the preferred transport protocol for peering between SBE1 and SBE2.
In our example, SBE1 uses TCP, the SRV lookup for _sip._tcp.example.com would return this list of available servers :
;; Priority Weight Port Target IN SRV 0 1 5060 server1.example.com IN SRV 0 2 5060 server2.example.com
| Figure 5 |
Alternatively, if no NAPTR records are found, then SBE1 uses the preferred transport protocol and issues an SRV query for that specific transport using "sips" for SIPS URI and SIP URI with TLS and "sip" for SIP URI as the SRV domain prefix.
In our example, SBE1 prefers to use TCP and target SIP URI of SP2 is sip:alice@example2.com, it sends a SRV query for _sip._tcp.example2.com.
The SRV responses MAY also include A and/or AAAA records with it.
| TOC |
If A records or AAAA records are not returned with the SRV responses, procedures from [RFC2782] (Gulbrandsen, A., Vixie, P., and L. Esibov, “A DNS RR for specifying the location of services (DNS SRV),” February 2000.) describes how to use and interpret the results obtained from the SRV query. The target entry of the SRV RRs is looked up by querying the DNS for address records. If the SRV response from DNS includes A or AAAA records with it, it will cut down on round trips and lookup of DNS again for target entry. On determining the transport protocol, service, port and address record from the SRV RRs as described above, the SBE1 will try to connect to the (protocol, address, service). Once the connection is established to an available instance of SBE2, SBE1 sends the SIP request to SBE2. SBE1 MUST act in a stateful manner and any retransmission of SIP requests for a specific SIP transaction, including ACKS for non-2xx response or CANCEL for that SIP transaction MUST go to the same server instance of SBE2.
When SBE1 sends the SIP request to SBE2, it SHOULD set the sent-by parameter of the topmost Via header in the SIP request to a domain that identifies SBE1. It MUST NOT specify the port.
[EDITORIAL NOTE: SHOULD THE ABOVE SENTENCE SAY MUST NOT OR SOMETHING ELSE?]
[EDITORIAL NOTE: ALEX SUGGESTS HAVING A SIP EXPERT REVIEW 4.2 ON SENT-BY]
[EDITORIAL NOTE: ALEX SUGGESTS A CALL FLOW FOR 4.2 - DOES WG AGREE?]
| TOC |
High Availability is ensured by detecting failures in the ability to connect to SBE1 and SBE2 server instances. In the event of a failure, when SBE1 tries to send SIP INVITE to SBE2, the following failures could occur:
| TOC |
A 503 error response is reported by the transaction layer, or failure can occur at the transport layer due to TCP disconnect in connection, ICMP error in UDP or time out at transport layer or SIP layer timeout when its not receiving any SIP response. In such situations, SBE1 tries a new SIP request transaction to the next available server instance of SBE2 as determined by SRV RRs entry. The SIP T1 timer on SBE1 SHOULD be configurable with a upper limit value of 500ms. A shorter value of T1, say 100ms, reflects a faster fail-over support.
[EDITORIAL NOTE: ALEX FINDS FIRST SENTENCE ABOVE CONFUSING - CONSIDER REWORDING IT]
[EDITORIAL NOTE: ALEX SUGGESTS ADDING A REF TO PROPER SECTION OF SRV DOC, AS WELL AS REF FOR SIP TIMERS]
| TOC |
Failure may also occur after the request is received by SBE2 from SBE1 due to closure of the transport connection the request came in on at SBE2, before the response can be sent back to SBE1. In this situation, SBE2 uses the domain value present in the 'sent-by' parameter in the top most Via header of the received SIP INVITE, and queries for SRV records at this domain name using the service identifier "_sips" if the Via transport is "TLS", "_sip" otherwise. The sorted list of SRV RRs are obtained and used as described in [RFC2782] (Gulbrandsen, A., Vixie, P., and L. Esibov, “A DNS RR for specifying the location of services (DNS SRV),” February 2000.) to send the response back to SBE1. If the topmost element in the list of server instances of SBE1 fails, the next available one is tried.
[EDITORIAL NOTE: FOR NEXT REV - SHOULD WE ADD CALL FLOW FOR FAILURE SCENARIOS DESCRIBED IN 4.1 AND 4.2?]
[EDITORIAL NOTE: ALEX RECOMMENDS WE CLARIFY WHETHER THE DESCRIBED FUNCTIONALITY HERE IS 'ON TOP' OF STANDARD SIP OR NEW FUNCTIONALITY, AND THAT A REF TO THE PROPER SECTION OF RFC3261 IS CONSIDERED]
| TOC |
| TOC |
SBE SHOULD use caching of DNS results to eliminate unnecessary DNS queries.
| TOC |
SRV RRs have a TTL value based on which the SBE1 caches the entry for that duration, if it supports caching, and any further requests to the same TARGET domain are delivered to the cached server instance. The TTL recommended for SRV is about 1 hr. The TTL for NAPTR is much higher, about 1 day (24hrs) since the NAPTR records do not vary that often as compared to SRV.
| TOC |
This document introduces no new security considerations.
| TOC |
There are no IANA considerations in this document.
| TOC |
Special thanks go to Yiu Lee for his valuable input to this document, as well as John Elwell, Alexander Mayrhofer, and Chris Griffiths for their detailed reviews of this document.
| TOC |
| TOC |
| TOC |
| [RFC2434] | Narten, T. and H. Alvestrand, “Guidelines for Writing an IANA Considerations Section in RFCs,” BCP 26, RFC 2434, October 1998 (TXT, HTML, XML). |
| [RFC3761] | Faltstrom, P. and M. Mealling, “The E.164 to Uniform Resource Identifiers (URI) Dynamic Delegation Discovery System (DDDS) Application (ENUM),” RFC 3761, April 2004 (TXT). |
| TOC |
[RFC Editor: This section is to be removed before publication]
draft-ietf-speermint-srv-naptr-use-05:
draft-ietf-speermint-srv-naptr-use-04:
draft-ietf-speermint-srv-naptr-use-03:
| TOC |
Decide what we want to do in Using SRV Results section
Benny Rodrig suggests adding some more description of how this works in the indirect case
Several open issues with John Elwell
Several open issues with Alex Mayrhofer
Several editorial notes to close out
| TOC |
| Tom Creighton | |
| Comcast Cable Communications | |
| One Comcast Center | |
| 1701 John F. Kennedy Boulevard | |
| Philadelphia, PA 19103 | |
| US | |
| Email: | tom_creighton@cable.comcast.com |
| URI: | http://www.comcast.com |
| Jason Livingood | |
| Comcast Cable Communications | |
| One Comcast Center | |
| 1701 John F. Kennedy Boulevard | |
| Philadelphia, PA 19103 | |
| US | |
| Email: | jason_livingood@cable.comcast.com |
| URI: | http://www.comcast.com |