You are here: irt.org | RFCs | RFC5853 [ previous next ]
Internet Engineering Task Force (IETF) J. Hautakorpi, Ed. Request for Comments: 5853 G. Camarillo Category: Informational Ericsson ISSN: 2070-1721 R. Penfield Acme Packet A. Hawrylyshen Skype, Inc. M. Bhatia 3CLogic April 2010 Requirements from Session Initiation Protocol (SIP) Session Border Control (SBC) Deployments Abstract This document describes functions implemented in Session Initiation Protocol (SIP) intermediaries known as Session Border Controllers (SBCs). The goal of this document is to describe the commonly provided functions of SBCs. A special focus is given to those practices that are viewed to be in conflict with SIP architectural principles. This document also explores the underlying requirements of network operators that have led to the use of these functions and practices in order to identify protocol requirements and determine whether those requirements are satisfied by existing specifications or if additional standards work is required. 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/rfc5853. Hautakorpi, et al. Informational [Page 1]
RFC 5853 Requirements from SIP SBC Deployments April 2010 Copyright Notice Copyright (c) 2010 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 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. Hautakorpi, et al. Informational [Page 2]
RFC 5853 Requirements from SIP SBC Deployments April 2010 Table of Contents 1. Introduction ....................................................4 2. Background on SBCs ..............................................4 2.1. Peering Scenario ...........................................6 2.2. Access Scenario ............................................6 3. Functions of SBCs ...............................................8 3.1. Topology Hiding ............................................8 3.1.1. General Information and Requirements ................8 3.1.2. Architectural Issues ................................9 3.1.3. Example .............................................9 3.2. Media Traffic Management ..................................11 3.2.1. General Information and Requirements ...............11 3.2.2. Architectural Issues ...............................12 3.2.3. Example ............................................13 3.3. Fixing Capability Mismatches ..............................14 3.3.1. General Information and Requirements ...............14 3.3.2. Architectural Issues ...............................14 3.3.3. Example ............................................15 3.4. Maintaining SIP-Related NAT Bindings ......................15 3.4.1. General Information and Requirements ...............15 3.4.2. Architectural Issues ...............................16 3.4.3. Example ............................................17 3.5. Access Control ............................................18 3.5.1. General Information and Requirements ...............18 3.5.2. Architectural Issues ...............................19 3.5.3. Example ............................................19 3.6. Protocol Repair ...........................................20 3.6.1. General Information and Requirements ...............20 3.6.2. Architectural Issues ...............................21 3.6.3. Examples ...........................................21 3.7. Media Encryption ..........................................21 3.7.1. General Information and Requirements ...............21 3.7.2. Architectural Issues ...............................22 3.7.3. Example ............................................22 4. Derived Requirements for Future SIP Standardization Work .......23 5. Security Considerations ........................................23 6. Acknowledgements ...............................................24 7. References .....................................................25 7.1. Normative References ......................................25 7.2. Informative References ....................................25 Hautakorpi, et al. Informational [Page 3]
RFC 5853 Requirements from SIP SBC Deployments April 2010 1. Introduction In the past few years, there has been a rapid adoption of the Session Initiation Protocol (SIP) [1] and deployment of SIP-based communications networks. This has often outpaced the development and implementation of protocol specifications to meet network operator requirements. This has led to the development of proprietary solutions. Often, these proprietary solutions are implemented in network intermediaries known in the marketplace as Session Border Controllers (SBCs) because they typically are deployed at the border between two networks. The reason for this is that network policies are typically enforced at the edge of the network. Even though many SBCs currently behave in ways that can break end-to- end security and impact feature negotiations, there is clearly a market for them. Network operators need many of the features current SBCs provide, and often there are no standard mechanisms available to provide them. The purpose of this document is to describe functions implemented in SBCs. A special focus is given to those practices that conflict with SIP architectural principles in some way. The document also explores the underlying requirements of network operators that have led to the use of these functions and practices in order to identify protocol requirements and determine whether those requirements are satisfied by existing specifications or if additional standards work is required. 2. Background on SBCs The term SBC is relatively non-specific, since it is not standardized or defined anywhere. Nodes that may be referred to as SBCs but do not implement SIP are outside the scope of this document. SBCs usually sit between two service provider networks in a peering environment, or between an access network and a backbone network to provide service to residential and/or enterprise customers. They provide a variety of functions to enable or enhance session-based multi-media services (e.g., Voice over IP). These functions include: a) perimeter defense (access control, topology hiding, and denial-of- service prevention and detection); b) functionality not available in the endpoints (NAT traversal, protocol interworking or repair); and c) traffic management (media monitoring and Quality of Service (QoS)). Some of these functions may also get integrated into other SIP elements (like pre-paid platforms, Third Generation Partnership Project (3GPP) Proxy CSCF (P-CSCF) [6], 3GPP I-CSCF, etc.). Hautakorpi, et al. Informational [Page 4]
RFC 5853 Requirements from SIP SBC Deployments April 2010 SIP-based SBCs typically handle both signaling and media and can implement behavior that is equivalent to a "privacy service" (as described in [2]) performing both Header Privacy and Session Privacy). SBCs often modify certain SIP headers and message bodies that proxies are not allowed to modify. Consequently, they are, by definition, B2BUAs (Back-to-Back User Agents). The transparency of these B2BUAs varies depending on the functions they perform. For example, some SBCs modify the session description carried in the message and insert a Record-Route entry. Other SBCs replace the value of the Contact header field with the SBCs' address and generate a new Call-ID and new To and From tags. +-----------------+ | SBC | [signaling] | +-----------+ | <------------|->| signaling |<-|----------> outer | +-----------+ | inner network | | | network | +-----------+ | <------------|->| media |<-|----------> [media] | +-----------+ | +-----------------+ Figure 1: SBC Architecture Figure 1 shows the logical architecture of an SBC, which includes a signaling and a media component. In this document, the terms outer and inner network are used for describing these two networks. An SBC is logically associated with the inner network, and it typically provides functions such as controlling and protecting access to the inner network from the outer network. The SBC itself is configured and managed by the organization operating the inner network. In some scenarios, SBCs operate with users' (implicit or explicit) consent; while in others, they operate without users' consent (this latter case can potentially cause problems). For example, if an SBC in the same administrative domain as a set of enterprise users performs topology hiding (see Section 3.1), the enterprise users can choose to route their SIP messages through it. If they choose to route through the SBC, then the SBC can be seen as having the users' implicit consent. Another example is a scenario where a service provider has broken gateways and it deploys an SBC in front of them for protocol repair reasons (see Section 3.6). Users can choose to configure the SBC as their gateway and, so, the SBC can be seen as having the users' implicit consent. Hautakorpi, et al. Informational [Page 5]
RFC 5853 Requirements from SIP SBC Deployments April 2010 2.1. Peering Scenario A typical peering scenario involves two network operators who exchange traffic with each other. An example peering scenario is illustrated in Figure 2. An originating gateway (GW-A1) in Operator A's network sends an INVITE that is routed to the SBC in Operator B's network. Then, the SBC forward it to the softswitch (SS-B). The softswitch responds with a redirect (3xx) message back to the SBC that points to the appropriate terminating gateway (GW-B1) in Operator B's network. If Operator B does not have an SBC, the redirect message would go to the Operator A's originating gateway. After receiving the redirect message, the SBC sends the INVITE to the terminating gateway. Operator A . Operator B . . 2) INVITE +-----+ . /--------------->+-----+ |SS-A | . / 3) 3xx (redir.) |SS-B | +-----+ . / /---------------+-----+ . / / +-----+ 1) INVITE +-----+--/ / +-----+ |GW-A1|---------------->| SBC |<---/ 4) INVITE |GW-B1| +-----+ +-----+---------------------->+-----+ . +-----+ . +-----+ |GW-A2| . |GW-B2| +-----+ . +-----+ Figure 2: Peering with SBC From the SBC's perspective the Operator A is the outer network, and Operator B is the inner network. Operator B can use the SBC, for example, to control access to its network, protect its gateways and softswitches from unauthorized use and denial-of-service (DoS) attacks, and monitor the signaling and media traffic. It also simplifies network management by minimizing the number of ACL (Access Control List) entries in the gateways. The gateways do not need to be exposed to the peer network, and they can restrict access (both media and signaling) to the SBCs. The SBC helps ensure that only media from sessions the SBC authorizes will reach the gateway. 2.2. Access Scenario In an access scenario, presented in Figure 3, the SBC is placed at the border between the access network (outer network) and the operator's network (inner network) to control access to the operator's network, protect its components (media servers, Hautakorpi, et al. Informational [Page 6]
RFC 5853 Requirements from SIP SBC Deployments April 2010 application servers, gateways, etc.) from unauthorized use and DoS attacks, and monitor the signaling and media traffic. Also, since the SBC is call stateful, it may provide access control functions to prevent over-subscription of the access links. Endpoints are configured with the SBC as their outbound proxy address. The SBC routes requests to one or more proxies in the operator network. Access Network Operator Network +-----+ | UA1 |<---------\ +-----+ \ \ +-----+ \------->+-----+ +-------+ | UA2 |<-------------------->| SBC |<----->| proxy |<-- - +-----+ /--->+-----+ +-------+ / +-----+ +-----+ / | UA3 +---+ NAT |<---/ +-----+ +-----+ Figure 3: Access Scenario with SBC The SBC may be hosted in the access network (e.g., this is common when the access network is an enterprise network), or in the operator network (e.g., this is common when the access network is a residential or small business network). Despite where the SBC is hosted, it is managed by the organization maintaining the operator network. Some endpoints may be behind enterprise or residential NATs. In cases where the access network is a private network, the SBC is a NAT for all traffic. It is noteworthy that SIP traffic may have to traverse more than one NAT. The proxy usually does authentication and/or authorization for registrations and outbound calls. The SBC modifies the REGISTER request so that subsequent requests to the registered address-of-record are routed to the SBC. This is done either with a Path header field [3] or by modifying the Contact to point at the SBC. The scenario presented in this section is a general one, and it applies also to other similar settings. One example from a similar setting is the one where an access network is the open internet, and the operator network is the network of a SIP service provider. Hautakorpi, et al. Informational [Page 7]
RFC 5853 Requirements from SIP SBC Deployments April 2010 3. Functions of SBCs This section lists those functions that are used in SBC deployments in current communication networks. Each subsection describes a particular function or feature, the operators' requirements for having it, explanation of any impact to the end-to-end SIP architecture, and a concrete implementation example. Each section also discusses potential concerns specific to that particular implementation technique. Suggestions for alternative implementation techniques that may be more architecturally compatible with SIP are outside the scope of this document. All the examples given in this section are simplified; only the relevant header lines from SIP and SDP (Session Description Protocol) [7] messages are displayed. 3.1. Topology Hiding 3.1.1. General Information and Requirements Topology hiding consists of limiting the amount of topology information given to external parties. Operators have a requirement for this functionality because they do not want the IP addresses of their equipment (proxies, gateways, application servers, etc.) to be exposed to outside parties. This may be because they do not want to expose their equipment to DoS attacks, they may use other carriers for certain traffic and do not want their customers to be aware of it, or they may want to hide their internal network architecture from competitors or partners. In some environments, the operator's customers may wish to hide the addresses of their equipment or the SIP messages may contain private, non-routable addresses. The most common form of topology hiding is the application of header privacy (see Section 5.1 of [2]), which involves stripping Via and Record-Route headers, replacing the Contact header, and even changing Call-IDs. However, in deployments that use IP addresses instead of domain names in headers that cannot be removed (e.g., From and To headers), the SBC may replace these IP addresses with its own IP address or domain name. For a reference, there are also other ways of hiding topology information than inserting an intermediary, like an SBC, to the signaling path. One of the ways is the UA-driven privacy mechanism [8], where the UA can facilitate the concealment of topology information. Hautakorpi, et al. Informational [Page 8]
RFC 5853 Requirements from SIP SBC Deployments April 2010 3.1.2. Architectural Issues Performing topology hiding, as described above, by SBCs that do not have the users' consent presents some issues. This functionality is based on a hop-by-hop trust model as opposed to an end-to-end trust model. The messages are modified without the subscriber's consent and could potentially modify or remove information about the user's privacy, security requirements, and higher-layer applications that are communicated end-to-end using SIP. Neither user agent in an end- to-end call has any way to distinguish the SBC actions from a man-in- the-middle (MITM) attack. The topology hiding function does not work well with Authenticated Identity Management [4] in scenarios where the SBC does not have any kind of consent from the users. The Authenticated Identity Management mechanism is based on a hash value that is calculated from parts of From, To, Call-ID, CSeq, Date, and Contact header fields plus from the whole message body. If the authentication service is not provided by the SBC itself, the modification of the aforementioned header fields and the message body is in violation of [4]. Some forms of topology hiding are in violation, because they are, e.g., replacing the Contact header of a SIP message. 3.1.3. Example The current way of implementing topology hiding consists of having an SBC act as a B2BUA (Back-to-Back User Agent) and remove all traces of topology information (e.g., Via and Record-Route entries) from outgoing messages. Imagine the following example scenario: the SBC (p4.domain.example.com) receives an INVITE request from the inner network, which in this case is an operator network. The received SIP message is shown in Figure 4. Hautakorpi, et al. Informational [Page 9]
RFC 5853 Requirements from SIP SBC Deployments April 2010 INVITE sip:callee@u2.domain.example.com SIP/2.0 Via: SIP/2.0/UDP p3.middle.example.com;branch=z9hG4bK48jq9w174131.1 Via: SIP/2.0/UDP p2.example.com;branch=z9hG4bK18an6i9234172.1 Via: SIP/2.0/UDP p1.example.com;branch=z9hG4bK39bn2e5239289.1 Via: SIP/2.0/UDP u1.example.com;branch=z9hG4bK92fj4u7283927.1 Contact: sip:caller@u1.example.com Record-Route: <sip:p3.middle.example.com;lr> Record-Route: <sip:p2.example.com;lr> Record-Route: <sip:p1.example.com;lr> Figure 4: INVITE Request Prior to Topology Hiding Then, the SBC performs a topology hiding function. In this scenario, the SBC removes and stores all existing Via and Record-Route headers, and then inserts Via and Record-Route header fields with its own SIP URI. After the topology hiding function, the message could appear as shown in Figure 5. INVITE sip:callee@u2.domain.example.com SIP/2.0 Via: SIP/2.0/UDP p4.domain.example.com;branch=z9hG4bK92es3w230129.1 Contact: sip:caller@u1.example.com Record-Route: <sip:p4.domain.example.com;lr> Figure 5: INVITE Request after Topology Hiding Like a regular proxy server that inserts a Record-Route entry, the SBC handles every single message of a given SIP dialog. If the SBC loses state (e.g., SBC restarts for some reason), it may not be able to route messages properly (note: some SBCs preserve the state information also on restart). For example, if the SBC removes Via entries from a request and then restarts, thus losing state; the SBC may not be able to route responses to that request, depending on the information that was lost when the SBC restarted. This is only one example of topology hiding. Besides topology hiding (i.e., information related to the network elements is being hidden), SBCs may also do identity hiding (i.e., information related to identity of subscribers is being hidden). While performing identity hiding, SBCs may modify Contact header field values and other header fields containing identity information. The header fields containing identity information is listed in Section 4.1 of [2]. Since the publication of [2], the following header fields containing identity information have been defined: "P-Asserted-Identity", "Referred-By", "Identity", and "Identity-Info". Hautakorpi, et al. Informational [Page 10]
RFC 5853 Requirements from SIP SBC Deployments April 2010 3.2. Media Traffic Management 3.2.1. General Information and Requirements Media traffic management is the function of controlling media traffic. Network operators may require this functionality in order to control the traffic being carried on their network on behalf of their subscribers. Traffic management helps the creation of different kinds of billing models (e.g., video telephony can be priced differently than voice-only calls) and it also makes it possible for operators to enforce the usage of selected codecs. One of the use cases for media traffic management is the implementation of intercept capabilities that are required to support audit or legal obligations. It is noteworthy that the legal obligations mainly apply to operators providing voice services, and those operators typically have infrastructure (e.g., SIP proxies acting as B2BUAs) for providing intercept capabilities even without SBCs. Since the media path is independent of the signaling path, the media may not traverse through the operator's network unless the SBC modifies the session description. By modifying the session description, the SBC can force the media to be sent through a media relay which may be co-located with the SBC. This kind of traffic management can be done, for example, to ensure a certain QoS level, or to ensure that subscribers are using only allowed codecs. It is noteworthy that the SBCs do not have direct ties to routing topology and they do not, for example, change bandwidth reservations on Traffic Engineering (TE) tunnels, nor do they have direct interaction with routing protocol. Some operators do not want to manage the traffic, but only to monitor it to collect statistics and make sure that they are able to meet any business service level agreements with their subscribers and/or partners. The protocol techniques, from the SBC's viewpoint, needed for monitoring media traffic are the same as for managing media traffic. SBCs on the media path are also capable of dealing with the "lost BYE" issue if either endpoint dies in the middle of the session. The SBC can detect that the media has stopped flowing and issue a BYE to both sides to clean up any state in other intermediate elements and the endpoints. One possible form of media traffic management is that SBCs terminate media streams and SIP dialogs by generating BYE requests. This kind of procedure can take place, for example, in a situation where the Hautakorpi, et al. Informational [Page 11]
RFC 5853 Requirements from SIP SBC Deployments April 2010 subscriber runs out of credits. Media management is needed to ensure that the subscriber cannot just ignore the BYE request generated by the SBC and continue its media sessions. 3.2.2. Architectural Issues Implementing traffic management in this manner requires the SBC to access and modify the session descriptions (i.e., offers and answers) exchanged between the user agents. Consequently, this approach does not work if user agents encrypt or integrity-protect their message bodies end-to-end. Again, messages are modified without subscriber consent, and user agents do not have any way to distinguish the SBC actions from an attack by a MITM. Furthermore, this is in violation of Authenticated Identity Management [4], see Section 3.1.2. The insertion of a media relay can prevent "non-media" uses of the media path, for example, the media path key agreement. Sometimes this type of prevention is intentional, but it is not always necessary. For example, if an SBC is used just for enabling media monitoring, but not for interception. There are some possible issues related to the media relaying. If the media relaying is not done in the correct manner, it may break functions like Explicit Congestion Notification (ECN) and Path MTU Discovery (PMTUD), for example. The media relays easily break such IP and transport layer functionalities that rely on the correct handling of the protocol fields. Some especially sensitive fields are, for example, ECN and Type of Service (ToS) fields and the Don't Fragment (DF) bit. The way in which media traffic management functions impedes innovation. The reason for the impediment is that in many cases, SBCs need to be able to support new forms of communication (e.g., extensions to the SDP protocol) before new services can be put into use, which slows the adoption of new innovations. If an SBC directs many media streams through a central point in the network, it is likely to cause a significant amount of additional traffic to a path to that central point. This might create possible bottleneck in the path. In this application, the SBC may originate messages that the user may not be able to authenticate as coming from the dialog peer or the SIP Registrar/Proxy. Hautakorpi, et al. Informational [Page 12]
RFC 5853 Requirements from SIP SBC Deployments April 2010 3.2.3. Example Traffic management may be performed in the following way: The SBC behaves as a B2BUA and inserts itself, or some other entity under the operator's control, in the media path. In practice, the SBC modifies the session descriptions carried in the SIP messages. As a result, the SBC receives media from one user agent and relays it to the other user agent and performs the identical operation with media traveling in the reverse direction. As mentioned in Section 3.2.1, codec restriction is a form of traffic management. The SBC restricts the codec set negotiated in the offer/ answer exchange [5] between the user agents. After modifying the session descriptions, the SBC can check whether or not the media stream corresponds to what was negotiated in the offer/answer exchange. If it differs, the SBC has the ability to terminate the media stream or take other appropriate (configured) actions (e.g., raise an alarm). Consider the following example scenario: the SBC receives an INVITE request from the outer network, which in this case is an access network. The received SIP message contains the SDP session descriptor shown in Figure 6. v=0 o=owner 2890844526 2890842807 IN IP4 192.0.2.4 c=IN IP4 192.0.2.4 m=audio 49230 RTP/AVP 96 98 a=rtpmap:96 L8/8000 a=rtpmap:98 L16/16000/2 Figure 6: Request Prior to Media Management In this example, the SBC performs the media traffic management function by rewriting the "m=" line, and removing one "a=" line according to some (external) policy. Figure 7 shows the session description after the traffic management function. v=0 o=owner 2890844526 2890842807 IN IP4 192.0.2.4 c=IN IP4 192.0.2.4 m=audio 49230 RTP/AVP 96 a=rtpmap:96 L8/8000 Figure 7: Request Body After Media Management Hautakorpi, et al. Informational [Page 13]
RFC 5853 Requirements from SIP SBC Deployments April 2010 Media traffic management has a problem where the SBC needs to understand the session description protocol and all extensions used by the user agents. This means that in order to use a new extension (e.g., an extension to implement a new service) or a new session description protocol, SBCs in the network may need to be upgraded in conjunction with the endpoints. It is noteworthy that a similar problem, but with header fields, applies to, for example, topology hiding function, see Section 3.1. Certain extensions that do not require active manipulation of the session descriptors to facilitate traffic management will be able to be deployed without upgrading existing SBCs, depending on the degree of transparency the SBC implementation affords. In cases requiring an SBC modification to support the new protocol features, the rate of service deployment may be affected. 3.3. Fixing Capability Mismatches 3.3.1. General Information and Requirements SBCs fixing capability mismatches enable communications between user agents with different capabilities or extensions. For example, an SBC can enable a plain SIP [1] user agent to connect to a 3GPP network, or enable a connection between user agents that support different IP versions, different codecs, or that are in different address realms. Operators have a requirement and a strong motivation for performing capability mismatch fixing, so that they can provide transparent communication across different domains. In some cases, different SIP extensions or methods to implement the same SIP application (like monitoring session liveness, call history/diversion etc.) may also be interworked through the SBC. 3.3.2. Architectural Issues SBCs that are fixing capability mismatches do it by inserting a media element into the media path using the procedures described in Section 3.2. Therefore, these SBCs have the same concerns as SBCs performing traffic management: the SBC may modify SIP messages without consent from any of the user agents. This may break end-to- end security and application extensions negotiation. The capability mismatch fixing is a fragile function in the long term. The number of incompatibilities built into various network elements is increasing the fragility and complexity over time. This might lead to a situation where SBCs need to be able to handle a large number of capability mismatches in parallel. Hautakorpi, et al. Informational [Page 14]
RFC 5853 Requirements from SIP SBC Deployments April 2010 3.3.3. Example Consider the following example scenario where the inner network is an access network using IPv4 and the outer network is using IPv6. The SBC receives an INVITE request with a session description from the access network: INVITE sip:callee@ipv6.domain.example.com SIP/2.0 Via: SIP/2.0/UDP 192.0.2.4 Contact: sip:caller@u1.example.com v=0 o=owner 2890844526 2890842807 IN IP4 192.0.2.4 c=IN IP4 192.0.2.4 m=audio 49230 RTP/AVP 96 a=rtpmap:96 L8/8000 Figure 8: Request Prior to Capabilities Match Then, the SBC performs a capability mismatch fixing function. In this scenario, the SBC inserts Record-Route and Via headers and rewrites the "c=" line from the sessions descriptor. Figure 9 shows the request after the capability mismatch adjustment. INVITE sip:callee@ipv6.domain.com SIP/2.0 Record-Route: <sip:[2001:DB8::801:201:2ff:fe94:8e10];lr> Via: SIP/2.0/UDP sip:[2001:DB8::801:201:2ff:fe94:8e10] Via: SIP/2.0/UDP 192.0.2.4 Contact: sip:caller@u1.example.com v=0 o=owner 2890844526 2890842807 IN IP4 192.0.2.4 c=IN IP6 2001:DB8::801:201:2ff:fe94:8e10 m=audio 49230 RTP/AVP 96 a=rtpmap:96 L8/8000 Figure 9: Request after Capability Match This message is then sent by the SBC to the onward IPv6 network. 3.4. Maintaining SIP-Related NAT Bindings 3.4.1. General Information and Requirements NAT traversal in this instance refers to the specific message modifications required to assist a user agent in maintaining SIP and media connectivity when there is a NAT device located between a user agent and a proxy/registrar and, possibly, any other user agent. The Hautakorpi, et al. Informational [Page 15]
RFC 5853 Requirements from SIP SBC Deployments April 2010 primary purpose of NAT traversal function is to keep up a control connection to user agents behind NATs. This can, for example, be achieved by generating periodic network traffic that keeps bindings in NATs alive. SBCs' NAT traversal function is required in scenarios where the NAT is outside the SBC (i.e., not in cases where SBC itself acts as a NAT). An SBC performing a NAT (Network Address Translator) traversal function for a user agent behind a NAT sits between the user agent and the registrar of the domain. NATs are widely deployed in various access networks today, so operators have a requirement to support it. When the registrar receives a REGISTER request from the user agent and responds with a 200 (OK) response, the SBC modifies such a response decreasing the validity of the registration (i.e., the registration expires sooner). This forces the user agent to send a new REGISTER to refresh the registration sooner that it would have done on receiving the original response from the registrar. The REGISTER requests sent by the user agent refresh the binding of the NAT before the binding expires. Note that the SBC does not need to relay all the REGISTER requests received from the user agent to the registrar. The SBC can generate responses to REGISTER requests received before the registration is about to expire at the registrar. Moreover, the SBC needs to deregister the user agent if this fails to refresh its registration in time, even if the registration at the registrar would still be valid. SBCs can also force traffic to go through a media relay for NAT traversal purposes (more about media traffic management in Section 3.2). A typical call has media streams to two directions. Even though SBCs can force media streams from both directions to go through a media relay, in some cases, it is enough to relay only the media from one direction (e.g., in a scenario where only the other endpoint is behind a NAT). 3.4.2. Architectural Issues This approach to NAT traversal does not work if end-to-end confidentiality or integrity-protection mechanisms are used (e.g., Secure/Multipurpose Internet Mail Extensions (S/MIME)). The SBC would be seen as a MITM modifying the messages between the user agent and the registrar. There is also a problem related to the method of how SBCs choose the value for the validity of a registration period. This value should be as high as possible, but it still needs to be low enough to maintain the NAT binding. Some SBCs do not have any deterministic Hautakorpi, et al. Informational [Page 16]
RFC 5853 Requirements from SIP SBC Deployments April 2010 method for choosing a suitable value. However, SBCs can just use a sub-optimal, relatively small value that usually works. An example from such value is 15 seconds (see [9]). NAT traversal for media using SBCs poses few issues as well. For example, an SBC normally guesses the recipient's public IP address on one of the media streams relayed by the SBC by snooping on the source IP address of another media stream relayed by the same SBC. This causes security and interoperability issues since the SBC can end up associating wrong destination IP addresses on media streams it is relaying. For example, an attacker may snoop on the local IP address and ports used by the SBC for media relaying the streams and send a few packets from a malicious IP address to these destinations. In most cases, this can cause media streams in the opposite directions to divert traffic to the attacker resulting in a successful MITM or DoS attack. A similar example of an interoperability issue is caused when an endpoint behind a NAT attempts to switch the IP address of the media streams by using a re-INVITE. If any media packets are re- ordered or delayed in the network, they can cause the SBC to block the switch from happening even if the re-INVITE successfully goes through. 3.4.3. Example Consider the following example scenario: The SBC resides between the UA and Registrar. Previously, the UA has sent a REGISTER request to the Registrar, and the SBC receives the registration response shown in Figure 10. SIP/2.0 200 OK From: Bob <sip:bob@biloxi.example.com>;tag=a73kszlfl To: Bob <sip:bob@biloxi.example.com>;tag=34095828jh CSeq: 1 REGISTER Contact: <sips:bob@client.biloxi.example.com>;expires=3600 Figure 10: Response Prior to NAT Maintenance Function When performing the NAT traversal function, the SBC may rewrite the expiry time to coax the UA to re-register prior to the intermediating NAT deciding to close the pinhole. Figure 11 shows a possible modification of the response from Figure 10. Hautakorpi, et al. Informational [Page 17]
RFC 5853 Requirements from SIP SBC Deployments April 2010 SIP/2.0 200 OK From: Bob <sip:bob@biloxi.example.com>;tag=a73kszlfl To: Bob <sip:bob@biloxi.example.com>;tag=34095828jh CSeq: 1 REGISTER Contact: <sips:bob@client.biloxi.example.com>;expires=60 Figure 11: Manipulated Response for NAT Traversal Naturally, other measures could be taken in order to enable the NAT traversal (e.g., non-SIP keepalive messages), but this example illustrates only one mechanism for preserving the SIP-related NAT bindings. 3.5. Access Control 3.5.1. General Information and Requirements Network operators may wish to control what kind of signaling and media traffic their network carries. There is strong motivation and a requirement to do access control on the edge of an operator's network. Access control can be based on, for example, link-layer identifiers, IP addresses or SIP identities. This function can be implemented by protecting the inner network with firewalls and configuring them so that they only accept SIP traffic from the SBC. This way, all the SIP traffic entering the inner network needs to be routed though the SBC, which only routes messages from authorized parties or traffic that meets a specific policy that is expressed in the SBC administratively. Access control can be applied to either only the signaling or both the signaling and media. If it is applied only to the signaling, then the SBC might behave as a proxy server. If access control is applied to both the signaling and media, then the SBC behaves in a similar manner as explained in Section 3.2. A key part of media- layer access control is that only media for authorized sessions is allowed to pass through the SBC and/or associated media relay devices. Operators implement some functionalities, like NAT traversal for example, in an SBC instead of other elements in the inner network for several reasons: (i) preventing packets from unregistered users to prevent chances of DoS attack, (ii) prioritization and/or re-routing of traffic (based on user or service, like E911) as it enters the network, and (iii) performing a load balancing function or reducing the load on other network equipment. Hautakorpi, et al. Informational [Page 18]
RFC 5853 Requirements from SIP SBC Deployments April 2010 In environments where there is limited bandwidth on the access links, the SBC can compute the potential bandwidth use by examining the codecs present in SDP offers and answers. With this information, the SBC can reject sessions before the available bandwidth is exhausted to allow existing sessions to maintain acceptable quality of service. Otherwise, the link could become over-subscribed and all sessions would experience a deterioration in quality of service. SBCs may contact a policy server to determine whether sufficient bandwidth is available on a per-session basis. 3.5.2. Architectural Issues Since the SBC needs to handle all SIP messages, this function has scalability implications. In addition, the SBC is a single point of failure from an architectural point of view. Although, in practice, many current SBCs have the capability to support redundant configuration, which prevents the loss of calls and/or sessions in the event of a failure on a single node. If access control is performed only on behalf of signaling, then the SBC is compatible with general SIP architectural principles, but if it is performed for signaling and for media, then there are similar problems as described in Section 3.2.2. 3.5.3. Example Figure 12 shows a callflow where the SBC is providing both signaling and media access control (ACKs omitted for brevity). Hautakorpi, et al. Informational [Page 19]
RFC 5853 Requirements from SIP SBC Deployments April 2010 caller SBC callee | | | | Identify the caller | | |<- - - - - - - - - - - >| | | | | | INVITE + SDP | | |----------------------->| | | [Modify the SDP] | | | INVITE + modified SDP | | |----------------------->| | | | | | 200 OK + SDP | | |<-----------------------| | [Modify the SDP] | | | | | 200 OK + modified SDP | | |<-----------------------| | | | | | Media [Media inspection] Media | |<======================>|<======================>| | | | Figure 12: Example Access Callflow In this scenario, the SBC first identifies the caller, so it can determine whether or not to give signaling access to the caller. This might be achieved using information gathered during registration, or by other means. Some SBCs may rely on the proxy to authenticate the user agent placing the call. After identification, the SBC modifies the session descriptors in INVITE and 200 OK messages in a way so that the media is going to flow through the SBC itself. When the media starts flowing, the SBC can inspect whether the callee and caller use the codec(s) upon which they had previously agreed. 3.6. Protocol Repair 3.6.1. General Information and Requirements SBCs are also used to repair protocol messages generated by not- fully-standard-compliant or badly implemented clients. Operators may wish to support protocol repair, if they want to support as many clients as possible. It is noteworthy that this function affects only the signaling component of an SBC, and that the protocol repair function is not the same as protocol conversion (i.e., making translation between two completely different protocols). Hautakorpi, et al. Informational [Page 20]
RFC 5853 Requirements from SIP SBC Deployments April 2010 3.6.2. Architectural Issues In many cases, doing protocol repair for SIP header fields can be seen as being compatible with SIP architectural principles, and it does not violate the end-to-end model of SIP. The SBC repairing protocol messages behaves as a proxy server that is liberal in what it accepts and strict in what it sends. However, protocol repair may break security mechanism that do cryptographical computations on SIP header values. Attempting protocol repair for SIP message bodies (SDP) is incompatible with Authenticated Identity Management [4] and end-to-end security mechanisms such as S/MIME. A similar problem related to increasing complexity, as explained in Section 3.3.2, also affects protocol repair function. 3.6.3. Examples The SBC can, for example, receive an INVITE message from a relatively new SIP UA as illustrated in Figure 13. INVITE sip:callee@sbchost.example.com Via: SIP/2.0/UDP u1.example.com:5060;lr From: Caller <sip:caller@one.example.com> To: Callee <sip:callee@two.example.com> Call-ID: 18293281@u1.example.com CSeq: 1 INVITE Contact: sip:caller@u1.example.com Figure 13: Request from a Relatively New Client If the SBC does protocol repair, it can rewrite the 'lr' parameter on the Via header field into the form 'lr=true' in order to support some older, badly implemented SIP stacks. It could also remove excess white spaces to make the SIP message more human readable. 3.7. Media Encryption 3.7.1. General Information and Requirements SBCs are used to perform media encryption/decryption at the edge of the network. This is the case when media encryption (e.g., Secure Real-time Transport Protocol (SRTP)) is used only on the access network (outer network) side and the media is carried unencrypted in the inner network. Some operators provide the ability to do legal Hautakorpi, et al. Informational [Page 21]
RFC 5853 Requirements from SIP SBC Deployments April 2010 interception while still giving their customers the ability to encrypt media in the access network. One possible way to do this is to perform media encryption function. 3.7.2. Architectural Issues While performing a media encryption function, SBCs need to be able to inject either themselves, or some other entity to the media path. It must be noted that this kind of behavior is the same as a classical MITM attack. Due to this, the SBCs have the same architectural issues as explained in Section 3.2. 3.7.3. Example Figure 14 shows an example where the SBC is performing media- encryption-related functions (ACKs omitted for brevity). caller SBC#1 SBC#2 callee | | | | | INVITE + SDP | | | |------------------->| | | | [Modify the SDP] | | | | | | | | INVITE + mod. SDP | | | |------------------->| | | | [Modify the SDP] | | | | | | | | INVITE + mod. SDP | | | |------------------->| | | | | | | | 200 OK + SDP | | | |<-------------------| | | [Modify the SDP] | | | | | | | 200 OK + mod. SDP | | | |<-------------------| | | [Modify the SDP] | | | | | | | 200 OK + mod. SDP | | | |<-------------------| | | | | | | | Encrypted | Plain | Encrypted | | media [enc./dec.] media [enc./dec.] media | |<==================>|<- - - - - - - - ->|<==================>| | | | | Figure 14: Media Encryption Example Hautakorpi, et al. Informational [Page 22]
RFC 5853 Requirements from SIP SBC Deployments April 2010 First, the UAC sends an INVITE request, and the first SBC modifies the session descriptor in a way that it injects itself to the media path. The same happens in the second SBC. Then, the User Agent Server (UAS) replies with a 200 OK response and the SBCs inject themselves in the returning media path. After signaling, the media starts flowing, and both SBCs perform media encryption and decryption. 4. Derived Requirements for Future SIP Standardization Work Some of the functions listed in this document are more SIP-unfriendly than others. This list of requirements is derived from the functions that break the principles of SIP in one way or another when performed by SBCs that do not have the users' consent. The derived requirements are: Req-1: There should be a SIP-friendly way to hide network topology information. Currently, this is done by stripping and replacing header fields, which is against the principles of SIP on behalf of some header fields (see Section 3.1). Req-2: There should be a SIP-friendly way to direct media traffic through intermediaries. Currently, this is done by modifying session descriptors, which is against the principles of SIP (see Sections 3.2, 3.4, 3.5, and 3.7). Req-3: There should be a SIP-friendly way to fix capability mismatches in SIP messages. This requirement is harder to fulfill on complex mismatch cases, like the 3GPP/SIP [1] network mismatch. Currently, this is done by modifying SIP messages, which may violate end-to-end security (see Sections 3.3 and 3.6), on behalf of some header fields. Req-1 and Req-3 do not have an existing, standardized solution today. There is ongoing work in the IETF for addressing Req-2, such as SIP session policies [10], Traversal Using Relays around NAT (TURN) [11], and Interactive Connectivity Establishment (ICE) [12]. Nonetheless, future work is needed in order to develop solutions to these requirements. 5. Security Considerations Many of the functions this document describes have important security and privacy implications. One major security problem is that many functions implemented by SBCs (e.g., topology hiding and media traffic management) modify SIP messages and their bodies without the user agents' consent. The result is that the user agents may Hautakorpi, et al. Informational [Page 23]
RFC 5853 Requirements from SIP SBC Deployments April 2010 interpret the actions taken by an SBC as a MITM attack. SBCs modify SIP messages because it allows them to, for example, protect elements in the inner network from direct attacks. SBCs that place themselves (or another entity) on the media path can be used to eavesdrop on conversations. Since, often, user agents cannot distinguish between the actions of an attacker and those of an SBC, users cannot know whether they are being eavesdropped on or if an SBC on the path is performing some other function. SBCs place themselves on the media path because it allows them to, for example, perform legal interception. On a general level, SBCs prevent the use of end-to-end authentication. This is because SBCs need to be able to perform actions that look like MITM attacks, and in order for user agents to communicate, they must allow those type of attacks. It other words, user agents cannot use end-to-end security. This is especially harmful because other network elements, besides SBCs, are then able to do similar attacks. However, in some cases, user agents can establish encrypted media connections between one another. One example is a scenario where SBC is used for enabling media monitoring but not for interception. An SBC is a single point of failure from the architectural point of view. This makes it an attractive target for DoS attacks. The fact that some functions of SBCs require those SBCs to maintain session- specific information makes the situation even worse. If the SBC crashes (or is brought down by an attacker), ongoing sessions experience undetermined behavior. If the IETF decides to develop standard mechanisms to address the requirements presented in Section 4, the security and privacy-related aspects of those mechanisms will, of course, need to be taken into consideration. 6. Acknowledgements The ad hoc meeting about SBCs, held on November 9, 2004 in Washington DC during the 61st IETF meeting, provided valuable input to this document. The authors would also like to thank Sridhar Ramachandran, Gaurav Kulshreshtha, and Rakendu Devdhar. Reviewers Spencer Dawkins and Francois Audet also deserve special thanks. Hautakorpi, et al. Informational [Page 24]
RFC 5853 Requirements from SIP SBC Deployments April 2010 7. References 7.1. Normative References [1] 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. [2] Peterson, J., "A Privacy Mechanism for the Session Initiation Protocol (SIP)", RFC 3323, November 2002. [3] Willis, D. and B. Hoeneisen, "Session Initiation Protocol (SIP) Extension Header Field for Registering Non-Adjacent Contacts", RFC 3327, December 2002. [4] Peterson, J. and C. Jennings, "Enhancements for Authenticated Identity Management in the Session Initiation Protocol (SIP)", RFC 4474, August 2006. [5] Rosenberg, J. and H. Schulzrinne, "An Offer/Answer Model with Session Description Protocol (SDP)", RFC 3264, June 2002. 7.2. Informative References [6] 3GPP, "IP Multimedia Subsystem (IMS); Stage 2", 3GPP TS 23.228 10.0.0, March 2010. [7] Handley, M., Jacobson, V., and C. Perkins, "SDP: Session Description Protocol", RFC 4566, July 2006. [8] Munakata, M., Schubert, S., and T. Ohba, "User-Agent-Driven Privacy Mechanism for SIP", RFC 5767, April 2010. [9] Eggert, L. and G. Fairhurst, "Unicast UDP Usage Guidelines for Application Designers", BCP 145, RFC 5405, November 2008. [10] Hilt, V., Camarillo, G., and J. Rosenberg, "A Framework for Session Initiation Protocol (SIP) Session Policies", Work in Progress, February 2010. [11] 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. [12] Rosenberg, J., "Interactive Connectivity Establishment (ICE): A Protocol for Network Address Translator (NAT) Traversal for Offer/Answer Protocols", RFC 5245, MonthTBD 2010. Hautakorpi, et al. Informational [Page 25]
RFC 5853 Requirements from SIP SBC Deployments April 2010 Authors' Addresses Jani Hautakorpi (editor) Ericsson Hirsalantie 11 Jorvas 02420 Finland EMail: Jani.Hautakorpi@ericsson.com Gonzalo Camarillo Ericsson Hirsalantie 11 Jorvas 02420 Finland EMail: Gonzalo.Camarillo@ericsson.com Robert F. Penfield Acme Packet 71 Third Avenue Burlington, MA 01803 US EMail: bpenfield@acmepacket.com Alan Hawrylyshen Skype, Inc. 2055 E. Hamilton Ave San Jose, CA 95125 US EMail: alan.ietf@polyphase.ca Medhavi Bhatia 3CLogic 9700 Great Seneca Hwy. Rockville, MD 20850 US EMail: mbhatia@3clogic.com Hautakorpi, et al. Informational [Page 26]