Internet Engineering Task Force (IETF) C. Boulton Request for Comments: 6314 NS-Technologies Category: Informational J. Rosenberg ISSN: 2070-1721 Skype G. Camarillo Ericsson F. Audet Skype July 2011
NAT Traversal Practices for Client-Server SIP
Abstract
Traversal of the Session Initiation Protocol (SIP) and the sessions it establishes through Network Address Translators (NATs) is a complex problem. Currently, there are many deployment scenarios and traversal mechanisms for media traffic. This document provides concrete recommendations and a unified method for NAT traversal as well as documents corresponding flows.
Status of This Memo
This document is not an Internet Standards Track specification; it is published for informational purposes.
This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Not all documents approved by the IESG are a candidate for any level of Internet Standard; see Section 2 of RFC 5741.
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc6314.
Copyright Notice
Copyright (c) 2011 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 (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect
Boulton, et al. Informational [Page 1]
RFC 6314 NAT Scenarios July 2011
to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
NAT (Network Address Translator) traversal has long been identified as a complex problem when considered in the context of the Session Initiation Protocol (SIP) [RFC3261] and its associated media such as the Real-time Transport Protocol (RTP) [RFC3550]. The problem is exacerbated by the variety of NATs that are available in the marketplace today and the large number of potential deployment scenarios. Details of different NATs behavior can be found in "NAT Behavioral Requirements for Unicast UDP" [RFC4787].
The IETF has been active on many specifications for the traversal of NATs, including Session Traversal Utilities for NAT (STUN) [RFC5389], Interactive Connectivity Establishment (ICE) [RFC5245], symmetric response [RFC3581], symmetric RTP [RFC4961], Traversal Using Relay NAT (TURN) [RFC5766], SIP Outbound [RFC5626], the Session Description Protocol (SDP) attribute for RTP Control Protocol (RTCP) [RFC3605], "Multiplexing RTP Data and Control Packets on a Single Port" [RFC5761], and others. Each of these represents a part of the solution, but none of them gives the overall context for how the NAT traversal problem is decomposed and solved through this collection of specifications. This document serves to meet that need. It should be noted that this document intentionally does not invoke 'Best Current Practice' machinery as defined in RFC 2026 [RFC2026].
The document is split into two distinct sections as follows:
o Section 4 provides a definitive set of best common practices to demonstrate the traversal of SIP and its associated media through NAT devices.
o Section 5 provides non-normative examples representing interactions of SIP using various NAT type deployments.
The document does not propose any new functionality but does draw on existing solutions for both core SIP signaling and media traversal (as defined in Section 4).
The best practices described in this document are for traditional "client-server"-style SIP. This term refers to the traditional use of the SIP protocol where User Agents talk to a series of intermediaries on a path to connect to a remote User Agent. It seems likely that other groups using SIP, for example, peer-to-peer SIP (P2PSIP), will recommend these same practices between a P2PSIP client and a P2PSIP peer, but will recommend different practices for use between peers in a peer-to-peer network.
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 [RFC2119].
It should be noted that the use of the term 'Endpoint-Independent NAT' in this document refers to a NAT that is both Endpoint- Independent Filtering and Endpoint-Independent Mapping per the definitions in RFC 4787 [RFC4787].
The traversal of SIP through NATs can be split into two categories that both require attention: the core SIP signaling and associated media traversal. This document assumes NATs that do not contain SIP- aware Application Layer Gateways (ALGs), which makes much of the issues discussed in the document not applicable. ALGs have limitations (as per RFC 4787 [RFC4787] Section 7, RFC 3424 [RFC3424], and [RFC5245] Section 18.6), and experience shows they can have an adverse impact on the functionality of SIP. This includes problems such as requiring the media and signaling to traverse the same device and not working with encrypted signaling and/or payload.
The use of non-TURN-based media intermediaries is not considered in this document. More information can be obtained from [RFC5853] and [MIDDLEBOXES].
The core SIP signaling has a number of issues when traversing through NATs.
SIP response routing over UDP as defined in RFC 3261 [RFC3261] without extensions causes the response to be delivered to the source IP address specified in the topmost Via header, or the 'received' parameter of the topmost 'Via' header. The port is extracted from the SIP 'Via' header to complete the IP address/port combination for returning the SIP response. While the destination for the response is correct, the port contained in the SIP 'Via' header represents the listening port of the originating client and not the port representing the open pinhole on the NAT. This results in responses being sent back to the NAT but to a port that is likely not open for SIP traffic. The SIP response will then be dropped at the NAT. This is illustrated in Figure 1, which depicts a SIP response being returned to port 5060.
Secondly, there are two cases where new requests reuse existing connections. The first is when using a reliable, connection-oriented transport protocol such as TCP, SIP has an inherent mechanism that results in SIP responses reusing the connection that was created/used for the corresponding transactional request. The SIP protocol does not provide a mechanism that allows new requests generated in the reverse direction of the originating client to use, for example, the existing TCP connection created between the client and the server during registration. This results in the registered contact address not being bound to the "connection" in the case of TCP. Requests are then blocked at the NAT, as illustrated in Figure 2. The second case is when using an unreliable transport protocol such as UDP where external NAT mappings need to be reused to reach a SIP entity on the private side of the network.
In Figure 2, the original REGISTER request is sent from the client on port 8023 and received by the proxy on port 5060, establishing a connection and opening a pinhole in the NAT. The generation of a new request from the proxy results in a request destined for the registered entity (contact IP address) that is not reachable from the public network. This results in the new SIP request attempting to create a connection to a private network address. This problem would be solved if the original connection were reused. While this problem has been discussed in the context of connection-oriented protocols such as TCP, the problem exists for SIP signaling using any transport protocol. The impact of connection reuse of connection-oriented transports (TCP, TLS, etc.) is discussed in more detail in the connection reuse specification [RFC5923]. The approach proposed for this problem in Section 4 of this document is relevant for all SIP signaling in conjunction with connection reuse, regardless of the transport protocol.
NAT policy can dictate that connections should be closed after a period of inactivity. This period of inactivity may vary from a number of seconds to hours. SIP signaling cannot be relied upon to keep connections alive for the following two reasons. Firstly, SIP entities can sometimes have no signaling traffic for long periods of time, which has the potential to exceed the inactivity timer, and this can lead to problems where endpoints are not available to receive incoming requests as the connection has been closed. Secondly, if a low inactivity timer is specified, SIP signaling is not appropriate as a keep-alive mechanism as it has the potential to add a large amount of traffic to the network, which uses up valuable resources and also requires processing at a SIP stack, which is also a waste of processing resources.
Media associated with SIP calls also has problems traversing NAT. RTP [RFC3550] runs over UDP and is one of the most common media transport types used in SIP signaling. Negotiation of RTP occurs with a SIP session establishment using the Session Description Protocol (SDP) [RFC4566] and a SIP offer/answer exchange [RFC3264]. During a SIP offer/answer exchange, an IP address and port combination are specified by each client in a session as a means of receiving media such as RTP. The problem arises when a client advertises its address to receive media and it exists in a private network that is not accessible from outside the NAT. Figure 3 illustrates this problem.
The connection addresses of the clients behind the NATs will nominally contain a private IPv4 address that is not routable across the public Internet. Exacerbating matters even more would be the tendency of Client A to send media to a destination address it received in the signaling confirmation message -- an address that may actually correspond to a host within the private network who is suddenly faced with incoming RTP packets (likewise, Client B may send media to a host within its private network who did not solicit these packets). Finally, to complicate the problem even further, a number of different NAT topologies with different default behaviors increases the difficulty of arriving at a unified approach. This problem exists for all media transport protocols that might be NATted (e.g., TCP, UDP, the Stream Control Transmission Protocol (SCTP), the Datagram Congestion Control Protocol (DCCP)).
In general, the problems associated with NAT traversal can be categorized as follows.
For signaling:
o Responses do not reuse the NAT mapping and filtering entries created by the request.
o Inbound requests are filtered out by the NAT because there is no long-term connection between the client and the proxy.
Boulton, et al. Informational [Page 7]
RFC 6314 NAT Scenarios July 2011
For media:
o Each endpoint has a variety of addresses that can be used to reach it (e.g., native interface address, public NATted address). In different situations, a different pair of (local endpoint, remote endpoint) addresses should be used, and it is not clear when to use which pair.
o Many NATs filter inbound packets if the local endpoint has not recently sent an outbound packet to the sender.
o Classic RTCP usage is to run RTCP on the next highest port. However, NATs do not necessarily preserve port adjacency.
o Classic RTP and RTCP usage is to use different 5-tuples for traffic in each direction. Though not really a problem, doing this through NATs is more work than using the same 5-tuple in both directions.
As mentioned previously, the traversal of SIP through existing NATs can be divided into two discrete problem areas: getting the SIP signaling across NATs and enabling media as specified by SDP in a SIP offer/answer exchange to flow between endpoints.
SIP signaling has two areas that result in transactional failure when traversing through NATs, as described in Section 3 of this document. The remaining sub-sections describe appropriate solutions that result in SIP signaling traversal through NATs, regardless of transport protocol. It is advised that SIP-compliant entities follow the guidelines presented in this section to enable traversal of SIP signaling through NATs.
As described in Section 3 of this document, when using an unreliable transport protocol such as UDP, SIP responses are sent to the IP address and port combination contained in the SIP 'Via' header field (or default port for the appropriate transport protocol if not present). Figure 4 illustrates the response traversal through the open pinhole using Symmetric techniques defined in RFC 3581 [RFC3581].
The outgoing request from Client A opens a pinhole in the NAT. The SIP Proxy would normally respond to the port available in the SIP 'Via' header, as illustrated in Figure 1. The SIP Proxy honors the 'rport' parameter in the SIP 'Via' header and routes the response to the port from which it was sent. The exact functionality for this method of response traversal is called 'Symmetric Response', and the details are documented in RFC 3581 [RFC3581]. Additional requirements are imposed on SIP entities in RFC 3581 [RFC3581] such as listening and sending SIP requests/responses from the same port.
The second problem with SIP signaling, as defined in Section 3 and illustrated in Figure 2, is to allow incoming requests to be properly routed.
Guidelines for devices such as User Agents that can only generate outbound connections through NATs are documented in "Managing Client- Initiated Connections in the Session Initiation Protocol (SIP)" [RFC5626]. The document provides techniques that use a unique User Agent instance identifier (instance-id) in association with a flow identifier (reg-id). The combination of the two identifiers provides a key to a particular connection (both UDP and TCP) that is stored in association with registration bindings. On receiving an incoming request to a SIP Address-Of-Record (AOR), a proxy/registrar routes to the associated flow created by the registration and thus a route through NATs. It also provides a keep-alive mechanism for clients to keep NAT bindings alive. This is achieved by multiplexing a ping- pong mechanism over the SIP signaling connection (STUN for UDP and
Boulton, et al. Informational [Page 9]
RFC 6314 NAT Scenarios July 2011
CRLF/operating system keepalive for reliable transports like TCP). Usage of [RFC5626] is RECOMMENDED. This mechanism is not transport specific and should be used for any transport protocol.
Even if the SIP Outbound mechanism is not used, clients generating SIP requests SHOULD use the same IP address and port (i.e., socket) for both transmission and receipt of SIP messages. Doing so allows for the vast majority of industry provided solutions to properly function (e.g., NAT traversal that is Session Border Control (SBC) hosted). Deployments should also consider the mechanism described in the Connection Reuse [RFC5923] specification for routing bidirectional messages securely between trusted SIP Proxy servers.
The issues of media traversal through NATs is not straightforward and requires the combination of a number of traversal methodologies. The technologies outlined in the remainder of this section provide the required solution set.
The primary problem identified in Section 3 of this document is that internal IP address/port combinations cannot be reached from the public side of NATs. In the case of media such as RTP, this will result in no audio traversing NATs (as illustrated in Figure 3). To overcome this problem, a technique called 'Symmetric RTP/RTCP' [RFC4961] can be used. This involves a SIP endpoint both sending and receiving RTP/RTCP traffic from the same IP address/port combination. When operating behind a NAT and using the 'latching' technique described in [MIDDLEBOXES], SIP User Agents MUST implement Symmetric RTP/RTCP. This allows traversal of RTP across the NAT.
Normal practice when selecting a port for defining RTP Control Protocol (RTCP) [RFC3550] is for consecutive-order numbering (i.e., select an incremented port for RTCP from that used for RTP). This assumption causes RTCP traffic to break when traversing certain types of NATs due to various reasons (e.g., already allocated port, randomized port allocation). To combat this problem, a specific address and port need to be specified in the SDP rather than relying on such assumptions. RFC 3605 [RFC3605] defines an SDP attribute that is included to explicitly specify transport connection information for RTCP so a separate, explicit NAT binding can be set up for the purpose. The address details can be obtained using any appropriate method including those detailed in this section (e.g., STUN, TURN, ICE).
Boulton, et al. Informational [Page 10]
RFC 6314 NAT Scenarios July 2011
A further enhancement to RFC 3605 [RFC3605] is defined in [RFC5761], which specifies 'muxing' both RTP and RTCP on the same IP/PORT combination.
ICE, STUN, and TURN are a suite of 3 inter-related protocols that combine to provide a complete media traversal solution for NATs. The following sections provide details of each component part.
Session Traversal Utilities for NAT or STUN is defined in RFC 5389 [RFC5389]. STUN is a lightweight tool kit and protocol that provides details of the external IP address/port combination used by the NAT device to represent the internal entity on the public facing side of NATs. On learning of such an external representation, a client can use it accordingly as the connection address in SDP to provide NAT traversal. Using terminology defined in "NAT Behavioral Requirements for Unicast UDP" [RFC4787], STUN does work with Endpoint-Independent Mapping but does not work with either Address-Dependent Mapping or Address and Port-Dependent Mapping type NATs. Using STUN with either of the previous two NAT mappings to probe for the external IP address/port representation will provide a different result to that required for traversal by an alternative SIP entity. The IP address/ port combination deduced for the STUN server would be blocked for RTP packets from the remote SIP User Agent.
As mentioned in Section 4.1.2, STUN is also used as a client-to- server keep-alive mechanism to refresh NAT bindings.
As described in Section 4.2.3.1, the STUN protocol does not work for UDP traversal through certain identified NAT mappings. 'Traversal Using Relays around NAT' is a usage of the STUN protocol for deriving (from a TURN server) an address that will be used to relay packets towards a client. TURN provides an external address (globally routable) at a TURN server that will act as a media relay that attempts to allow traffic to reach the associated internal address. The full details of the TURN specification are defined in [RFC5766]. A TURN service will almost always provide media traffic to a SIP entity, but it is RECOMMENDED that this method would only be used as a last resort and not as a general mechanism for NAT traversal. This is because using TURN has high performance costs when relaying media traffic and can lead to unwanted latency.
Interactive Connectivity Establishment (ICE) is the RECOMMENDED method for traversal of existing NATs if Symmetric RTP and media latching are not sufficient. ICE is a methodology for using existing technologies such as STUN, TURN, and any other protocol compliant with Unilateral Self-Address Fixing (NSAF) [RFC3424] to provide a unified solution. This is achieved by obtaining as many representative IP address/port combinations as possible using technologies such as STUN/TURN (note: an ICE endpoint can also use other mechanisms (e.g., the NAT Port Mapping Protocol [NAT-PMP], Universal Plug and Play Internet Gateway Device [UPnP-IGD]) to learn public IP addresses and ports, and populate a=candidate lines with that information). Once the addresses are accumulated, they are all included in the SDP exchange in a new media attribute called 'candidate'. Each candidate SDP attribute entry has detailed connection information including a media address, priority, and transport protocol. The appropriate IP address/port combinations are used in the order specified by the priority. A client compliant to the ICE specification will then locally run STUN servers on all addresses being advertised using ICE. Each instance will undertake connectivity checks to ensure that a client can successfully receive media on the advertised address. Only connections that pass the relevant connectivity checks are used for media exchange. The full details of the ICE methodology are in [RFC5245].
This section of the document includes detailed NAT traversal scenarios for both SIP signaling and the associated media. Signaling NAT traversal is achieved using [RFC5626].
The following sub-sections concentrate on SIP signaling traversal of NATs. The scenarios include traversal for both reliable and unreliable transport protocols.
In this example, the client sends a SIP REGISTER request through a NAT. The client will include an 'rport' parameter as described in Section 4.1.1 of this document for allowing traversal of UDP responses. The original request as illustrated in (1) in Figure 5 is a standard SIP REGISTER message:
Message 1:
REGISTER sip:example.com SIP/2.0 Via: SIP/2.0/UDP 192.168.1.2;rport;branch=z9hG4bKnashds7 Max-Forwards: 70 From: Bob <sip:bob@example.com>;tag=7F94778B653B To: Bob <sip:bob@example.com> Call-ID: 16CB75F21C70 CSeq: 1 REGISTER Supported: path, outbound Contact: <sip:bob@192.168.1.2 >;reg-id=1 ;+sip.instance="<urn:uuid:00000000-0000-1000-8000-AABBCCDDEEFF>" Content-Length: 0
Boulton, et al. Informational [Page 13]
RFC 6314 NAT Scenarios July 2011
This SIP transaction now generates a SIP 200 OK response, as depicted in (2) from Figure 5:
Message 2:
SIP/2.0 200 OK Via: SIP/2.0/UDP 192.168.1.2;rport=8050;branch=z9hG4bKnashds7; received=172.16.3.4 From: Bob <sip:bob@example.com>;tag=7F94778B653B To: Bob <sip:bob@example.com>;tag=6AF99445E44A Call-ID: 16CB75F21C70 CSeq: 1 REGISTER Supported: path, outbound Require: outbound Contact: <sip:bob@192.168.1.2 >;reg-id=1;expires=3600 ;+sip.instance="<urn:uuid:00000000-0000-1000-8000-AABBCCDDEEFF>" Content-Length: 0
The response will be sent to the address appearing in the 'received' parameter of the SIP 'Via' header (address 172.16.3.4). The response will not be sent to the port deduced from the SIP 'Via' header, as per standard SIP operation but will be sent to the value that has been stamped in the 'rport' parameter of the SIP 'Via' header (port 8050). For the response to successfully traverse the NAT, all of the conventions defined in RFC 3581 [RFC3581] are to be obeyed. Make note of both the 'reg-id' and 'sip.instance' contact header parameters. They are used to establish an outbound connection tuple as defined in [RFC5626]. The connection tuple creation is clearly shown in Figure 5. This ensures that any inbound request that causes a registration lookup will result in the reuse of the connection path established by the registration. This removes the need to manipulate contact header URIs to represent a globally routable address as perceived on the public side of a NAT.
Traversal of SIP REGISTER requests/responses using a reliable, connection-oriented protocol such as TCP does not require any additional core SIP signaling extensions, beyond the procedures defined in [RFC5626]. SIP responses will reuse the connection created for the initial REGISTER request, (1) from Figure 6:
Message 1:
REGISTER sip:example.com SIP/2.0 Via: SIP/2.0/TCP 192.168.1.2;branch=z9hG4bKnashds7 Max-Forwards: 70 From: Bob <sip:bob@example.com>;tag=7F94778B653B To: Bob <sip:bob@example.com> Call-ID: 16CB75F21C70 CSeq: 1 REGISTER Supported: path, outbound Contact: <sip:bob@192.168.1.2;transport=tcp>;reg-id=1 ;+sip.instance="<urn:uuid:00000000-0000-1000-8000-AABBCCDDEEFF>" Content-Length: 0
Boulton, et al. Informational [Page 15]
RFC 6314 NAT Scenarios July 2011
Message 2:
SIP/2.0 200 OK Via: SIP/2.0/TCP 192.168.1.2;branch=z9hG4bKnashds7 From: Bob <sip:bob@example.com>;tag=7F94778B653B To: Bob <sip:bob@example.com>;tag=6AF99445E44A Call-ID: 16CB75F21C70 CSeq: 1 REGISTER Supported: path, outbound Require: outbound Contact: <sip:bob@192.168.1.2;transport=tcp>;reg-id=1;expires=3600 ;+sip.instance="<urn:uuid:00000000-0000-1000-8000-AABBCCDDEEFF>" Content-Length: 0
This example was included to show the inclusion of the 'sip.instance' contact header parameter as defined in the SIP Outbound specification [RFC5626]. This creates an association tuple as described in the previous example for future inbound requests directed at the newly created registration binding with the only difference that the association is with a TCP connection, not a UDP pinhole binding.
5.1.2. Registration(Registrar/Edge Proxy Not Co-Located)
This section demonstrates traversal mechanisms when the Registrar component is not co-located with the edge proxy element. The procedures described in this section are identical, regardless of transport protocol, so only one example will be documented in the form of TCP.
Figure 7: Registration (Registrar/Proxy Not Co-Located)
This scenario builds on the previous example in Section 5.1.1.2. The primary difference is that the REGISTER request is routed onwards from a proxy server to a separated Registrar. The important message to note is (1) in Figure 7. The edge proxy, on receiving a REGISTER request that contains a 'sip.instance' media feature tag, forms a unique flow identifier token as discussed in [RFC5626]. At this point, the proxy server routes the SIP REGISTER message to the Registrar. The proxy will create the connection tuple as described in SIP Outbound at the same moment as the co-located example, but for subsequent messages to arrive at the proxy, the proxy needs to indicate its need to remain in the SIP signaling path. To achieve this, the proxy inserts to REGISTER message (2) a SIP 'Path' extension header, as defined in RFC 3327 [RFC3327]. The previously created flow association token is inserted in a position within the Path header where it can easily be retrieved at a later point when receiving messages to be routed to the registration binding (in this case the user part of the SIP URI). The REGISTER message of (1) includes a SIP 'Route' header for the edge proxy.
Boulton, et al. Informational [Page 17]
RFC 6314 NAT Scenarios July 2011
Message 1:
REGISTER sip:example.com SIP/2.0 Via: SIP/2.0/TCP 192.168.1.2;branch=z9hG4bKnashds7 Max-Forwards: 70 From: Bob <sip:bob@example.com>;tag=7F94778B653B To: Bob <sip:bob@example.com> Call-ID: 16CB75F21C70 CSeq: 1 REGISTER Supported: path, outbound Route: <sip:ep1.example.com;lr> Contact: <sip:bob@192.168.1.2;transport=tcp>;reg-id=1 ;+sip.instance="<urn:uuid:00000000-0000-1000-8000-AABBCCDDEEFF>" Content-Length: 0
This REGISTER request results in the Path header being stored along with the AOR and its associated binding at the Registrar. The URI contained in the Path header will be inserted as a pre-loaded SIP 'Route' header into any request that arrives at the Registrar and is directed towards the associated AOR binding. This all but guarantees that all requests for the new registration will be forwarded to the edge proxy. In our example, the user part of the SIP 'Path' header URI that was inserted by the edge proxy contains the unique token identifying the flow to the client. On receiving subsequent requests, the edge proxy will examine the user part of the pre-loaded SIP 'Route' header and extract the unique flow token for use in its connection tuple comparison, as defined in the SIP Outbound specification [RFC5626]. An example that builds on this scenario (showing an inbound request to the AOR) is detailed in Section 5.1.4.2 of this document.
The initiating client generates an INVITE request that is to be sent through the NAT to a proxy server. The INVITE message is represented in Figure 8 by (1) and is as follows:
Message 1:
INVITE sip:alice@a.example SIP/2.0 Via: SIP/2.0/UDP 192.168.1.2;rport;branch=z9hG4bKnashds7 Max-Forwards: 70 From: Bob <sip:bob@example.com>;tag=ldw22z To: Alice <sip:alice@a.example> Call-ID: 95KGsk2V/Eis9LcpBYy3 CSeq: 1 INVITE Supported: outbound Route: <sip:ep1.example.com;lr> Contact: <sip:bob@192.168.1.2;ob> Content-Type: application/sdp Content-Length: ...
[SDP not shown]
There are a number of points to note with this message:
1. Firstly, as with the registration example in Section 5.1.1.1, responses to this request will not automatically pass back through a NAT, so the SIP 'Via' header 'rport' is included as described in the Section 4.1.1 ("Symmetric Response") and defined in RFC 3581 [RFC3581].
2. Secondly, the 'ob' parameter is added to the 'Contact' header to ensure that all subsequent requests are sent to the same flow; alternatively, a Globally Routable User Agent URI (GRUU) might have been used. See Section 4.3 of [RFC5626].
In (2), the proxy inserts itself in the 'Via' header, adds the 'rport' port number and the 'received' parameter in the previous 'Via' header, removes the 'Route' header, and inserts a Record-Route with a token.
When using a reliable transport such as TCP, the call flow and procedures for traversing a NAT are almost identical to those described in Section 5.1.3.1. The primary difference when using reliable transport protocols is that symmetric response [RFC3581] is not required for SIP responses to traverse a NAT. RFC 3261 [RFC3261] defines procedures for SIP response messages to be sent back on the same connection on which the request arrived. See Section 9.5 of [RFC5626] for an example flow of an outgoing call.
This section details scenarios where a client behind a NAT receives an inbound request through a NAT. These scenarios build on the previous registration scenario from Sections 5.1.1 and 5.1.2 in this document.
The SIP signaling on the interior of the network (behind the user's proxy) is not impacted directly by the transport protocol, so only one example scenario is necessary. The example uses UDP and follows on from the registration installed in the example from Section 5.1.1.1.
An INVITE request arrives at the authoritative proxy with a destination pointing to the AOR of that inserted in Section 5.1.1.1. The message is illustrated by (1) in Figure 9 and looks as follows:
INVITE sip:bob@example.com SIP/2.0 Via: SIP/2.0/UDP 172.16.1.4;branch=z9hG4bK74huHJ37d Max-Forwards: 70 From: External Alice <sip:alice@example.com>;tag=02935 To: Bob <sip:bob@example.com> Call-ID: klmvCxVWGp6MxJp2T2mb CSeq: 1 INVITE Contact: <sip:alice@172.16.1.4> Content-Type: application/sdp Content-Length: ..
[SDP not shown]
The INVITE request matches the registration binding previously installed at the Registrar and the INVITE Request-URI is rewritten to the selected onward address. The proxy then examines the Request-URI of the INVITE and compares with its list of connection tuples. It uses the incoming AOR to commence the check for associated open connections/mappings. Once matched, the proxy checks to see if the unique instance identifier (+sip.instance) associated with the binding equals the same instance identifier associated with that connection tuple. The request is then dispatched on the appropriate binding. This is message (2) from Figure 9 and is as follows:
Boulton, et al. Informational [Page 23]
RFC 6314 NAT Scenarios July 2011
INVITE sip:bob@192.168.1.2 SIP/2.0 Via: SIP/2.0/UDP ep1.example.com;branch=z9hG4kmlds893jhsd Via: SIP/2.0/UDP 172.16.1.4;branch=z9hG4bK74huHJ37d Max-Forwards: 69 From: Alice <sip:alice@example.com>;tag=02935 To: client bob <sip:bob@example.com> Call-ID: klmvCxVWGp6MxJp2T2mb CSeq: 1 INVITE Contact: <sip:alice@172.16.1.4> Content-Type: application/sdp Content-Length: ..
[SDP not shown]
It is a standard SIP INVITE request with no additional functionality. The major difference is that this request will not be forwarded to the address specified in the Request-URI, as standard SIP rules would enforce, but will be sent on the flow associated with the registration binding (lookup procedures in RFC 3263 [RFC3263] are overridden by RFC 5626 [RFC5626]). This then allows the original connection/mapping from the initial registration process to be reused.
5.1.4.2. Edge Proxy/Authoritative Proxy Not Co-Located
The core SIP signaling associated with this call flow is not impacted directly by the transport protocol, so only one example scenario is necessary. The example uses UDP and follows on from the registration installed in the example from Section 5.1.2.
An INVITE request arrives at the authoritative proxy with a destination pointing to the AOR of that inserted in Section 5.1.2. The message is illustrated by (1) in Figure 10 and looks as follows:
INVITE sip:bob@example.com SIP/2.0 Via: SIP/2.0/UDP 172.16.1.4;branch=z9hG4bK74huHJ37d Max-Forwards: 70 From: Alice <sip:alice@example.com>;tag=02935 To: Bob <sip:bob@example.com> Call-ID: klmvCxVWGp6MxJp2T2mb CSeq: 1 INVITE Contact: <sip:external@172.16.1.4> Content-Type: application/sdp Content-Length: ..
[SDP not shown]
The INVITE request matches the registration binding previously installed at the Registrar and the INVITE Request-URI is rewritten to the selected onward address. The Registrar also identifies that a SIP 'Path' header was associated with the registration and pushes it into the INVITE request in the form of a pre-loaded SIP Route header. It then forwards the request on to the proxy identified in the SIP Route header as shown in (2) from Figure 10:
Boulton, et al. Informational [Page 25]
RFC 6314 NAT Scenarios July 2011
INVITE sip:bob@client.example.com SIP/2.0 Via: SIP/2.0/UDP proxy.example.com;branch=z9hG4bK74fmljnc Via: SIP/2.0/UDP 172.16.1.4;branch=z9hG4bK74huHJ37d Route: <sip:VskztcQ/S8p4WPbOnHbuyh5iJvJIW3ib@ep1.example.com;lr;ob> Max-Forwards: 69 From: Alice <sip:alice@example.net>;tag=02935 To: Bob <sip:Bob@example.com> Call-ID: klmvCxVWGp6MxJp2T2mb CSeq: 1 INVITE Contact: <sip:alice@172.16.1.4> Content-Type: application/sdp Content-Length: ..
[SDP not shown]
The request then arrives at the outbound proxy for the client. The proxy examines the Request-URI of the INVITE in conjunction with the flow token that it previously inserted into the user part of the 'Path' header SIP URI (which now appears in the user part of the Route header in the incoming INVITE). The proxy locates the appropriate flow and sends the message to the client, as shown in (3) from Figure 10:
It is a standard SIP INVITE request with no additional functionality at the originator. The major difference is that this request will not follow the address specified in the Request-URI when it reaches the outbound proxy, as standard SIP rules would enforce, but will be sent on the flow associated with the registration binding as indicated in the Route header (lookup procedures in RFC 3263 [RFC3263] are overridden). This then allows the original connection/ mapping from the initial registration to the outbound proxy to be reused.
This section provides example scenarios to demonstrate basic media traversal using the techniques outlined earlier in this document.
In the flow diagrams, STUN messages have been annotated for simplicity as follows:
o The "Src" attribute represents the source transport address of the message.
o The "Dest" attribute represents the destination transport address of the message.
o The "Map" attribute represents the server reflexive (XOR-MAPPED- ADDRESS STUN attribute) transport address.
o The "Rel" attribute represents the relayed (RELAY-ADDRESS STUN attribute) transport address.
The meaning of each STUN attribute is extensively explained in the core STUN [RFC5389] and TURN [RFC5766] specifications.
A number of ICE SDP attributes have also been included in some of the examples. Detailed information on individual attributes can be obtained from the core ICE specification [RFC5245].
The examples also contain a mechanism for representing transport addresses. It would be confusing to include representations of network addresses in the call flows and would make them hard to follow. For this reason, network addresses will be represented using the following annotation. The first component will contain the representation of the client responsible for the address. For example, in the majority of the examples "L" (left client), "R" (right client), "NAT-PUB" (NAT public), "PRIV" (Private), and "STUN- PUB" (STUN public) are used. To allow for multiple addresses from the same network element, each representation can also be followed by a number. These can also be used in combination. For example, "L-NAT-PUB-1" would represent a public network address of the left- hand side NAT while "R-NAT-PUB-1" would represent a public network address of the right-hand side of the NAT. "L-PRIV-1" would represent a private network address of the left-hand side of the NAT while "R-PRIV-1" represents a private address of the right-hand side of the NAT.
Boulton, et al. Informational [Page 27]
RFC 6314 NAT Scenarios July 2011
It should also be noted that, during the examples, it might be appropriate to signify an explicit part of a transport address. This is achieved by adding either the '.address' or '.port' tag on the end of the representation -- for example, 'L-PRIV-1.address' and 'L-PRIV- 1.port'.
The use of '$' signifies variable parts in example SIP messages.
This section demonstrates an example of a client both initiating and receiving calls behind an Endpoint-Independent NAT. An example is included for both STUN and ICE with ICE being the RECOMMENDED mechanism for media traversal.
At this time, there is no reliable test to determine if a host is behind an Endpoint-Independent Filtering NAT or an Endpoint- Independent Mapping NAT [RFC5780], and the sort of failure that occurs in this situation is described in Section 5.2.2.1. For this reason, ICE is RECOMMENDED over the mechanism described in this section.
It is possible to traverse media through an Endpoint-Independent NAT using STUN. The remainder of this section provides simplified examples of the 'Binding Discovery' STUN as defined in [RFC5389]. The STUN messages have been simplified and do not include 'Shared Secret' requests used to obtain the temporary username and password.
The following example demonstrates media traversal through a NAT with Endpoint-Independent Mapping properties using the STUN 'Binding Discovery' usage. It is assumed in this example that the STUN client and SIP Client are co-located on the same physical machine. Note that some SIP signaling messages have been left out for simplicity.
|(9)SIP INVITE | | | |----------------->| | | | | | | | |(10)SIP INVITE | | | |------------------------------------>| | | | | | | |(11)SIP 200 OK | | |<------------------------------------| | | | | |(12)SIP 200 OK | | | |<-----------------| | | | | | | |========================================================| |>>>>>>>>>>>>Outgoing Media sent from L-PRIV-1>>>>>>>>>>>| |========================================================| | | |========================================================| |<<<<<<<<<<<<Incoming Media sent to NAT-PUB-1<<<<<<<<<<<<| |========================================================| | | |========================================================| |>>>>>>>>>>>>Outgoing RTCP sent from L-PRIV-2>>>>>>>>>>>>| |========================================================| | | |========================================================| |<<<<<<<<<<<<Incoming RTCP sent to NAT-PUB-2<<<<<<<<<<<<<| |========================================================| | | | | |(13)SIP ACK | | | |----------------->| | | | | | | | |(14) SIP ACK | | | |------------------------------------>| | | | |
Figure 11: Endpoint-Independent NAT - Initiating
o On deciding to initiate a SIP voice session, the client starts a local STUN client on the interface and port that is to be used for media (send/receive). The STUN client generates a standard 'Binding Discovery' request as indicated in (1) from Figure 11 that also highlights the source address and port for which the client device wishes to obtain a mapping. The 'Binding Discovery' request is sent through the NAT towards the public Internet and STUN server.
Boulton, et al. Informational [Page 30]
RFC 6314 NAT Scenarios July 2011
o Message (2) traverses the NAT and breaks out onto the public Internet towards the public STUN server. Note that the source address of the 'Binding Discovery' request now represents the public address and port from the public side of the NAT.
o The STUN server receives the request and processes it appropriately. This results in a successful 'Binding Discovery' response being generated and returned (3). The message contains details of the XOR-mapped public address (contained in the STUN XOR-MAPPED-ADDRESS attribute) that is to be used by the originating client to receive media (see 'Map=NAT-PUB-1' from (3)).
o The 'Binding Discovery' response traverses back through the NAT using the path created by the 'Binding Discovery' request and presents the new XOR-mapped address to the client (4). At this point, the process is repeated to obtain a second XOR-mapped address (as shown in (5)-(8)) for a second local address (the address has changed from "L-PRIV-1" to "L-PRIV-2") for an RTCP port.
o The client now constructs a SIP INVITE message (9). Note that traversal of SIP is not covered in this example and is discussed in Section 5.1. The INVITE request will use the addresses it has obtained in the previous STUN transactions to populate the SDP of the SIP INVITE as shown below:
o Note that the XOR-mapped address obtained from the 'Binding Discovery' transactions are inserted as the connection address for the SDP (c=$NAT-PUB-1.address). The Primary port for RTP is also inserted in the SDP (m=audio $NAT-PUB-1.port RTP/AVP 0). Finally, the port gained from the additional 'Binding Discovery' is placed in the RTCP attribute (as discussed in Section 4.2.2) for traversal of RTCP (a=rtcp:$NAT-PUB-2.port).
o The SIP signaling then traverses the NAT and sets up the SIP session (9-12). Note that the left-hand client transmits media as soon as the 200 OK to the INVITE arrives at the client (12). Up until this point, the incoming media and RTCP to the left-hand
Boulton, et al. Informational [Page 31]
RFC 6314 NAT Scenarios July 2011
client will not pass through the NAT as no outbound association has been created with the far-end client. Two-way media communication has now been established.
Receiving a session for an Endpoint-Independent NAT using the STUN 'Binding Discovery' usage is very similar to the example outlined in Section 5.2.1.1.1. Figure 12 illustrates the associated flow of messages.
o On receiving an invitation to a SIP voice session (SIP INVITE request), the User Agent starts a local STUN client on the appropriate port on which it is to receive media. The STUN client generates a standard 'Binding Discovery' request as indicated in (3) from Figure 12 that also highlights the source address and port for which the client device wishes to obtain a mapping. The 'Binding Discovery' request is sent through the NAT towards the public Internet and STUN server.
o 'Binding Discovery' message (4) traverses the NAT and breaks out onto the public Internet towards the public STUN server. Note that the source address of the STUN requests now represents the public address and port from the public side of the NAT.
o The STUN server receives the request and processes it appropriately. This results in a successful 'Binding Discovery' response being generated and returned (5). The message contains details of the mapped public address (contained in the STUN XOR- MAPPED-ADDRESS attribute) that is to be used by the originating client to receive media (see 'Map=NAT-PUB-1' from (5)).
o The 'Binding Discovery' response traverses back through the NAT using the path created by the outgoing 'Binding Discovery' request and presents the new XOR-mapped address to the client (6). At this point, the process is repeated to obtain a second XOR-mapped address (as shown in (7)-(10)) for a second local address (local port has now changed and is represented by L-PRIV-2 in (7)) for an RTCP port.
o The client now constructs a SIP 200 OK message (11) in response to the original SIP INVITE requests. Note that traversal of SIP is not covered in this example and is discussed in Section 5.1. SIP Provisional responses are also left out for simplicity. The 200 OK response will use the addresses it has obtained in the previous STUN transactions to populate the SDP of the SIP 200 OK as shown below:
o Note that the XOR-mapped address obtained from the initial 'Binding Discovery' transaction is inserted as the connection address for the SDP (c=NAT-PUB-1.address). The Primary port for RTP is also inserted in the SDP (m=audio NAT-PUB-1.port RTP/AVP
Boulton, et al. Informational [Page 34]
RFC 6314 NAT Scenarios July 2011
0). Finally, the port gained from the second 'Binding Discovery' is placed in the RTCP attribute (as discussed in Section 4.2.2) for traversal of RTCP (a=rtcp:NAT-PUB-2.port).
o The SIP signaling then traverses the NAT and sets up the SIP session (11-14). Note that the left-hand client transmits media as soon as the 200 OK to the INVITE is sent to the User Agent Client (UAC) (11). Up until this point, the incoming media from the right-hand client will not pass through the NAT as no outbound association has been created with the far-end client. Two-way media communication has now been established.
The preferred solution for media traversal of NAT is using ICE, as described in Section 4.2.3.3, regardless of the NAT type. The following examples illustrate the traversal of an Endpoint- Independent NAT when initiating the session. The example only covers ICE in association with the 'Binding Discovery' and TURN. It is worth noting that the TURN server provides both STUN functions (to learn your public mapping) and TURN functions (media relaying). It is also worth noting that in the example described in Section 5.2.1.2.1, both SIP clients L and R are contacting the same TURN server. This is not necessary for ICE, STUN, TURN to function; all that is necessary is that the STUN and TURN server(s) be in the same addressing domain that is accessible on the Internet.
o On deciding to initiate a SIP voice session, the SIP client L starts a local STUN client. The STUN client generates a TURN Allocate request as indicated in (1) from Figure 13 that also highlights the source address and port combination for which the client device wishes to obtain a mapping. The Allocate request is sent through the NAT towards the public Internet.
Boulton, et al. Informational [Page 43]
RFC 6314 NAT Scenarios July 2011
o The Allocate message (2) traverses the NAT to the public Internet towards the public TURN server. Note that the source address of the Allocate request now represents the public address and port from the public side of the NAT (L-NAT-PUB-1).
o The TURN server receives the Allocate request and processes it appropriately. This results in a successful Allocate response being generated and returned (3). The message contains details of the server reflexive address that is to be used by the originating client to receive media (see 'Map=L-NAT-PUB-1') from (3)). It also contains an appropriate TURN-relayed address that can be used at the STUN server (see 'Rel=TURN-PUB-2').
o The Allocate response traverses back through the NAT using the binding created by the initial Allocate request and presents the new mapped address to the client (4). The process is repeated and a second STUN derived set of addresses is obtained, as illustrated in (5)-(8) in Figure 13. At this point, the User Agent behind the NAT has pairs of derived external server reflexive and relayed representations. The client can also gather IP addresses and ports via other mechanisms (e.g., NAT-PMP [NAT-PMP], UPnP IGD [UPnP-IGD]) or similar.
o The client now constructs a SIP INVITE message (9). The INVITE request will use the addresses it has obtained in the previous STUN/TURN interactions to populate the SDP of the SIP INVITE. This should be carried out in accordance with the semantics defined in the ICE specification [RFC5245], as shown below in Figure 14:
o The SDP has been constructed to include all the available candidates that have been assembled. The first set of candidates (as identified by Foundation $L1) contains two local addresses that have the highest priority. They are also encoded into the connection (c=) and media (m=) lines of the SDP. The second set of candidates, as identified by Foundation $L2, contains the two server reflexive addresses obtained from the STUN server for both RTP and RTCP traffic (identified by candidate-id $L2). This entry has been given a priority lower than the pair $L1 by the client. The third and final set of candidates represents the relayed addresses (as identified by $L3) obtained from the STUN server. This pair has the lowest priority and will be used as a last resort if both $L1 and $L2 fail.
o The SIP signaling then traverses the NAT and sets up the SIP session (9)-(10). On advertising a candidate address, the client should have a local STUN server running on each advertised candidate address. This is for the purpose of responding to incoming STUN connectivity checks.
o On receiving the SIP INVITE request (10) client R also starts local STUN servers on appropriate address/port combinations and gathers potential candidate addresses to be encoded into the SDP (as the originating client did). Steps (11-18) involve client R
Boulton, et al. Informational [Page 45]
RFC 6314 NAT Scenarios July 2011
carrying out the same steps as client L. This involves obtaining local, server reflexive, and relayed addresses. Client R is now ready to generate an appropriate answer in the SIP 200 OK message (19). The example answer follows in Figure 15:
o The two clients have now exchanged SDP using offer/answer and can now continue with the ICE processing -- User Agent L assuming the role controlling agent, as specified by ICE. The clients are now required to form their Candidate check lists to determine which will be used for the media streams. In this example, User Agent L's Foundation 1 is paired with User Agent R's Foundation 1, User Agent L's Foundation 2 is paired with User Agent R's Foundation 2, and finally User Agent L's Foundation 3 is paired with User Agent R's Foundation 3. User Agents L and R now have a complete candidate checklist. Both clients now use the algorithm provided in ICE to determine candidate pair priorities and sort into a list of decreasing priorities. In this example, both User Agents L and R will have lists that firstly specify the host address (Foundation $L1), then the server reflexive address (Foundation $L2), and lastly the relayed address (Foundation $L3). All candidate pairs have an associate state as specified in ICE. At this stage, all of the candidate pairs for User Agents L and R are initialized to the 'Frozen' state. The User Agents then scan the list and move the candidates to the 'Waiting' state. At this point, both clients will periodically, starting with the highest
Boulton, et al. Informational [Page 46]
RFC 6314 NAT Scenarios July 2011
candidate pair priority, work their way down the list issuing STUN checks from the local candidate to the remote candidate (of the candidate pair). As a STUN check is attempted from each local candidate in the list, the candidate pair state transitions to 'In-Progress'. As illustrated in (23), client L constructs a STUN connectivity check in an attempt to validate the remote candidate address received in the SDP of the 200 OK (20) for the highest priority in the checklist. As a private address was specified in the active address in the SDP, the STUN connectivity check fails to reach its destination causing a STUN failure. Client L transitions the state for this candidate pair to 'Failed'. In the meantime, client L is attempting a STUN connectivity check for the second candidate pair in the returned SDP with the second highest priority (24). As can be seen from messages (24) to (29), the STUN Bind request is successful and returns a positive outcome for the connectivity check. Client L is now free to send media to the peer using the candidate pair. Immediately after sending its 200 OK, client R also carries out the same set of binding requests. It firstly (in parallel) tries to contact the active address contained in the SDP (30) which results in failure.
o In the meantime, a successful response to a STUN connectivity check by User Agent R (27) results in a tentative check in the reverse direction -- this is illustrated by messages (31) to (36). Once this check has succeeded, User Agent R can transition the state of the appropriate candidate to 'Succeeded', and media can be sent (RTP). The previously (31-36) described check confirm on both sides (User Agents L and R) that connectivity can be achieved using the appropriate candidate pair. User Agent L, as the controlling client now sends another connectivity check for the candidate pair, this time including the 'USE-CANDIDATE' attribute as specified in ICE to signal the chosen candidate. This exchange is illustrated in messages (37) to (42).
o As part of the process in this example, both L and R will now complete the same connectivity checks for part 2 of the component named for the favored 'Foundation' selected for use with RTCP. The connectivity checks for part 2 of the candidate component are shown in L (43-48) and R (49-54). Once this has succeeded, User Agent L as the controlling client sends another connectivity check for the candidate pair. This time the 'USE-CANDIDATE' attribute is again specified to signal the chosen candidate for component 2.
o The candidates have now been fully verified (and selected), and as they are the highest priority, an updated offer (61-62) is now sent from the offerer (client L) to the answerer (client R) representing the new active candidates. The new offer would look as follows:
This section highlights that although using STUN techniques is the preferred mechanism for traversal of NAT, it does not solve every case. The use of basic STUN on its own will not guarantee traversal through every NAT type, hence the recommendation that ICE is the preferred option.
Figure 18: Address/Port-Dependent NAT with STUN - Failure
Boulton, et al. Informational [Page 49]
RFC 6314 NAT Scenarios July 2011
The example in Figure 18 is conveyed in the context of a client behind the Address/Port-Dependent NAT initiating a call. It should be noted that the same problem applies when a client receives a SIP invitation and is behind a Address/Port-Dependent NAT.
o In Figure 18, the client behind the NAT obtains a server reflexive representation using standard STUN mechanisms (1)-(4) that have been used in previous examples in this document (e.g., Section 5.2.1.1.1).
o The external mapped address (server reflexive) obtained is also used in the outgoing SDP contained in the SIP INVITE request (5).
o In this example, the client is still able to send media to the external client. The problem occurs when the client outside the NAT tries to use the reflexive address supplied in the outgoing INVITE request to traverse media back through the Address/ Port-Dependent NAT.
o A Address/Port-Dependent NAT has differing rules from the Endpoint-Independent type of NAT (as defined in RFC 4787 [RFC4787]). For any internal IP address and port combination, data sent to a different external destination does not provide the same public mapping at the NAT. In Figure 18, the STUN query produced a valid external mapping for receiving media. This mapping, however, can only be used in the context of the original STUN request that was sent to the STUN server. Any packets that attempt to use the mapped address and that do not originate from the STUN server IP address and optionally port will be dropped at the NAT. Figure 18 shows the media being dropped at the NAT after (7) and before (8). This then leads to one-way audio.
As identified in Section 5.2.2.1, STUN provides a useful tool for the traversal of the majority of NATs but fails with Address/ Port-Dependent NAT. The TURN extensions [RFC5766] address this scenario. TURN extends STUN to allow a client to request a relayed address at the TURN server rather than a reflexive representation. This then introduces a media relay in the path for NAT traversal (as described in Section 4.2.3.2). The following example explains how TURN solves the previous failure when using STUN to traverse a Address/Port-Dependent NAT. It should be noted that TURN works most effectively when used in conjunction with ICE. Using TURN on its own results in all media being relayed through a TURN server; this is not efficient.
Figure 19: Address/Port-Dependent NAT with TURN - Success
Boulton, et al. Informational [Page 52]
RFC 6314 NAT Scenarios July 2011
o In this example, client L issues a TURN allocate request (1) to obtained a relay address at the STUN server. The request traverses through the Address/Port-Dependent NAT and reaches the STUN server (2). The STUN server generates an Allocate response (3) that contains both a server reflexive address (Map=NAT-PUB-1) of the client and also a relayed address (Rel=STUN-PUB-2). The relayed address maps to an address mapping on the STUN server that is bound to the public pinhole that has been opened on the NAT by the Allocate request. This results in any traffic sent to the TURN server relayed address (Rel=STUN-PUB-2) being forwarded to the external representation of the pinhole created by the Allocate request (NAT-PUB-1).
o The TURN derived address (STUN-PUB-2) arrives back at the originating client (4) in an Allocate response. This address can then be used in the SDP for the outgoing SIP INVITE request as shown in the following example (note that the example also includes client L obtaining a second relay address for use in the RTCP attribute (5-8)):
o On receiving the INVITE request, the User Agent Server (UAS) is able to stream media and RTCP to the relay address (STUN-PUB-2 and STUN-PUB-3) at the STUN server. As shown in Figure 19 (between messages (12) and (13), the media from the UAS is directed to the relayed address at the STUN server. The STUN server then forwards the traffic to the open pinholes in the Address/Port-Dependent NAT (NAT-PUB-1 and NAT-PUB-2). The media traffic is then able to traverse the Address/Port-Dependent NAT and arrives back at client L.
o TURN on its own will work for Address/Port-Dependent and other types of NAT mentioned in this specification but should only be used as a last resort. The relaying of media through an external entity is not an efficient mechanism for NAT traversal and comes at a high processing cost.
The previous two examples have highlighted the problem with using core STUN for all forms of NAT traversal and a solution using TURN for the Address/Port-Dependent NAT case. The RECOMMENDED mechanism
Boulton, et al. Informational [Page 53]
RFC 6314 NAT Scenarios July 2011
for traversing all varieties of NAT is using ICE, as detailed in Section 4.2.3.3. ICE makes use of core STUN, TURN and any other mechanism (e.g., NAT-PMP[NAT-PMP], UPnP IGD[UPnP-IGD]) to provide a list of prioritized addresses that can be used for media traffic. Detailed examples of ICE can be found in Section 5.2.1.2.1. These examples are associated with an Endpoint-Independent type NAT but can be applied to any NAT type variation, including Address/ Port-Dependent type NAT. The ICE procedures carried out are the same. For a list of candidate addresses, a client will choose where to send media dependent on the results of the STUN connectivity checks and associated priority (highest priority wins). It should be noted that the inclusion of a NAT displaying Address/Port-Dependent properties does not automatically result in relayed media. In fact, ICE processing will avoid use of media relay with the exception of two clients that both happen to be behind a NAT using Address/ Port-Dependent characteristics. The connectivity checks and associated selection algorithm enable traversal in this case. Figure 20 and the following description provide a guide as to how this is achieved using the ICE connectivity checks. This is an abbreviated example that assumes successful SIP offer/answer exchange and illustrates the connectivity check flow.
Figure 20: Single Address/Port-Dependent NAT - Success
In this abbreviated example, client R has already received a SIP INVITE request and is starting its connectivity checks with client L. Client R generates a connectivity check (1) and sends to client L's information as presented in the SDP offer. The request arrives at client L's Address/Port-Dependent NAT and fails to traverse as there is no NAT binding. This would then move the connectivity check to a failed state. In the meantime, client L has received the SDP answer in the SIP request and will also commence connectivity checks. A check is dispatched (3) to client R. The check is able to traverse the NAT due to the association set up in the previously failed check (1). The full Bind request/response is shown in steps (3)-(8). As part of a candidate pair, client R will now successfully be able to complete the checks, as illustrated in steps (9)-(14). The result is a successful pair of candidates that can be used without the need to relay any media.
Boulton, et al. Informational [Page 56]
RFC 6314 NAT Scenarios July 2011
In conclusion, the only time media needs to be relayed is a result of clients both behind Address/Port-Dependent NATs. As you can see from the example in this section, neither side would be able to complete connectivity checks with the exception of the Relayed candidates.
This section describes how IPv6-only SIP User Agents can communicate with IPv4-only SIP User Agents. While the techniques discussed in this document primarily contain examples of traversing NATs to allow communications between hosts in private and public networks, they are by no means limited to such scenarios. The same NAT traversal techniques can also be used to establish communication in a heterogeneous network environment -- e.g., communication between an IPv4 host and an IPv6 host.
IPv4-IPv6 translations at the SIP level usually take place at dual- stack proxies that have both IPv4 and IPv6 DNS entries. Since these translations do not involve NATs that are placed in the middle of two SIP entities, they fall outside the scope of this document. A detailed description of this type of translation can be found in [RFC6157].
The authors would like to thank the members of the IETF SIPPING WG for their comments and suggestions. Expert review and detailed contribution including text was provided by Dan Wing, who was supportive throughout.
Detailed comments were provided by Vijay Gurbani, Kaiduan Xie, Remi Denis-Courmont, Hadriel Kaplan, Phillip Matthews, Spencer Dawkins, and Hans Persson.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, A., Peterson, J., Sparks, R., Handley, M., and E. Schooler, "SIP: Session Initiation Protocol", RFC 3261, June 2002.
[RFC3263] Rosenberg, J. and H. Schulzrinne, "Session Initiation Protocol (SIP): Locating SIP Servers", RFC 3263, June 2002.
[RFC3264] Rosenberg, J. and H. Schulzrinne, "An Offer/Answer Model with Session Description Protocol (SDP)", RFC 3264, June 2002.
[RFC3327] Willis, D. and B. Hoeneisen, "Session Initiation Protocol (SIP) Extension Header Field for Registering Non-Adjacent Contacts", RFC 3327, December 2002.
[RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. Jacobson, "RTP: A Transport Protocol for Real-Time Applications", STD 64, RFC 3550, July 2003.
[RFC3581] Rosenberg, J. and H. Schulzrinne, "An Extension to the Session Initiation Protocol (SIP) for Symmetric Response Routing", RFC 3581, August 2003.
[RFC3605] Huitema, C., "Real Time Control Protocol (RTCP) attribute in Session Description Protocol (SDP)", RFC 3605, October 2003.
[RFC4566] Handley, M., Jacobson, V., and C. Perkins, "SDP: Session Description Protocol", RFC 4566, July 2006.
[RFC4787] Audet, F. and C. Jennings, "Network Address Translation (NAT) Behavioral Requirements for Unicast UDP", BCP 127, RFC 4787, January 2007.
[RFC4961] Wing, D., "Symmetric RTP / RTP Control Protocol (RTCP)", BCP 131, RFC 4961, July 2007.
Boulton, et al. Informational [Page 58]
RFC 6314 NAT Scenarios July 2011
[RFC5245] Rosenberg, J., "Interactive Connectivity Establishment (ICE): A Protocol for Network Address Translator (NAT) Traversal for Offer/Answer Protocols", RFC 5245, April 2010.
[RFC5389] Rosenberg, J., Mahy, R., Matthews, P., and D. Wing, "Session Traversal Utilities for NAT (STUN)", RFC 5389, October 2008.
[RFC5626] Jennings, C., Mahy, R., and F. Audet, "Managing Client-Initiated Connections in the Session Initiation Protocol (SIP)", RFC 5626, October 2009.
[RFC5761] Perkins, C. and M. Westerlund, "Multiplexing RTP Data and Control Packets on a Single Port", RFC 5761, April 2010.
[RFC5766] Mahy, R., Matthews, P., and J. Rosenberg, "Traversal Using Relays around NAT (TURN): Relay Extensions to Session Traversal Utilities for NAT (STUN)", RFC 5766, April 2010.
[RFC5923] Gurbani, V., Mahy, R., and B. Tate, "Connection Reuse in the Session Initiation Protocol (SIP)", RFC 5923, June 2010.
[MIDDLEBOXES] Stucker, B. and H. Tschofenig, "Analysis of Middlebox Interactions for Signaling Protocol Communication along the Media Path", Work in Progress, July 2010.
[NAT-PMP] Cheshire, S., "NAT Port Mapping Protocol (NAT-PMP)", Work in Progress, April 2008.
[RFC2026] Bradner, S., "The Internet Standards Process -- Revision 3", BCP 9, RFC 2026, October 1996.
[RFC3424] Daigle, L. and IAB, "IAB Considerations for UNilateral Self-Address Fixing (UNSAF) Across Network Address Translation", RFC 3424, November 2002.
[RFC5780] MacDonald, D. and B. Lowekamp, "NAT Behavior Discovery Using Session Traversal Utilities for NAT (STUN)", RFC 5780, May 2010.
Boulton, et al. Informational [Page 59]
RFC 6314 NAT Scenarios July 2011
[RFC5853] Hautakorpi, J., Camarillo, G., Penfield, R., Hawrylyshen, A., and M. Bhatia, "Requirements from Session Initiation Protocol (SIP) Session Border Control (SBC) Deployments", RFC 5853, April 2010.
[RFC6157] Camarillo, G., El Malki, K., and V. Gurbani, "IPv6 Transition in the Session Initiation Protocol (SIP)", RFC 6157, April 2011.