Internet Engineering Task Force (IETF) B. Ver Steeg Request for Comments: 6285 A. Begen Category: Standards Track Cisco ISSN: 2070-1721 T. Van Caenegem Alcatel-Lucent Z. Vax Magnum Semiconductor June 2011
Unicast-Based Rapid Acquisition of Multicast RTP Sessions
When an RTP receiver joins a multicast session, it may need to acquire and parse certain Reference Information before it can process any data sent in the multicast session. Depending on the join time, length of the Reference Information repetition (or appearance) interval, size of the Reference Information, and the application and transport properties, the time lag before an RTP receiver can usefully consume the multicast data, which we refer to as the Acquisition Delay, varies and can be large. This is an undesirable phenomenon for receivers that frequently switch among different multicast sessions, such as video broadcasts.
In this document, we describe a method using the existing RTP and RTP Control Protocol (RTCP) machinery that reduces the acquisition delay. In this method, an auxiliary unicast RTP session carrying the Reference Information to the receiver precedes or accompanies the multicast stream. This unicast RTP flow can be transmitted at a faster than natural bitrate to further accelerate the acquisition. The motivating use case for this capability is multicast applications that carry real-time compressed audio and video. However, this method can also be used in other types of multicast applications where the acquisition delay is long enough to be a problem.
Status of This Memo
This is an Internet Standards Track document.
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). Further information on Internet Standards is available in Section 2 of RFC 5741.
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 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.
This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English.
Most multicast flows carry a stream of inter-related data. Receivers need to acquire certain information to start processing any data sent in the multicast session. This document refers to this information as Reference Information. The Reference Information is conventionally sent periodically in the multicast session (although its content can change over time) and usually consists of items such as a description of the schema for the rest of the data, references to which data to process, encryption information including keys, and any other information required to process the data in the multicast stream [IC2009].
Real-time multicast applications require receivers to buffer data. Receivers may have to buffer data to smooth out the network jitter, to allow loss-repair methods such as Forward Error Correction and retransmission to recover the missing packets, and to satisfy the data-processing requirements of the application layer.
When a receiver joins a multicast session, it has no control over what point in the flow is currently being transmitted. Sometimes the receiver might join the session right before the Reference
Ver Steeg, et al. Standards Track [Page 3]
RFC 6285 RAMS June 2011
Information is sent in the session. In this case, the required waiting time is usually minimal. Other times, the receiver might join the session right after the Reference Information has been transmitted. In this case, the receiver has to wait for the Reference Information to appear again in the flow before it can start processing any multicast data. In some other cases, the Reference Information is not contiguous in the flow but dispersed over a large period, which forces the receiver to wait for the whole Reference Information to arrive before starting to process the rest of the data.
The net effect of waiting for the Reference Information and waiting for various buffers to fill up is that receivers can experience significantly large delays in data processing. In this document, we refer to the difference between the time an RTP receiver wants to join the multicast session and the time the RTP receiver acquires all the necessary Reference Information as the Acquisition Delay. The acquisition delay might not be the same for different receivers; it usually varies depending on the join time, length of the Reference Information repetition (or appearance) interval, and size of the Reference Information, as well as the application and transport properties.
The varying nature of the acquisition delay adversely affects the receivers that frequently switch among multicast sessions. While this problem equally applies to both any-source multicast (ASM) and source-specific multicast (SSM) applications, in this specification we address it for the SSM-based applications by describing a method that uses the fundamental tools offered by the existing RTP and RTCP protocols [RFC3550]. In this method, either the multicast source (or the distribution source in an SSM session) retains the Reference Information for a period after its transmission, or an intermediary network element (that we refer to as Retransmission Server) joins the multicast session and continuously caches the Reference Information as it is sent in the session and acts as a feedback target (see [RFC5760]) for the session. When an RTP receiver wishes to join the same multicast session, instead of simply issuing a Source Filtering Group Management Protocol (SFGMP) Join message, it sends a request to the feedback target for the session and asks for the Reference Information. The retransmission server starts a new unicast RTP (retransmission) session and sends the Reference Information to the RTP receiver over that session. If there is residual bandwidth, the retransmission server might burst the Reference Information faster than its natural rate. As soon as the receiver acquires the Reference Information, it can join the multicast session and start processing the multicast data. A simplified network diagram showing this method through an intermediary network element is depicted in Figure 1.
Ver Steeg, et al. Standards Track [Page 4]
RFC 6285 RAMS June 2011
This method potentially reduces the acquisition delay. We refer to this method as Unicast-Based Rapid Acquisition of Multicast RTP Sessions. A primary use case for this method is to reduce the channel-change times in IPTV networks where compressed video streams are multicast in different SSM sessions and viewers randomly join these sessions.
Figure 1: Rapid Acquisition through an Intermediary Network Element
A principle design goal in this solution is to use the existing tools in the RTP/RTCP protocol family. This improves the versatility of the existing implementations and promotes faster deployment and better interoperability. To this effect, we use the unicast retransmission support of RTP [RFC4588] and the capabilities of RTCP to handle the signaling needed to accomplish the acquisition.
A reasonable effort has been made in this specification to design a solution that reliably works in both engineered and best-effort networks. However, a proper congestion control combined with the desired behavior of this solution is difficult to achieve. Rather, this solution has been designed based on the assumption that the retransmission server and the RTP receivers have some knowledge about where the bottleneck between them is. This assumption does not generally hold unless both the retransmission server and the RTP receivers are in the same edge network. Thus, this solution should
Ver Steeg, et al. Standards Track [Page 5]
RFC 6285 RAMS June 2011
not be used across any best-effort path of the Internet. Furthermore, this solution should only be used in networks that are already carrying non-congestion-responsive multicast traffic and have throttling mechanisms in the retransmission servers to ensure the (unicast) burst traffic is a known constant upper-bound multiplier on the multicast load.
In this memo, we describe a protocol that handles the rapid acquisition of a single multicast RTP session (called a primary multicast RTP session) carrying one or more RTP streams (called primary multicast streams). If desired, multiple instances of this protocol may be run in parallel to acquire multiple RTP sessions simultaneously.
When an RTP receiver requests the Reference Information from the retransmission server, it can opt to rapidly acquire a specific subset of the available RTP streams in the primary multicast RTP session. Alternatively, the RTP receiver can request the rapid acquisition of all of the RTP streams in that RTP session. Regardless of how many RTP streams are requested by the RTP receiver or how many will be actually sent by the retransmission server, only one unicast RTP session will be established by the retransmission server. This unicast RTP session is separate from the associated primary multicast RTP session. As a result, there are always two different RTP sessions in a single instance of the rapid acquisition protocol: (i) the primary multicast RTP session with its associated unicast feedback and (ii) the unicast RTP session.
If the RTP receiver wants to rapidly acquire multiple RTP sessions simultaneously, separate unicast RTP sessions will be established for each of them.
The rest of this specification is as follows. Section 3 provides a list of the definitions frequently used in this document. In Section 4, we describe the delay components in generic multicast applications. Section 5 presents an overview of the protocol design considerations for rapid acquisition. We provide the protocol details of the rapid acquisition method in Sections 6 and 7. Sections 8 and 9 discuss the Session Description Protocol (SDP) signaling issues with examples and NAT-related issues, respectively. Finally, Section 10 discusses the security considerations, and Section 11 details the IANA considerations.
This document uses the following acronyms and definitions frequently:
(Primary) SSM Session: The multicast session to which RTP receivers can join at a random point in time. A primary SSM session can carry multiple RTP streams.
Primary Multicast RTP Session: The multicast RTP session an RTP receiver is interested in acquiring rapidly. From the RTP receiver's viewpoint, the primary multicast RTP session has one associated unicast RTCP feedback stream to a Feedback Target, in addition to the primary multicast RTP stream(s).
Primary Multicast (RTP) Streams: The RTP stream(s) carried in the primary multicast RTP session.
Source Filtering Group Management Protocol (SFGMP): Following the definition in [RFC4604], SFGMP refers to the Internet Group Management Protocol (IGMP) version 3 [RFC3376] and the Multicast Listener Discovery Protocol (MLD) version 2 [RFC3810] in the IPv4 and IPv6 networks, respectively. However, the rapid acquisition method introduced in this document does not depend on a specific version of either of these group management protocols. In the remainder of this document, SFGMP will refer to any group management protocol that has Join and Leave functionalities.
Feedback Target (FT): Unicast RTCP feedback target as defined in [RFC5760]. FT_Ap denotes a specific feedback target running on a particular address and port.
Retransmission (or Burst) Packet: An RTP packet that is formatted as defined in Section 4 of [RFC4588]. The payload of a retransmission or burst packet comprises the retransmission payload header followed by the payload of the original RTP packet.
Reference Information: The set of certain media content and metadata information that is sufficient for an RTP receiver to start usefully consuming a media stream. The meaning, format, and size of this information are specific to the application and are out of the scope of this document.
Ver Steeg, et al. Standards Track [Page 7]
RFC 6285 RAMS June 2011
Preamble Information: A more compact form of the whole or a subset of the Reference Information transmitted out-of-band.
(Unicast) Burst (or Retransmission) RTP Session: The unicast RTP session used to send one or more unicast burst RTP streams and their associated RTCP messages. The terms "burst RTP session" and "retransmission RTP session" can be used interchangeably.
(Unicast) Burst (Stream): A unicast stream of RTP retransmission packets that enable an RTP receiver to rapidly acquire the Reference Information associated with a primary multicast stream. Each burst stream is identified by its Synchronization Source (SSRC) identifier that is unique in the primary multicast RTP session. Following the session-multiplexing guidelines in [RFC4588], each unicast burst stream will use the same SSRC and Canonical Name (CNAME) as its primary multicast RTP stream.
Retransmission Server (RS): The RTP/RTCP endpoint that can generate the retransmission packets and the burst streams. The RS may also generate other non-retransmission packets to aid rapid acquisition.
In a source-specific multicast (SSM) delivery system, there are three major elements that contribute to the acquisition delay when an RTP receiver switches from one multicast session to another one. These are:
o Multicast-switching delay
o Reference Information latency
o Buffering delays
Multicast-switching delay is the delay that is experienced when leaving the current multicast session (if any) and joining the new multicast session. In typical systems, the multicast join and leave operations are handled by a group management protocol. For example, the receivers and routers participating in a multicast session can use the Internet Group Management Protocol (IGMP) version 3 [RFC3376] or the Multicast Listener Discovery Protocol (MLD) version 2 [RFC3810]. In either of these protocols, when a receiver wants to join a multicast session, it sends a message to its upstream router and the routing infrastructure sets up the multicast forwarding state to deliver the packets of the multicast session to the new receiver. The join times vary depending on the proximity of the upstream router, the current state of the multicast tree, the load on the system, and the protocol implementation. Current systems provide
Ver Steeg, et al. Standards Track [Page 8]
RFC 6285 RAMS June 2011
join latencies, usually less than 200 milliseconds (ms). If the receiver had been participating in another multicast session before joining the new session, it needs to send a Leave message to its upstream router to leave the session. In common multicast routing protocols, the leave times are usually smaller than the join times; however, it is possible that the Leave and Join messages might get lost, in which case the multicast-switching delay inevitably increases.
Reference Information latency is the time it takes the receiver to acquire the Reference Information. It is highly dependent on the proximity of the actual time the receiver joined the session to the next time the Reference Information will be sent to the receivers in the session, whether or not the Reference Information is sent contiguously, and the size of the Reference Information. For some multicast flows, there is a little or no interdependency in the data, in which case the Reference Information latency will be nil or negligible. For other multicast flows, there is a high degree of interdependency. One example of interest is the multicast flows that carry compressed audio/video. For these flows, the Reference Information latency can become quite large and be a major contributor to the overall delay.
The buffering component of the overall acquisition delay is driven by the way the application layer processes the payload. In many multicast applications, an unreliable transport protocol such as UDP [RFC0768] is often used to transmit the data packets, and the reliability, if needed, is usually addressed through other means such as Forward Error Correction (e.g., [RFC6015]) and retransmission. These loss-repair methods require buffering at the receiver side to function properly. In many applications, it is also often necessary to de-jitter the incoming data packets before feeding them to the application. The de-jittering process also increases the buffering delays. Besides these network-related buffering delays, there are also specific buffering needs that are required by the individual applications. For example, standard video decoders typically require a certain amount, sometimes up to a few seconds, of coded video data to be available in the pre-decoding buffers prior to starting to decode the video bitstream.
Ver Steeg, et al. Standards Track [Page 9]
RFC 6285 RAMS June 2011
5. Protocol Design Considerations and Their Effect on Resource Management for Rapid Acquisition
This section is for informational purposes and does not contain requirements for implementations.
Rapid acquisition is an optimization of a system that is expected to continue to work correctly and properly whether or not the optimization is effective or even fails due to lost control and feedback messages, congestion, or other problems. This is fundamental to the overall design requirements surrounding the protocol definition and to the resource management schemes to be employed together with the protocol (e.g., Quality of Service (QoS) machinery, server load management, etc). In particular, the system needs to operate within a number of constraints:
o First, a rapid acquisition operation must fail gracefully. The user experience must not be significantly worse for trying and failing to complete rapid acquisition compared to simply joining the multicast session.
o Second, providing the rapid acquisition optimizations must not cause collateral damage to either the multicast session being joined or other multicast sessions sharing resources with the rapid acquisition operation. In particular, the rapid acquisition operation must avoid interference with the multicast session that might be simultaneously being received by other hosts. In addition, it must also avoid interference with other multicast and non-multicast sessions sharing the same network resources. These properties are possible but are usually difficult to achieve.
One challenge is the existence of multiple bandwidth bottlenecks between the receiver and the server(s) in the network providing the rapid acquisition service. In commercial IPTV deployments, for example, bottlenecks are often present in the aggregation network connecting the IPTV servers to the network edge, the access links (e.g., DSL, Data Over Cable Service Interface Specification (DOCSIS)), and the home network of the subscribers. Some of these links might serve only a single subscriber, limiting congestion impact to the traffic of only that subscriber, but others can be shared links carrying multicast sessions of many subscribers. Also note that the state of these links can vary over time. The receiver might have knowledge of a portion of this network or might have partial knowledge of the entire network. The methods employed by the devices to acquire this network state information is out of the scope of this document. The receiver should be able to signal the server with the bandwidth that it believes it can handle. The server also needs to be able to rate limit the flow in order to stay within the
Ver Steeg, et al. Standards Track [Page 10]
RFC 6285 RAMS June 2011
performance envelope that it knows about. Both the server and receiver need to be able to inform the other of changes in the requested and delivered rates. However, the protocol must be robust in the presence of packet loss, so this signaling must include the appropriate default behaviors.
A second challenge is that for some uses (e.g., high-bitrate video) the unicast burst bitrate is high while the flow duration of the unicast burst is short. This is because the purpose of the unicast burst is to allow the RTP receiver to join the multicast quickly and thereby limit the overall resources consumed by the burst. Such high-bitrate, short-duration flows are not amenable to conventional admission-control techniques. For example, end-to-end per-flow signaled admission-control techniques such as Resource Reservation Protocol (RSVP) have too much latency and control channel overhead to be a good fit for rapid acquisition. Similarly, using a TCP (or TCP- like) approach with a 3-way handshake and slow-start to avoid inducing congestion would defeat the purpose of attempting rapid acquisition in the first place by introducing many round-trip times (RTTs) of delay.
These observations lead to certain unavoidable requirements and goals for a rapid acquisition protocol. These are:
o The protocol must be designed to allow a deterministic upper bound on the extra bandwidth used (compared to just joining the multicast session). A reasonable size bound is e*B, where B is the nominal bandwidth of the primary multicast streams and e is an excess-bandwidth coefficient. The total duration of the unicast burst must have a reasonable bound; long unicast bursts devolve to the bandwidth profile of multi-unicast for the whole system.
o The scheme should minimize (or better eliminate) the overlap of the unicast burst and the primary multicast stream. This minimizes the window during which congestion could be induced on a bottleneck link compared to just carrying the multicast or unicast packets alone.
o The scheme must minimize (or better eliminate) any gap between the unicast burst and the primary multicast stream, which has to be repaired later or, in the absence of repair, will result in loss being experienced by the application.
In addition to the above, there are some other protocol design issues to be considered. First, there is at least one RTT of "slop" in the control loop. In starting a rapid acquisition burst, this manifests as the time between the client requesting the unicast burst and the burst description and/or the first unicast burst packets arriving at
Ver Steeg, et al. Standards Track [Page 11]
RFC 6285 RAMS June 2011
the receiver. For managing and terminating the unicast burst, there are two possible approaches for the control loop. First, the receiver can adapt to the unicast burst as received, converge based on observation, and explicitly terminate the unicast burst with a second control loop exchange (which takes a minimum of one RTT, just as starting the unicast burst does). Alternatively, the server generating the unicast burst can precompute the burst parameters based on the information in the initial request and tell the receiver the burst duration.
The protocol described in the next section allows either method of controlling the rapid acquisition unicast burst.
6. Rapid Acquisition of Multicast RTP Sessions (RAMS)
We start this section with an overview of the Rapid Acquisition of Multicast RTP Sessions (RAMS) method.
[RFC5760] specifies an extension to the RTP Control Protocol (RTCP) to use unicast feedback in an SSM session. It defines an architecture that introduces the concept of Distribution Source, which, in an SSM context, distributes the RTP data and redistributes RTCP information to all RTP receivers. This RTCP information is retrieved from the Feedback Target, to which RTCP unicast feedback traffic is sent. One or more entities different from the Distribution Source MAY implement the feedback target functionality, and different RTP receivers MAY use different feedback targets.
This document builds further on these concepts to reduce the acquisition delay when an RTP receiver joins a multicast session at a random point in time by introducing the concept of the Burst Source and new RTCP feedback messages. The Burst Source has a cache where the most recent packets from the primary multicast RTP session are continuously stored. When an RTP receiver wants to receive a primary multicast stream, it can first request a unicast burst from the Burst Source before it joins the SSM session. In this burst, the packets are formatted as RTP retransmission packets [RFC4588] and carry Reference Information. This information allows the RTP receiver to start usefully consuming the RTP packets sent in the primary multicast RTP session.
Using an accelerated bitrate (as compared to the nominal bitrate of the primary multicast stream) for the unicast burst implies that at a certain point in time, the payload transmitted in the unicast burst is going to be the same as the payload in the associated multicast stream, i.e., the unicast burst will catch up with the primary
Ver Steeg, et al. Standards Track [Page 12]
RFC 6285 RAMS June 2011
multicast stream. At this point, the RTP receiver no longer needs to receive the unicast burst and can join the SSM session. This method is referred to as the Rapid Acquisition of Multicast RTP Sessions (RAMS).
This document defines extensions to [RFC4585] for an RTP receiver to request a unicast burst as well as for additional control messaging that can be leveraged during the acquisition process.
Figure 2: Flow Diagram for Unicast-Based Rapid Acquisition
As defined in [RFC5760], the feedback target (FT) is the entity to which the RTP_Rx sends its RTCP feedback messages indicating packet loss by means of an RTCP NACK message or indicating RTP_Rx's desire to rapidly acquire the primary multicast RTP session by means of an RTCP feedback message defined in this document. While the Burst/ Retransmission Source (BRS) is responsible for responding to these messages and for further RTCP interaction with the RTP_Rx in the case of a rapid acquisition process, it is assumed in the remainder of this document that these two logical entities (FT and BRS) are combined in a single physical entity and they share state. In the remainder of the text, the term Retransmission Server (RS) is used whenever appropriate, to refer to this single physical entity.
Ver Steeg, et al. Standards Track [Page 14]
RFC 6285 RAMS June 2011
The FT is involved in the primary multicast RTP session and receives unicast feedback for that session as in [RFC5760]. The BRS is involved in the unicast burst (or retransmission) RTP session and transmits the unicast burst and retransmission packets formatted as RTP retransmission packets [RFC4588] in a single separate unicast RTP session to each RTP_Rx. In the unicast burst RTP session, as in any other RTP session, the BRS and RTP_Rx regularly send the periodic sender and receiver reports, respectively.
The unicast burst is started by an RTCP feedback message that is defined in this document based on the common packet format provided in [RFC4585]. An RTP retransmission is triggered by an RTCP NACK message defined in [RFC4585]. Both of these messages are sent to the FT of the primary multicast RTP session and can start the unicast burst/retransmission RTP session.
In the extended RTP profile for RTCP-based feedback (RTP/Audio-Visual Profile with Feedback (AVPF)), there are certain rules that apply to scheduling of both of these messages sent to the FT in the primary multicast RTP session; these are detailed in Section 3.5 of [RFC4585]. One of the rules states that in a multi-party session such as the SSM sessions we are considering in this specification, an RTP_Rx cannot send an RTCP feedback message for a minimum of one second after joining the session (i.e., Tmin=1.0 second). While this rule has the goal of avoiding problems associated with flash crowds in typical multi-party sessions, it defeats the purpose of rapid acquisition. Furthermore, when RTP receivers delay their messages requesting a burst by a deterministic or even a random amount, it still does not make a difference since such messages are not related and their timings are independent from each other. Thus, in this specification, we initialize Tmin to zero and allow the RTP receivers to send a burst request message right at the beginning. For the subsequent messages (e.g., updated requests) during rapid acquisition, the timing rules of [RFC4585] still apply.
Figure 3 depicts an example of messaging flow for rapid acquisition. The RTCP feedback messages are explained below. The optional messages are indicated in parentheses, and they might or might not be present during rapid acquisition. In a scenario where rapid acquisition is performed by a feedback target co-located with the media sender, the same method (with the identical message flows) still applies.
Figure 3: Message Flows for Unicast-Based Rapid Acquisition
This document defines the expected behaviors of the RS and RTP_Rx. It is instructive to consider independently operating implementations on the RS and RTP_Rx that request the burst, describe the burst, start the burst, join the multicast session, and stop the burst. These implementations send messages to each other, but provisions are needed for the cases where the control messages get lost, or reordered, or are not being delivered to their destinations.
The following steps describe rapid acquisition in detail:
1. Port Mapping Setup: For the primary multicast RTP session, the RTP and RTCP destination ports are declaratively specified (refer to Section 8 for examples in SDP). However, the RTP_Rx needs to choose its RTP and RTCP receive ports for the unicast burst RTP session. Since this unicast session is established after the RTP_Rx has sent a RAMS Request (RAMS-R) message as unicast feedback in the primary multicast RTP session, the RTP_Rx MUST first set up the port mappings between the unicast and multicast sessions and send this mapping information to the FT along with the RAMS-R message so that the BRS knows how to communicate with the RTP_Rx.
The details of this setup procedure are explained in [RFC6284]. Other NAT-related issues are left to Section 9 to keep the present discussion focused on the RAMS message flows.
2. Request: The RTP_Rx sends a rapid acquisition request (RAMS-R) for the primary multicast RTP session to the unicast feedback target of that session. The request contains the SSRC identifier of the RTP_Rx (aka SSRC of packet sender) and can contain the media sender SSRC identifier(s) of the primary multicast stream(s) of interest (aka SSRC of media source). The RAMS-R message can contain parameters that constrain the burst, such as the buffer and bandwidth limits.
Before joining the SSM session, the RTP_Rx learns the addresses for the multicast source, group, and RS by out-of-band means. If the RTP_Rx desires to rapidly acquire only a subset of the primary multicast streams available in the primary multicast RTP
Ver Steeg, et al. Standards Track [Page 17]
RFC 6285 RAMS June 2011
session, the RTP_Rx MUST also acquire the SSRC identifiers for the desired RTP streams out-of-band. Based on this information, the RTP_Rx populates the desired SSRC(s) in the RAMS-R message.
When the FT successfully receives the RAMS-R message, the BRS responds to it by accepting or rejecting the request. Immediately before the BRS sends any RTP or RTCP packet(s) described below, it establishes the unicast burst RTP session.
3. Response: The BRS sends RAMS Information (RAMS-I) message(s) to the RTP_Rx to convey the status for the burst(s) requested by the RTP_Rx.
If the primary multicast RTP session associated with the FT_Ap (a specific feedback target running on a particular address and port) on which the RAMS-R message was received contains only a single primary multicast stream, the BRS SHALL always use the SSRC of the RTP stream associated with the FT_Ap in the RAMS-I message(s) regardless of the media sender SSRC requested in the RAMS-R message. In such cases the 'ssrc' attribute MAY be omitted from the media description. If the requested SSRC and the actual media sender SSRC do not match, the BRS MUST explicitly populate the correct media sender SSRC in the initial RAMS-I message (see Section 7.3).
The FT_Ap could also be associated with an RTP session that carries two or more primary multicast streams. If the RTP_Rx wants to issue a collective request to receive the whole primary multicast RTP session, it does not need the 'ssrc' attributes to be described in the media description.
If the FT_Ap is associated with two or more RTP sessions, RTP_Rx's request will be ambiguous. To avoid any ambiguity, each FT_Ap MUST be only associated with a single RTP session.
If the RTP_Rx is willing to rapidly acquire only a subset of the primary multicast streams, the RTP_Rx MUST list all the media sender SSRC(s) denoting the stream(s) it wishes to acquire in the RAMS-R message. Upon receiving such a message, the BRS MAY accept the request for all or a subset of the media sender SSRC(s) that match the RTP stream(s) it serves. The BRS MUST reject all other requests with an appropriate response code.
* Reject Responses: The BRS MUST take into account any limitations that may have been specified by the RTP_Rx in the RAMS-R message when making a decision regarding the request. If the RTP_Rx has requested to acquire the whole primary multicast RTP session but the BRS cannot provide a rapid
Ver Steeg, et al. Standards Track [Page 18]
RFC 6285 RAMS June 2011
acquisition service for any of the primary multicast streams, the BRS MUST reject the request via a single RAMS-I message with a collective reject response code, which is defined as 510 in Section 11.6 and whose media sender SSRC field is set to one of SSRCs served by this FT_Ap. Upon receiving this RAMS-I message, the RTP_Rx abandons the rapid acquisition attempt and can immediately join the multicast session by sending an SFGMP Join message towards its upstream multicast router.
In all other cases, the BRS MUST send a separate RAMS-I message with the appropriate 5xx-level response code (as defined in Section 11.6) for each primary multicast stream that has been requested by the RTP_Rx but cannot be served by the BRS. There could be multiple reasons why the BRS has rejected the request; however, the BRS chooses the most appropriate response code to inform the RTP_Rx.
Upon receiving a reject response indicating a transient problem such as insufficient BRS or network resources, if the RTP_Rx wants to retry sending the same request, the RTP_Rx MUST follow the RTCP timer rules of [RFC4585] to allow the transient problems to go away. However, if the reject response indicates a non-transient problem (such as the ones reported by response codes 504, 505, and 506), the RTP_Rx MUST NOT attempt a retry. Different response codes have different scopes. Refer to Section 7.3.1 for details.
The BRS can employ a policing mechanism against the broken RTP_Rx implementations that are not following these rules. See Section 10 for details.
* Accept Responses: The BRS MUST send at least one separate RAMS-I message with the appropriate response code (either zero indicating a private response or response code 200 indicating success as listed in Section 11.6) for each primary multicast stream that has been requested by the RTP_Rx and will be served by the BRS. Such RAMS-I messages comprise fields that can be used to describe the individual unicast burst streams. When there is a RAMS-R request for multiple primary multicast streams, the BRS MUST include all the individual RAMS-I messages corresponding to that RAMS-R request in the same compound RTCP packet if these messages fit in the same packet. Note that if the BRS is sending only the preamble information but not the whole unicast burst stream, it will not accept the request but will send a response code 511 instead, as defined in Section 11.6.
Ver Steeg, et al. Standards Track [Page 19]
RFC 6285 RAMS June 2011
The RAMS-I message carries the RTP sequence number of the first packet transmitted in the respective RTP stream to allow the RTP_Rx to detect any missing initial packet(s). When the BRS accepts a request for a primary multicast stream, this field MUST always be populated in the RAMS-I message(s) sent for this particular primary multicast stream. It is RECOMMENDED that the BRS sends a RAMS-I message at the start of the burst so that the RTP_Rx can quickly detect any missing initial packet(s).
It is possible that the RAMS-I message for a primary multicast stream can get delayed or lost, and the RTP_Rx can start receiving RTP packets before receiving a RAMS-I message. An RTP_Rx implementation MUST NOT assume it will quickly receive the initial RAMS-I message. For redundancy purposes, it is RECOMMENDED that the BRS repeats the RAMS-I messages multiple times as long as it follows the RTCP timer rules defined in [RFC4585].
4. Unicast Burst: For the primary multicast stream(s) for which the request is accepted, the BRS starts sending the unicast burst(s) that comprises one or more RTP retransmission packets sent in the unicast burst RTP session. In some applications, the BRS can send preamble information data to the RTP_Rx in addition to the requested burst to prime the media decoder(s). However, for the BRS to send the preamble information in a particular format, explicit signaling from the RTP_Rx is required. In other words, the BRS MUST NOT send preamble information in a particular format unless the RTP_Rx has signaled support for that format in the RAMS-R message through a public or private extension as defined in Section 7.1.
The format of this preamble data is RTP-payload specific and not specified here.
5. Updated Request: The RTP_Rx MAY send an updated RAMS-R message (as unicast feedback in the primary multicast RTP session) with a different value for one or more fields of an earlier RAMS-R message. The BRS MUST be able to detect whether a burst is already planned for or being transmitted to this particular RTP_Rx for this particular media sender SSRC. If there is an existing burst planned for or being transmitted, the newly arriving RAMS-R is an updated request; otherwise, it is a new request. It is also possible that the RTP_Rx has sent an improperly formatted RAMS-R message or used an invalid value in the RAMS-R message. If notified by the BRS using a 4xx-level
Ver Steeg, et al. Standards Track [Page 20]
RFC 6285 RAMS June 2011
response code (as defined in Section 11.6) and only after following the timing rules of [RFC4585], the RTP_Rx MAY resend its corrected request.
The BRS determines the identity of the requesting RTP_Rx by looking at its canonical name identifier (CNAME item in the source description (SDES)). Thus, the RTP_Rx MUST choose a method that ensures it uses a unique CNAME identifier. Such methods are provided in [RFC6222]. In addition to one or more fields with updated values, an updated RAMS-R message may also include the fields whose values did not change.
Upon receiving an updated request, the BRS can use the updated values for sending/shaping the burst or refine the values and use the refined values for sending/shaping the burst. Subsequently, the BRS MAY send an updated RAMS-I message in the unicast burst RTP session to indicate the changes it made.
It is an implementation-dependent decision for an RTP_RX whether and when to send an updated request. However, note that the updated request messages can get delayed and arrive at the BRS after the initial unicast burst was completed. In this case, the updated request message can trigger a new unicast burst, and by then if the RTP_Rx has already started receiving multicast data, a congestion is likely to occur. In this case, the RTP_Rx can detect this only after a delay, and then it can try to terminate the new burst. However, in the meantime, the RTP_Rx can experience packet loss or other problems. This and other similar corner cases SHOULD be carefully examined if updated requests are to be used.
6. Updated Response: The BRS can send more than one RAMS-I message in the unicast burst RTP session, e.g., to update the value of one or more fields in an earlier RAMS-I message. The updated RAMS-I messages might or might not be a direct response to a RAMS-R message. The BRS can also send updated RAMS-I messages to signal the RTP_Rx in real time to join the SSM session when the BRS had already sent an initial RAMS-I message, e.g., at the start of the burst. The RTP_Rx depends on the BRS to learn the join time, which can be conveyed by the first RAMS-I message or can be sent/revised in a later RAMS-I message. If the BRS is not capable of determining the join time in the initial RAMS-I message, the BRS MUST send another RAMS-I message (with the join time information) later.
7. Multicast Join Signaling: The RAMS-I message allows the BRS to signal explicitly when the RTP_Rx needs to send the SFGMP Join message. The RTP_Rx SHOULD use this information from the most
Ver Steeg, et al. Standards Track [Page 21]
RFC 6285 RAMS June 2011
recent RAMS-I message unless it has more accurate information. If the request is accepted, this information MUST be conveyed in at least one RAMS-I message, and its value MAY be updated by subsequent RAMS-I messages.
There can be missing packets if the RTP_Rx joins the multicast session too early or too late. For example, if the RTP_Rx starts receiving the primary multicast stream while it is still receiving the unicast burst at a high excess bitrate, this can result in an increased risk of packet loss. Or, if the RTP_Rx joins the multicast session some time after the unicast burst is finished, there can be a gap between the burst and multicast data (a number of RTP packets might be missing). In both cases, the RTP_Rx can issue retransmission requests (via RTCP NACK messages sent as unicast feedback in the primary multicast RTP session) [RFC4585] to the FT entity of the RS to fill the gap. The BRS might or might not respond to such requests. When it responds and the response causes significant changes in one or more values reported earlier to the RTP_Rx, an updated RAMS-I SHOULD be sent to the RTP_Rx.
8. Multicast Receive: After the join, the RTP_Rx starts receiving the primary multicast stream(s).
9. Terminate: The BRS can know when it needs to ultimately stop the unicast burst based on its parameters. However, the RTP_Rx may need to ask the BRS to terminate the burst prematurely or at a specific sequence number. For this purpose, the RTP_Rx uses the RAMS Termination (RAMS-T) message sent as RTCP feedback in the unicast burst RTP session. A separate RAMS-T message is sent for each primary multicast stream served by the BRS unless an RTCP BYE message has been sent in the unicast burst RTP session as described in Step 10. For the burst requests that were rejected by the BRS, there is no need to send a RAMS-T message.
If the RTP_Rx wants to terminate a burst prematurely, it MUST send a RAMS-T message for the SSRC of the primary multicast stream it wishes to terminate. This message is sent in the unicast burst RTP session. Upon receiving this message, the BRS MUST terminate the unicast burst. If the RTP_Rx requested to acquire the entire primary multicast RTP session but wants to terminate this request before it learns the individual media sender SSRC(s) via RAMS-I message(s) or RTP packets, the RTP_Rx cannot use RAMS-T message(s) and thus MUST send an RTCP BYE message in the unicast burst RTP session to terminate the request.
Ver Steeg, et al. Standards Track [Page 22]
RFC 6285 RAMS June 2011
Otherwise, the default behavior for the RTP_Rx is to send a RAMS-T message in the unicast burst RTP session immediately after it joins the multicast session and has started receiving multicast packets. In that case, the RTP_Rx MUST send a RAMS-T message with the sequence number of the first RTP packet received in the primary multicast stream. Then, the BRS MUST terminate the respective burst after it sends the unicast burst packet whose Original Sequence Number (OSN) field in the RTP retransmission payload header matches this number minus one. If the BRS has already sent that unicast burst packet when the RAMS-T message arrives, the BRS MUST terminate the respective burst immediately.
If an RTCP BYE message has not been issued yet as described in Step 10, the RTP_Rx MUST send at least one RAMS-T message for each primary multicast stream served by the BRS. The RAMS-T message(s) MUST be sent to the BRS in the unicast burst RTP session. Against the possibility of a message loss, it is RECOMMENDED that the RTP_Rx repeats the RAMS-T messages multiple times as long as it follows the RTCP timer rules defined in [RFC4585].
The binding between RAMS-T and ongoing bursts is achieved through RTP_Rx's CNAME identifier and packet sender and media sender SSRCs. Choosing a globally unique CNAME makes sure that the RAMS-T messages are processed correctly.
10. Terminate with RTCP BYE: When the RTP_Rx is receiving one or more burst streams, if the RTP_Rx becomes no longer interested in acquiring any of the primary multicast streams, the RTP_Rx SHALL issue an RTCP BYE message for the unicast burst RTP session and another RTCP BYE message for the primary multicast RTP session. These RTCP BYE messages are sent to the BRS and FT logical entities, respectively.
Upon receiving an RTCP BYE message, the BRS MUST terminate the rapid acquisition operation and cease transmitting any further burst packets and retransmission packets. If support for [RFC5506] has been signaled, the RTCP BYE message MAY be sent in a reduced-size RTCP packet. Otherwise, Section 6.1 of [RFC3550] mandates the RTCP BYE message always be sent with a sender or receiver report in a compound RTCP packet. If no data has been received, an empty receiver report MUST be still included. With the information contained in the receiver report, the RS can figure out how many duplicate RTP packets have been delivered to the RTP_Rx (note that this will be an upper-bound estimate as one or more packets might have been lost during the burst
Ver Steeg, et al. Standards Track [Page 23]
RFC 6285 RAMS June 2011
transmission). The impact of duplicate packets and measures that can be taken to minimize the impact of receiving duplicate packets will be addressed in Section 6.4.
Since an RTCP BYE message issued for the unicast burst RTP session terminates that session and ceases transmitting any further packets in that session, there is no need for sending explicit RAMS-T messages, which would only terminate their respective bursts.
For the purpose of gathering detailed information about RTP_Rx's rapid acquisition experience, [MULTICAST-ACQ] defines an RTCP Extended Report (XR) Block. This report is designed to be payload- independent; thus, it can be used by any multicast application that supports rapid acquisition.
When an RTP_RX acquires multiple primary multicast streams, it might need to synchronize them for playout. This synchronization is achieved by the help of the RTCP sender reports [RFC3550]. If the playout will start before the RTP_Rx has joined the multicast session, the RTP_Rx needs to receive the information reflecting the synchronization among the primary multicast streams early enough so that it can play out the media in a synchronized fashion.
The suggested approach is to use the RTP header extension mechanism [RFC5285] and convey the synchronization information in a header extension as defined in [RFC6051]. Per [RFC4588], "if the original RTP header carried an RTP header extension, the retransmission packet SHOULD carry the same header extension". Thus, as long as the multicast source emits a header extension with the synchronization information frequently enough, there is no additional task that needs to be carried out by the BRS. The synchronization information will be sent to the RTP_Rx along with the burst packets. The frequent header extensions sent in the primary multicast RTP sessions also allow rapid synchronization of the RTP streams for the RTP receivers that do not support RAMS or that directly join the multicast session without running RAMS. Thus, in RAMS applications, it is RECOMMENDED that the multicast sources frequently send synchronization information by using header extensions following the rules presented in [RFC6051]. The regular sender reports are still sent in the unicast session by following the rules of [RFC3550].
This section provides informative guidelines about how the BRS can shape the transmission of the unicast burst and how congestion can be dealt with in the RAMS process. When the RTP_Rx detects that the unicast burst is causing severe congestion, it can prefer to send a RAMS-T message immediately to stop the unicast burst.
A higher bitrate for the unicast burst naturally conveys the Reference Information and media content to the RTP_Rx faster. This way, the RTP_Rx can start consuming the data sooner, which results in a faster acquisition. A higher bitrate also represents a better utilization of the BRS resources. As the burst may continue until it catches up with the primary multicast stream, the higher the bursting bitrate, the less data the BRS needs to transmit. However, a higher bitrate for the burst also increases the chances for congestion- caused packet loss. Thus, as discussed in Section 5, there has to be an upper bound on the bandwidth used by the burst.
When the BRS transmits the unicast burst, it is expected to take into account all available information to prevent any packet loss that might take place during the bursting as a result of buffer overflow on the path between the RS and RTP_Rx and at the RTP_Rx itself. The bursting bitrate can be determined by taking into account the following information, when available:
(a) Information obtained via the RAMS-R message, such as Max RAMS Buffer Fill Requirement and/or Max Receive Bitrate (see Section 7.2).
(b) Information obtained via RTCP receiver reports provided by the RTP_Rx in the retransmission session, allowing in-session bitrate adaptations for the burst. When these receiver reports indicate packet loss, this can indicate a certain congestion state in the path from the RS to the RTP_Rx.
(c) Information obtained via RTCP NACKs provided by the RTP_Rx in the primary multicast RTP session, allowing in-session bitrate adaptations for the burst. Such RTCP NACKs are transmitted by the RTP_Rx in response to packet loss detection in the burst. NACKs can indicate a certain congestion state on the path from the RS to RTP_Rx.
(d) There can be other feedback received from the RTP_Rx, e.g., in the form of Explicit Congestion Notification - Congestion Experienced (ECN-CE) markings [ECN-FOR-RTP] that can influence in-session bitrate adaptation.
Ver Steeg, et al. Standards Track [Page 25]
RFC 6285 RAMS June 2011
(e) Information obtained via updated RAMS-R messages, allowing in- session bitrate adaptations, if supported by the BRS.
(f) Transport protocol-specific information. For example, when Datagram Congestion Control Protocol (DCCP) is used to transport the RTP burst, the ACKs from the DCCP client can be leveraged by the BRS / DCCP server for burst shaping and congestion control.
(g) Preconfigured settings for each RTP_Rx or a set of RTP_Rxs that indicate the upper-bound bursting bitrates for which no packet loss will occur as a result of congestion along the path of the RS to RTP_Rx. For example, in managed IPTV networks, where the bottleneck bandwidth along the end-to-end path is known and where the network between the RS and this link is provisioned and dimensioned to carry the burst streams, the bursting bitrate does not exceed the provisioned value. These settings can also be dynamically adapted using application-aware knowledge.
The BRS chooses the initial burst bitrate as follows:
o When using RAMS in environments as described in (g), the BRS MUST transmit the burst packets at an initial bitrate higher than the nominal bitrate but within the engineered or reserved bandwidth limit.
o When the BRS cannot determine a reliable bitrate value for the unicast burst (using (a) through (g)), it is desirable for the BRS to choose an appropriate initial bitrate not above the nominal bitrate and increase it gradually unless a congestion is detected.
In both cases, during the burst transmission, the BRS MUST continuously monitor for packet losses as a result of congestion by means of one or more of the mechanisms described in (b) through (f). When the BRS relies on RTCP receiver reports, sufficient bandwidth needs to be provided to RTP_Rx for RTCP transmission in the unicast burst RTP session. To achieve a reasonable fast adaptation against congestion, it is recommended that the RTP_Rx sends a receiver report at least once every two RTTs between the RS and RTP_Rx. Although the specific heuristics and algorithms that deduce a congestion state and how the BRS acts subsequently are outside the scope of this specification, the following two methods are the best practices:
o Upon detection of a significant amount of packet loss, which the BRS attributes to congestion, the BRS decreases the burst bitrate. The rate by which the BRS increases and decreases the bitrate for the burst can be determined by a TCP-friendly bitrate adaptation algorithm for RTP over UDP or, in the case of (f), by the congestion control algorithms defined in DCCP [RFC5762].
Ver Steeg, et al. Standards Track [Page 26]
RFC 6285 RAMS June 2011
o If the congestion is persistent and the BRS has to reduce the burst bitrate to a point where the RTP_Rx buffer might underrun or the burst will consume too many resources, the BRS terminates the burst and transmits a RAMS-I message to RTP_Rx with the appropriate response code. It is then up to RTP_Rx to decide when to join the multicast session.
Even though there is no congestion experienced during the burst, congestion may anyway arise near the end of the burst as the RTP_Rx eventually needs to join the multicast session. During this brief period, both the burst packets and the multicast packets can be simultaneously received by the RTP_Rx, thus enhancing the risk of congestion.
Since the BRS signals the RTP_Rx when the BRS expects the RTP_Rx to send the SFGMP Join message, the BRS can have a rough estimate of when the RTP_Rx will start receiving multicast packets in the SSM session. The BRS can keep on sending burst packets but reduces the bitrate accordingly at the appropriate instant by taking the bitrate of the whole SSM session into account. If the BRS ceases transmitting the burst packets before the burst catches up, any gap resulting from this imperfect switch-over by the RTP_Rx can be later repaired by requesting retransmissions for the missing packets from the RS. The retransmissions can be shaped by the BRS to make sure that they do not cause collateral loss in the primary multicast RTP session and the unicast burst RTP session.
In this section, we examine the implications of losing the RAMS-R, RAMS-I, or RAMS-T messages and other failure cases.
When the RTP_Rx sends a RAMS-R message to initiate a rapid acquisition but the message gets lost and the FT does not receive it, the RTP_Rx will get neither a RAMS-I message nor a unicast burst. In this case, the RTP_Rx MAY resend the request when it is eligible to do so based on the RTCP timer rules defined in [RFC4585]. Or, after a reasonable amount of time, the RTP_Rx can time out (based on the previous observed response times) and immediately join the SSM session.
In the case where RTP_Rx starts receiving a unicast burst but does not receive a corresponding RAMS-I message within a reasonable amount of time, the RTP_Rx can either discard the burst data or decide not to interrupt the unicast burst and be prepared to join the SSM session at an appropriate time it determines or as indicated in a subsequent RAMS-I message (if available). If the network is subject
Ver Steeg, et al. Standards Track [Page 27]
RFC 6285 RAMS June 2011
to packet loss, it is RECOMMENDED that the BRS repeats the RAMS-I messages multiple times based on the RTCP timer rules defined in [RFC4585].
In the failure cases where the RAMS-I message is lost or the RAMS-R message is lost and the RTP_Rx gives up, the RTP_Rx MUST still terminate the burst(s) it requested by following the rules described in Section 6.2.
In the case where a RAMS-T message sent by the RTP_Rx does not reach its destination, the BRS can continue sending burst packets even though the RTP_Rx no longer needs them. If an RTP_Rx is receiving burst packets it no longer needs after sending a RAMS-T message, it is RECOMMENDED that the RTP_Rx repeats the RAMS-T message multiple times based on the RTCP timer rules defined in [RFC4585]. The BRS MUST be provisioned to terminate the burst when it can no longer send the burst packets faster than it receives the primary multicast stream packets.
Section 6.3.5 of [RFC3550] explains the rules pertaining to timing out an SSRC. When the BRS accepts to serve the requested burst(s) and establishes the retransmission session, the BRS needs to check the liveness of the RTP_Rx via the RTCP messages and reports the RTP_Rx sends. The default rules explained in [RFC3550] apply in RAMS as well.
This section defines the formats of the RTCP transport-layer feedback messages that are exchanged between the retransmission server (RS) and RTP receiver (RTP_Rx) during rapid acquisition. These messages are referred to as the RAMS Messages. They are payload-independent and MUST be used by all RTP-based multicast applications that support rapid acquisition regardless of the payload they carry.
Payload-specific feedback messages are not defined in this document. However, further optional payload-independent and payload-specific information can be included in the exchange.
The common packet format for the RTCP feedback messages is defined in Section 6.1 of [RFC4585] and is also sketched in Figure 4.
Figure 4: The Common Packet Format for the RTCP Feedback Messages
Each feedback message has a fixed-length field for version, padding, feedback message type (FMT), packet type (PT), length, SSRC of packet sender, SSRC of media source, and a variable-length field for feedback control information (FCI).
In the RAMS messages, the PT field is set to RTPFB (205) and the FMT field is set to RAMS (6). Individual RAMS messages are identified by a sub-field called sub-feedback message type (SFMT). Any Reserved field SHALL be set to zero and ignored.
Depending on the specific scenario and timeliness/importance of a RAMS message, it can be desirable to send it in a reduced-size RTCP packet [RFC5506]. However, unless support for [RFC5506] has been signaled, compound RTCP packets MUST be used by following [RFC3550] rules.
Following the rules specified in [RFC3550], all integer fields in the messages defined below are carried in network-byte order, that is, most significant byte (octet) first, also known as big-endian. Unless otherwise stated, numeric constants are in decimal (base 10).
To improve the functionality of the RAMS method in certain applications, it can be desirable to define new fields in the RAMS Request, Information, and Termination messages. Such fields MUST be encoded as Type-Length-Value (TLV) elements as described below and sketched in Figure 5:
o Type: A single-octet identifier that defines the type of the parameter represented in this TLV element.
Ver Steeg, et al. Standards Track [Page 29]
RFC 6285 RAMS June 2011
o Length: A two-octet field that indicates the length (in octets) of the TLV element excluding the Type and Length fields and the 8-bit Reserved field between them. This length does not include any padding that is required for alignment.
o Value: Variable-size set of octets that contains the specific value for the parameter.
In the extensions, the Reserved field SHALL be set to zero and ignored. If a TLV element does not fall on a 32-bit boundary, the last word MUST be padded to the boundary using further bits set to zero.
An RTP_Rx or BRS MAY include vendor-neutral and private extensions in any RAMS message. The RTP_Rx or BRS MUST place such extensions after the mandatory fields and mandatory TLV elements (if any) and MAY place such extensions in any order. The RTP_Rx or BRS MUST NOT include multiple TLV elements with the same Type value in a RAMS message.
The support for extensions (unless specified otherwise) is OPTIONAL. An RTP_Rx or BRS receiving a vendor-neutral or private extension that it does not understand MUST ignore that extension.
It is desirable to allow vendors to use private extensions in a TLV format. For interoperability, such extensions must not collide with each other.
Ver Steeg, et al. Standards Track [Page 30]
RFC 6285 RAMS June 2011
A certain range of TLV Types (between and including 128 and 254 ) is reserved for private extensions (refer to Section 11.5). IANA management for these extensions is unnecessary, and they are the responsibility of individual vendors.
The structure that MUST be used for the private extensions is depicted in Figure 6. Here, the enterprise numbers are as listed on http://www.iana.org. This will ensure the uniqueness of the private extensions and avoid any collision.
The RAMS Request (RAMS-R) message is identified by SFMT=1. This message is sent as unicast feedback in the primary multicast RTP session by the RTP_Rx to request rapid acquisition of a primary multicast RTP session or one or more primary multicast streams belonging to the same primary multicast RTP session. In the RAMS-R message, the RTP_Rx MUST set both the packet sender SSRC and media sender SSRC fields to its own SSRC since the media sender SSRC may not be known. The RTP_Rx MUST provide explicit signaling as described below to request stream(s). This minimizes the chances of accidentally requesting a wrong primary multicast stream.
The FCI field MUST contain only one RAMS Request. The FCI field has the structure depicted in Figure 7.
The semantics of the RAMS-R message is independent of the payload type carried in the primary multicast RTP session.
Figure 7: FCI Field Syntax for the RAMS Request Message
o Requested Media Sender SSRC(s): Mandatory TLV element that lists the media sender SSRC(s) requested by the RTP_Rx. The BRS MUST ignore the media sender SSRC specified in the header of the RAMS-R message.
If the Length field is set to zero (i.e., no media sender SSRC is listed), it means that the RTP_Rx is requesting to rapidly acquire the entire primary multicast RTP session. Otherwise, the RTP_Rx lists the individual media sender SSRCs in this TLV element and sets the Length field of the TLV element to 4*n, where n is the number of SSRC entries.
o Min RAMS Buffer Fill Requirement (32 bits): Optional TLV element that denotes the minimum milliseconds of data that the RTP_Rx desires to have in its buffer before allowing the data to be consumed by the application.
The RTP_Rx can have knowledge of its buffering requirements. These requirements can be application and/or device specific. For instance, the RTP_Rx might need to have a certain amount of data in its application buffer to handle transmission jitter and/or to be able to support error-control methods. If the BRS is told the minimum buffering requirement of the receiver, the BRS can tailor the burst(s) more precisely, e.g., by choosing an appropriate starting point. The methods used by the RTP_Rx to determine this value are application specific and thus out of the scope of this document.
Ver Steeg, et al. Standards Track [Page 32]
RFC 6285 RAMS June 2011
If specified, the amount of backfill that will be provided by the unicast bursts and any payload-specific information MUST NOT be smaller than the specified value. Otherwise, the backfill will not be able to build up the desired level of buffer at the RTP_Rx and can cause buffer underruns.
o Max RAMS Buffer Fill Requirement (32 bits): Optional TLV element that denotes the maximum milliseconds of data that the RTP_Rx can buffer without losing the data due to buffer overflow.
The RTP_Rx can have knowledge of its buffering requirements. These requirements can be application or device specific. For instance, one particular RTP_Rx might have more physical memory than another RTP_Rx and thus can buffer more data. If the BRS knows the buffering ability of the receiver, the BRS can tailor the burst(s) more precisely. The methods used by the receiver to determine this value are application specific and thus out of the scope of this document.
If specified, the amount of backfill that will be provided by the unicast bursts and any payload-specific information MUST NOT be larger than this value. Otherwise, the backfill may cause buffer overflows at the RTP_Rx.
o Max Receive Bitrate (64 bits): Optional TLV element that denotes the maximum bitrate (in bits per second) at which the RTP_Rx can process the aggregation of the unicast burst(s) and any payload- specific information that will be provided by the BRS. The limits can include local receiver limits as well as network limits that are known to the receiver.
If specified, the total bitrate of the unicast burst(s) plus any payload-specific information MUST NOT be larger than this value. Otherwise, congestion and packet loss are more likely to occur.
o Preamble-only Allowed (0 bits): Optional TLV element that indicates that the RTP_Rx is willing to receive only the preamble information for the desired primary multicast stream(s) in case the BRS cannot send the unicast burst stream(s). (In such cases, the BRS will not accept the request, but will send a response code 511 in the RAMS-I message as defined in Section 11.6.) Note that
Ver Steeg, et al. Standards Track [Page 33]
RFC 6285 RAMS June 2011
the RTP_Rx signals the particular preamble format(s) it supports through a public or a private extension in the same RAMS-R message.
If this TLV element does not exist in the RAMS-R message, the BRS MUST NOT respond to the RAMS-R message by sending the preamble information only. Since this TLV element does not carry a Value field, the Length field MUST be set to zero.
o Supported Enterprise Number(s): Optional TLV element that indicates the enterprise number(s) that the RTP_Rx implementation supports. Similar to the private extensions, the enterprise numbers here are as listed on http://www.iana.org. This TLV element, if exists, provides a hint to the BRS about which private extensions the RTP_Rx can potentially support. Note that an RTP_Rx does not necessarily support all the private extensions under a particular enterprise number. Unless the BRS explicitly knows which private extensions an RTP_Rx supports (through this or some out-of-band means), the BRS SHOULD NOT use private extensions in the RAMS-I messages. The BRS SHOULD rather use only vendor- neutral extensions. The Length field of this TLV element is set to 4*n, where n is the number of enterprise number entries.
The RAMS Information (RAMS-I) message is identified by SFMT=2. This message is used to describe the unicast burst that will be sent for rapid acquisition. It also includes other useful information for the RTP_Rx as described below.
A separate RAMS-I message with the appropriate response code is sent in the unicast burst RTP session by the BRS for each primary multicast stream that has been requested by the RTP_Rx. In each of these RAMS-I messages, the media sender SSRC and packet sender SSRC fields are both set to the SSRC of the BRS, which equals the SSRC of the respective primary multicast stream.
The FCI field MUST contain only one RAMS Information message. The FCI field has the structure depicted in Figure 8.
The semantics of the RAMS-I message is independent of the payload type carried in the primary multicast RTP session.
Figure 8: FCI Field Syntax for the RAMS Information Message
A RAMS-I message has the following fields:
o Message Sequence Number (MSN) (8 bits) : Mandatory field that denotes the sequence number of the RAMS-I message for the particular media sender SSRC specified in the message header. The MSN value SHALL be set to zero when a new RAMS request is received. During rapid acquisition, the same RAMS-I message MAY be repeated for redundancy purposes without incrementing the MSN value. If an updated RAMS-I message will be sent (either with new information or updated information), the MSN value SHALL be incremented by one. In the MSN field, the regular wrapping rules apply. Note that if the RTP_Rx has sent an updated request, it MUST check every RAMS-I message entirely, regardless of whether or not the MSN value has changed.
o Response (16 bits): Mandatory field that denotes the response code for this RAMS-I message. This document defines several initial response codes in Section 7.3.1 and registers them with IANA in Section 11.6. If a new vendor-neutral response code will be defined, it MUST be registered with IANA through the guidelines specified in Section 11.6. If the new response code is intended to be used privately by a vendor, there is no need for IANA management. Instead, the vendor MUST use the private extension mechanism (Section 7.1.2) to convey its message and MUST indicate this by putting zero in the Response field.
When the RTP_Rx receives a RAMS-I message with a response code that it does not understand, the RTP_Rx MUST send a RAMS-T message immediately to the BRS.
The following TLV elements have been defined for the RAMS-I messages:
o Media Sender SSRC (32 bits): Optional TLV element that specifies the media sender SSRC of the unicast burst stream. If the FT_Ap that received the RAMS-R message is associated with a single primary multicast stream but the requested media sender SSRC does not match the SSRC of the RTP stream associated with this FT_Ap,
Ver Steeg, et al. Standards Track [Page 35]
RFC 6285 RAMS June 2011
the BRS includes this TLV element in the initial RAMS-I message to let the RTP_Rx know that the media sender SSRC has changed. If the two SSRCs match, there is no need to include this TLV element.
o RTP Seqnum of the First Packet (16 bits): TLV element that specifies the RTP sequence number of the first packet that will be sent in the respective unicast RTP stream. This allows the RTP_Rx to know whether one or more packets sent by the BRS have been dropped at the beginning of the stream. If the BRS accepts the RAMS request, this element exists. If the BRS rejects the RAMS request, this element SHALL NOT exist.
o Earliest Multicast Join Time (32 bits): TLV element that specifies the delta time (in ms) between the arrival of the first RTP packet in the unicast RTP stream (which could be a burst packet or a payload-specific packet) and the earliest time instant when an RTP_Rx MAY send an SFGMP Join message to join the multicast session. A zero value indicated in this element means that the RTP_Rx MAY send the SFGMP Join message right away. If the RTP_Rx does not want to wait until the earliest multicast join time, it MAY send a RAMS-T message, and after a reasonable amount of time, it MAY join the SSM session.
If the RAMS request has been accepted, the BRS sends this element at least once so that the RTP_Rx knows when to join the multicast session. If the burst request has been rejected as indicated in the Response field, this element MUST indicate a zero value. In that case, it is up to the RTP_Rx when or whether to join the multicast session.
When the BRS serves two or more bursts and sends a separate RAMS-I message for each burst, the join times specified in these RAMS-I messages SHOULD correspond to more or less the same time instant, and the RTP_Rx sends the SFGMP Join message based on the earliest join time.
o Burst Duration (32 bits): Optional TLV element that denotes the time the burst will last (in ms), i.e., the difference between the expected transmission times of the first and the last burst packets that the BRS is planning to send in the respective unicast RTP stream. In the absence of additional stimulus, the BRS will
Ver Steeg, et al. Standards Track [Page 36]
RFC 6285 RAMS June 2011
send a burst of this duration. However, the burst duration can be modified by subsequent events, including changes in the primary multicast stream and reception of RAMS-T messages.
The BRS MUST terminate the flow in the timeframe indicated by this burst duration or the duration established by those subsequent events even if it does not get a RAMS-T or a BYE message from the RTP_Rx. It is OPTIONAL to send this element in a RAMS-I message when the burst request is accepted. If the burst request has been rejected as indicated in the Response field, this element MAY be omitted or indicate a zero value.
o Max Transmit Bitrate (64 bits): Optional TLV element that denotes the maximum bitrate (in bits per second) that will be used by the BRS for the RTP stream associated with this RAMS-I message.
1xx-Level Response Codes: These response codes are sent for informational purposes.
o 100: This is used when the BRS wants to update a value that was sent earlier to the RTP_Rx.
2xx-Level Response Codes: These response codes are sent to indicate success.
o 200: This is used when the server accepts the RAMS request.
o 201: This is used when the unicast burst has been completed and the BRS wants to notify the RTP_Rx.
4xx-Level Response Codes: These response codes are sent to indicate that the message sent by the RTP_Rx is either improperly formatted or contains an invalid value. The rules the RTP_Rx needs to follow upon receiving one of these response codes are outlined in Step 5 in Section 6.2. The 4xx-level response codes are also used as status codes in the Multicast Acquisition Report Block [MULTICAST-ACQ] in order to collect RTP_Rx-based error conditions.
o 400: This is used when the RAMS-R message is improperly formatted.
Ver Steeg, et al. Standards Track [Page 37]
RFC 6285 RAMS June 2011
o 401: This is used when the minimum RAMS buffer fill requirement value indicated in the RAMS-R message is invalid.
o 402: This is used when the maximum RAMS buffer fill requirement value indicated in the RAMS-R message is invalid.
o 403: This is used when the maximum receive bitrate value indicated in the RAMS-R message is insufficient.
o 404: This is used when the RAMS-T message is improperly formatted.
5xx-Level Response Codes: These response codes are sent to indicate an error on the BRS side. The rules the RTP_Rx needs to follow upon receiving one of these response codes are outlined in Step 3 in Section 6.2. The 5xx-level response codes are also used as status codes in the Multicast Acquisition Report Block [MULTICAST-ACQ] in order to collect BRS-based error conditions.
o 500: This is used when the BRS has experienced an internal error and cannot accept the RAMS request.
o 501: This is used when the BRS does not have enough bandwidth to send the unicast burst stream.
o 502: This is used when the BRS terminates the unicast burst stream due to network congestion.
o 503: This is used when the BRS does not have enough CPU resources to send the unicast burst stream.
o 504: This is used when the BRS does not support sending a unicast burst stream.
o 505: This is used when the requesting RTP_Rx is not eligible to receive a unicast burst stream.
o 506: This is used when RAMS functionality is not enabled for the requested multicast stream.
o 507: This is used when the BRS cannot find a valid starting point for the unicast burst stream that satisfies the RTP_Rx's requirements.
o 508: This is used when the BRS cannot find the essential reference information for the requested multicast stream.
Ver Steeg, et al. Standards Track [Page 38]
RFC 6285 RAMS June 2011
o 509: This is used when the BRS cannot match the requested SSRC to any of the streams it is serving.
o 510: This is used when the BRS cannot serve the requested entire session.
o 511: This is used when the BRS sends only the preamble information but not the whole unicast burst stream.
o 512: This is used when the RAMS request is denied due to a policy for the requested multicast stream, the RTP_Rx, or this particular BRS.
The RAMS Termination (RAMS-T) message is identified by SFMT=3.
The RAMS Termination is used to assist the BRS in determining when to stop the burst. A separate RAMS-T message is sent in the unicast burst RTP session by the RTP_Rx for each primary multicast stream that has been served by the BRS. Each of these RAMS-T message's media sender SSRC field SHALL BE populated with the SSRC of the media stream to be terminated. If the media sender SSRC populated in the RAMS-T message does not match the SSRC of the burst served by the BRS, BRS SHALL ignore the RAMS-T message.
If the RTP_Rx wants the BRS to stop a burst prematurely, it sends a RAMS-T message as described below. Upon receiving this message, the BRS stops the respective burst immediately. If the RTP_Rx wants the BRS to terminate all of the bursts, it needs to send all of the respective RAMS-T messages in a single compound RTCP packet.
The default behavior for the RTP_Rx is to send a RAMS-T message immediately after it joined the multicast session and started receiving multicast packets. In that case, the RTP_Rx includes the sequence number of the first RTP packet received in the primary multicast stream in the RAMS-T message. With this information, the BRS can decide when to terminate the unicast burst.
The FCI field MUST contain only one RAMS Termination. The FCI field has the structure depicted in Figure 9.
The semantics of the RAMS-T message is independent of the payload type carried in the primary multicast RTP session.
Figure 9: FCI field syntax for the RAMS Termination message
o Extended RTP Seqnum of First Multicast Packet (32 bits): Optional TLV element that specifies the extended RTP sequence number of the first packet received from the SSM session for a particular primary multicast stream. The low 16 bits contain the sequence number of the first packet received from the SSM session, and the most significant 16 bits extend that sequence number with the corresponding count of sequence number cycles, which can be maintained according to the algorithm in Appendix A.1 of [RFC3550].
The syntax of the 'rtcp-fb' attribute has been defined in [RFC4585]. Here we add the following syntax to the 'rtcp-fb' attribute (the feedback type and optional parameters are all case sensitive):
(In the following ABNF [RFC5234], rtcp-fb-nack-param is used as defined in [RFC4585].)
rtcp-fb-nack-param =/ SP "rai"
The following parameter is defined in this document for use with 'nack':
o 'rai' stands for Rapid Acquisition Indication and indicates the use of RAMS messages as defined in Section 7.
This document also defines a new media-level SDP attribute ('rams-updates') that indicates whether or not the BRS supports updated request messages. This attribute is used in a declarative manner and no Offer/Answer Model behavior is specified. If the BRS supports updated request messages and this attribute is included in the SDP description, the RTP_Rx can send updated requests. The BRS might or might not be able to accept value changes in every field in
Ver Steeg, et al. Standards Track [Page 40]
RFC 6285 RAMS June 2011
an updated RAMS-R message. However, if the 'rams-updates' attribute is not included in the SDP description, the RTP_Rx SHALL NOT send updated requests. The RTP_Rx MAY still repeat its initial request without changes, though.
Figure 10: Example SDP for a Single-Channel RAMS Scenario
In this example SDP description, we have a primary multicast (source) stream and a unicast retransmission stream. The source stream is multicast from a distribution source (with a source IP address of 198.51.100.1) to the multicast destination address of 188.8.131.52 and port 41000. The corresponding RTCP traffic is multicast to the same multicast destination address at port 42000. Here, we are assuming that the multicast RTP and RTCP ports are carefully chosen so that different RTP and RTCP streams do not collide with each other.
The feedback target (FT_Ap) residing on the RS (with an address of 192.0.2.1) at port 43000 is declared with the "a=rtcp" line [RFC3605]. The support for the conventional retransmission is indicated through the "a=rtcp-fb:98 nack" line. The RTP receiver(s) can report missing packets on the source stream to the feedback target and request retransmissions. The SDP above includes the "a=sendonly" line for the media description of the retransmission stream since the retransmissions are sent in only one direction.
The support for rapid acquisition is indicated through the "a=rtcp- fb:98 nack rai" line. The parameter 'rtx-time' applies to both the conventional retransmissions and rapid acquisition. However, when rapid acquisition is enabled, the standard definition for the parameter 'rtx-time' given in [RFC4588] is not followed. Instead, when rapid acquisition support is enabled, 'rtx-time' specifies the time in milliseconds that the BRS keeps an RTP packet in its cache
Ver Steeg, et al. Standards Track [Page 42]
RFC 6285 RAMS June 2011
available for retransmission (measured from the time the packet was received by the BRS, not from the time indicated in the packet timestamp).
Once an RTP_Rx has acquired an SDP description, it can ask for rapid acquisition before it joins a primary multicast RTP session. To do so, it sends a RAMS-R message to the feedback target of that primary multicast RTP session. If the FT_Ap is associated with only one RTP stream, the RTP_Rx does not need to learn the SSRC of that stream via an out-of-band method. If the BRS accepts the rapid acquisition request, it will send a RAMS-I message with the correct SSRC identifier. If the FT_Ap is associated with a multi-stream RTP session and the RTP_Rx is willing to request rapid acquisition for the entire session, the RTP_Rx again does not need to learn the SSRCs via an out-of-band method. However, if the RTP_Rx intends to request a particular subset of the primary multicast streams, it must learn their SSRC identifiers and list them in the RAMS-R message. Since this RTP_Rx has not yet received any RTP packets for the primary multicast stream(s), the RTP_Rx must in this case learn the SSRC value(s) from the 'ssrc' attribute of the media description [RFC5576]. In addition to the SSRC value, the 'cname' source attribute must also be present in the SDP description [RFC5576].
Listing the SSRC values for the primary multicast streams in the SDP file does not create a problem in SSM sessions when an SSRC collision occurs. This is because in SSM sessions, an RTP_Rx that observed an SSRC collision with a media sender must change its own SSRC [RFC5760] by following the rules defined in [RFC3550].
A feedback target that receives a RAMS-R message becomes aware that the RTP_Rx wants to rapidly catch up with a primary multicast RTP session. If the necessary conditions are satisfied (as outlined in Section 7 of [RFC4585]) and available resources exist, the BRS can react to the RAMS-R message by sending any transport-layer (and optional payload-specific, when allowed) feedback message(s) and starting the unicast burst.
In this section, we considered the simplest scenario where the primary multicast RTP session carried only one stream and the RTP_Rx wanted to rapidly acquire this stream only. Best practices for scenarios where the primary multicast RTP session carries two or more streams or the RTP_Rx wants to acquire one or more streams from multiple primary multicast RTP sessions at the same time are presented in [RAMS-SCENARIOS].
For a variety of reasons, one or more Network Address Port Translators (NAPT, hereafter simply called NAT) can exist between the RTP_Rx and RS. NATs have a variety of operating characteristics for UDP traffic [RFC4787]. For a NAT to permit traffic from the BRS to arrive at the RTP_Rx, the NAT(s) must first either:
a. See UDP (or DCCP) traffic sent from the RTP_Rx (which is on the 'inside' of the NAT) to the BRS (which is on the 'outside' of the NAT). This traffic has the same transport address as the subsequent response traffic, or
b. Be configured to forward certain ports (e.g., using HTML configuration or Universal Plug and Play (UPnP) Internet Gateway Device (IGD) [UPnP-IGD]). Details of this are out of the scope of this document.
For both (a) and (b), the RTP_Rx is responsible for maintaining the NAT's state if it wants to receive traffic from the BRS on that port. For (a), the RTP_Rx MUST send UDP traffic to keep the NAT binding alive, at least every 30 seconds [RFC4787]. While (a) is more like an automatic/dynamic configuration, (b) is more like a manual/static configuration.
When the RTP_Rx sends a request (RAMS-R) message to the FT as unicast feedback in the primary multicast RTP session and the request is received by the FT, a new unicast burst RTP session will be established between the BRS and RTP_Rx.
While the FT and BRS ports on the RS are already signaled via out-of- band means (e.g., SDP), the RTP_Rx needs to convey the RTP and RTCP ports it wants to use on its side for the new session. Since there are two RTP sessions (one multicast and one unicast) involved during this process and one of them is established upon a feedback message sent in the other one, this requires an explicit port mapping method.
Applications using RAMS MUST support the method presented in [RFC6284] both on the RS and RTP_Rx side to allow RTP receivers to use their desired ports and to support RAMS behind NAT devices. The port mapping message exchange needs to take place whenever the RTP_Rx intends to make use of the RAMS protocol for rapidly acquiring a specific multicast RTP session prior to joining the associated SSM session.
Applications that are using RAMS make heavy use of the unicast feedback mechanism described in [RFC5760], the payload format defined in [RFC4588], and the port mapping solution specified in [RFC6284]. Thus, these applications are subject to the general security considerations discussed in those documents. In particular, the threats and risks discussed in [RFC5760] need to be considered carefully. In this section, we give an overview of the guidelines and suggestions described in these specifications from a RAMS perspective. We also discuss the security considerations that explicitly apply to applications using RAMS.
First of all, much of the session description information is available in the SDP descriptions that are distributed to the media senders, retransmission servers, and RTP receivers. Adequate security measures are RECOMMENDED to ensure the integrity and authenticity of the SDP descriptions so that transport addresses of the media senders, distribution sources, feedback targets, and other session-specific information can be protected. See [RFC4566] for details.
Compared to an RTCP NACK message that triggers one or more retransmissions, a RAMS Request (RAMS-R) message can trigger a new burst stream to be sent by the retransmission server. Depending on the application-specific requirements and conditions existing at the time of the RAMS-R reception by the retransmission server, the resulting burst stream can potentially contain a large number of retransmission packets. Since these packets are sent faster than the nominal rate, RAMS consumes more resources on the retransmission servers, RTP receivers, and the network. In particular, this can make denial-of-service (DoS) attacks more intense and hence more harmful than attacks that target ordinary retransmission sessions.
As RAMS messages are sent as RTCP messages, counter-measures SHOULD be taken to prevent tampered or spoofed RTCP packets, following the suggestions given in [RFC4588]. Tampered RAMS-R messages can trigger inappropriate burst streams or alter the existing burst streams in an inappropriate way. For example, if the Max Receive Bitrate field is altered by a tampered RAMS-R message, the updated burst can overflow the buffer at the receiver side or, oppositely, can slow down the burst to the point that it becomes useless. Tampered RAMS Termination (RAMS-T) messages can terminate valid burst streams prematurely resulting in gaps in the received RTP packets. RAMS Information (RAMS-I) messages contain fields that are critical for a successful rapid acquisition. Any tampered information in the RAMS-I message can easily cause an RTP receiver to make wrong decisions. Consequently, the RAMS operation can fail.
Ver Steeg, et al. Standards Track [Page 45]
RFC 6285 RAMS June 2011
RTCP BYE messages are similar to RAMS-T messages in the sense that they can be used to stop an existing burst. The CNAME of an RTP receiver is used to bind the RTCP BYE message to an existing burst. Thus, one should be careful if the CNAMEs are reasonably easy to guess and off-path attacks can be performed. Also note that the CNAMEs might be redistributed to all participants in the multicast group (as in ASM or the simple feedback model of [RFC5760]).
The retransmission server has to consider if values indicated in a RAMS-R message are reasonable. For example, a request demanding a large value of many seconds in the Min RAMS Buffer Fill Requirement element should, unless special use cases exist, likely be rejected since it is likely to be an attempt to prolong a DoS attack on the retransmission server, RTP receiver, and/or the network. The Max Receive Bitrate could also be set to a very large value to try to get the retransmission server to cause massive congestion by bursting at a bitrate that will not be supported by the network. An RTP_Rx should also consider if the values for the Earliest Multicast Join Time and Burst Duration indicated by the retransmission server in a RAMS-I message are reasonable. For example, if the burst packets stop arriving and the join time is still significantly far into the future, this could be a sign of a man-in-the-middle attack where the RAMS-I message has been manipulated by an attacker.
A basic mitigation against DoS attacks introduced by an attacker injecting tampered RAMS messages is source address validation [RFC2827]. Also, most of the DoS attacks can be prevented by the integrity and authenticity checks enabled by Secure RTP (SRTP) [RFC3711]. However, an attack can still be started by legitimate endpoints that send several valid RAMS-R messages to a particular feedback target in a synchronized fashion and in a very short amount of time. Since a RAMS operation can temporarily consume a large amount of resources, a series of the RAMS-R messages can temporarily overload the retransmission server. In these circumstances, the retransmission server can, for example, reject incoming RAMS requests until its resources become available again. One means to ameliorate this threat is to apply a per-endpoint policing mechanism on the incoming RAMS requests. A reasonable policing mechanism should consider application-specific requirements and minimize false negatives.
In addition to the DoS attacks, man-in-the-middle and replay attacks will also be harmful. RAMS-R messages do not carry any information that allows the retransmission server to detect duplication or replay attacks. Thus, the possibility of a replay attack using a captured valid RAMS-R message exists unless a mitigation method such as Secure RTCP (SRTCP) is used. Similarly, RAMS-T messages can be replayed. The RAMS-I has a sequence number that makes replay attacks less
Ver Steeg, et al. Standards Track [Page 46]
RFC 6285 RAMS June 2011
likely but not impossible. Man-in-the-middle attacks that are capable of capturing, injecting, or modifying the RAMS messages can easily accomplish the attacks described above. Thus, cryptographic integrity and authentication is the only reliable protection. To protect the RTCP messages from man-in-the-middle and replay attacks, the RTP receivers and retransmission server SHOULD perform a Datagram Transport Layer Security (DTLS)-SRTP handshake [RFC5764] over the RTCP channel. Because there is no integrity-protected signaling channel between an RTP receiver and the retransmission server, the retransmission server MUST maintain a list of certificates owned by legitimate RTP receivers, or their certificates MUST be signed by a trusted Certificate Authority. Once the DTLS-SRTP security is established, non-SRTCP-protected messages received from a particular RTP receiver are ignored by the retransmission server. To reduce the impact of DTLS-SRTP overhead when communicating with different feedback targets on the same retransmission server, it is RECOMMENDED that RTP receivers and the retransmission server both support TLS Session Resumption without Server-side State [RFC5077]. To help scale SRTP to handle many RTP receivers asking for retransmissions of identical data, implementors may consider using the same SRTP key for SRTP data sent to the receivers [SRTP-EKT] and be aware that such key sharing allows those receivers to impersonate the sender. Thus, source address validation remains important.
[RFC4588] RECOMMENDS that cryptography mechanisms be used for the retransmission payload format to provide protection against known plain-text attacks. As discussed in [RFC4588], the retransmission payload format sets the timestamp field in the RTP header to the media timestamp of the original packet, and this does not compromise the confidentiality. Furthermore, if cryptography is used to provide security services on the original stream, then the same services, with equivalent cryptographic strength, MUST be provided on the retransmission stream per [RFC4588].
Finally, a retransmission server that has become subverted by an attacker is almost impossible to protect against as such a server can perform a large number of different actions to deny service to receivers.
This document registers a new attribute name in SDP.
SDP Attribute ("att-field"): Attribute name: rams-updates Long form: Support for Updated RAMS Request Messages Type of name: att-field Type of attribute: Media level Subject to charset: No Purpose: See this document Reference: [RFC6285] Values: None
This document creates a new sub-registry for the sub-feedback message type (SFMT) values to be used with the FMT value registered for RAMS messages. The registry is called the SFMT Values for RAMS Messages Registry. This registry is managed by the IANA according to the Specification Required policy of [RFC5226].
Ver Steeg, et al. Standards Track [Page 48]
RFC 6285 RAMS June 2011
The length of the SFMT field in the RAMS messages is a single octet, allowing 256 values. The registry is initialized with the following entries:
Value Name Reference ----- -------------------------------------------------- ------------ 0 Reserved [RFC6285] 1 RAMS Request [RFC6285] 2 RAMS Information [RFC6285] 3 RAMS Termination [RFC6285] 4-254 Unassigned - Specification Required 255 Reserved [RFC6285]
The SFMT values 0 and 255 are reserved for future use.
Any registration for an unassigned SFMT value needs to contain the following information:
o Contact information of the one doing the registration, including at least name, address, and email.
o A detailed description of what the new SFMT represents and how it shall be interpreted.
New RAMS functionality is intended to be introduced by using the extension mechanism within the existing RAMS message types not by introducing new message types unless it is absolutely necessary.
This document creates a new IANA TLV space registry for the RAMS extensions. The registry is called the RAMS TLV Space Registry. This registry is managed by the IANA according to the Specification Required policy of [RFC5226].
The length of the Type field in the TLV elements is a single octet, allowing 256 values. The Type values 0 and 255 are reserved for future use. The Type values between (and including) 128 and 254 are reserved for private extensions.
Ver Steeg, et al. Standards Track [Page 49]
RFC 6285 RAMS June 2011
The registry is initialized with the following entries:
Type Description Reference ---- ----------------------------------------------- ------------- 0 Reserved [RFC6285] 1 Requested Media Sender SSRC(s) [RFC6285] 2 Min RAMS Buffer Fill Requirement [RFC6285] 3 Max RAMS Buffer Fill Requirement [RFC6285] 4 Max Receive Bitrate [RFC6285] 5 Request for Preamble Only [RFC6285] 6 Supported Enterprise Number(s) [RFC6285] 7-30 Unassigned - Specification Required 31 Media Sender SSRC [RFC6285] 32 RTP Seqnum of the First Packet [RFC6285] 33 Earliest Multicast Join Time [RFC6285] 34 Burst Duration [RFC6285] 35 Max Transmit Bitrate [RFC6285] 36-60 Unassigned - Specification Required 61 Extended RTP Seqnum of First Multicast Packet [RFC6285] 62-127 Unassigned - Specification Required 128-254 Reserved for Private Use 255 Reserved [RFC6285]
Any registration for an unassigned Type value needs to contain the following information:
o Contact information of the one doing the registration, including at least name, address, and email.
o A detailed description of what the new TLV element represents and how it shall be interpreted.
This document creates a new IANA TLV space registry for the RAMS response codes. The registry is called the RAMS Response Code Space Registry. This registry is managed by the IANA according to the Specification Required policy of [RFC5226].
The length of the Response field is two octets, allowing 65536 codes. However, in this document, the response codes have been classified and registered following an HTTP-style code numbering. New response codes should be classified following the guidelines below:
200 RAMS request has been accepted [RFC6285] 201 Unicast burst has been completed [RFC6285]
400 Invalid RAMS-R message syntax [RFC6285] 401 Invalid min buffer requirement in RAMS-R message [RFC6285] 402 Invalid max buffer requirement in RAMS-R message [RFC6285] 403 Insufficient max bitrate requirement in RAMS-R message [RFC6285] 404 Invalid RAMS-T message syntax [RFC6285]
500 An unspecified BRS internal error has occurred [RFC6285] 501 BRS has insufficient bandwidth to start RAMS session [RFC6285] 502 Burst is terminated due to network congestion [RFC6285] 503 BRS has insufficient CPU cycles to start RAMS session [RFC6285] 504 RAMS functionality is not available on BRS [RFC6285] 505 RAMS functionality is not available for RTP_Rx [RFC6285] 506 RAMS functionality is not available for the requested multicast stream [RFC6285] 507 BRS has no valid starting point available for the requested multicast stream [RFC6285] 508 BRS has no reference information available for the requested multicast stream [RFC6285] 509 BRS has no RTP stream matching the requested SSRC [RFC6285] 510 RAMS request to acquire the entire session has been denied [RFC6285] 511 Only the preamble information is sent [RFC6285] 512 RAMS request has been denied due to a policy [RFC6285]
Ver Steeg, et al. Standards Track [Page 51]
RFC 6285 RAMS June 2011
Any registration for an unassigned Response code needs to contain the following information:
o Contact information of the one doing the registration, including at least name, address, and email.
o A detailed description of what the new Response code describes and how it shall be interpreted.
Dave Oran, Magnus Westerlund, and Colin Perkins have contributed significantly to this specification by providing text and solutions to some of the issues raised during the development of this specification.
The following individuals reviewed earlier versions of this specification and provided helpful comments: Joerg Ott, Roni Even, Dan Wing, Tony Faustini, Peilin Yang, Jeff Goldberg, Muriel Deschanel, Orit Levin, Guy Hirson, Tom Taylor, Xavier Marjou, Ye-Kui Wang, Zixuan Zou, Ingemar Johansson, Haibin Song, Ning Zong, Jonathan Lennox, Jose Rey, Sean Sheedy, and Keith Drage.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC2827] Ferguson, P. and D. Senie, "Network Ingress Filtering: Defeating Denial of Service Attacks which employ IP Source Address Spoofing", BCP 38, RFC 2827, May 2000.
[RFC3376] Cain, B., Deering, S., Kouvelas, I., Fenner, B., and A. Thyagarajan, "Internet Group Management Protocol, Version 3", RFC 3376, October 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.
[RFC3605] Huitema, C., "Real Time Control Protocol (RTCP) attribute in Session Description Protocol (SDP)", RFC 3605, October 2003.
Ver Steeg, et al. Standards Track [Page 52]
RFC 6285 RAMS June 2011
[RFC3711] Baugher, M., McGrew, D., Naslund, M., Carrara, E., and K. Norrman, "The Secure Real-time Transport Protocol (SRTP)", RFC 3711, March 2004.
[RFC3810] Vida, R. and L. Costa, "Multicast Listener Discovery Version 2 (MLDv2) for IPv6", RFC 3810, June 2004.
[RFC4566] Handley, M., Jacobson, V., and C. Perkins, "SDP: Session Description Protocol", RFC 4566, July 2006.
[RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, "Extended RTP Profile for Real-time Transport Control Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 2006.
[RFC4588] Rey, J., Leon, D., Miyazaki, A., Varsa, V., and R. Hakenberg, "RTP Retransmission Payload Format", RFC 4588, July 2006.
[RFC4604] Holbrook, H., Cain, B., and B. Haberman, "Using Internet Group Management Protocol Version 3 (IGMPv3) and Multicast Listener Discovery Protocol Version 2 (MLDv2) for Source- Specific Multicast", RFC 4604, August 2006.
[RFC5077] Salowey, J., Zhou, H., Eronen, P., and H. Tschofenig, "Transport Layer Security (TLS) Session Resumption without Server-Side State", RFC 5077, January 2008.
[RFC5234] Crocker, D. and P. Overell, "Augmented BNF for Syntax Specifications: ABNF", STD 68, RFC 5234, January 2008.
[RFC5285] Singer, D. and H. Desineni, "A General Mechanism for RTP Header Extensions", RFC 5285, July 2008.
[RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size Real-Time Transport Control Protocol (RTCP): Opportunities and Consequences", RFC 5506, April 2009.
[RFC5576] Lennox, J., Ott, J., and T. Schierl, "Source-Specific Media Attributes in the Session Description Protocol (SDP)", RFC 5576, June 2009.
[RFC5760] Ott, J., Chesterfield, J., and E. Schooler, "RTP Control Protocol (RTCP) Extensions for Single-Source Multicast Sessions with Unicast Feedback", RFC 5760, February 2010.
[RFC5761] Perkins, C. and M. Westerlund, "Multiplexing RTP Data and Control Packets on a Single Port", RFC 5761, April 2010.
Ver Steeg, et al. Standards Track [Page 53]
RFC 6285 RAMS June 2011
[RFC5764] McGrew, D. and E. Rescorla, "Datagram Transport Layer Security (DTLS) Extension to Establish Keys for the Secure Real-time Transport Protocol (SRTP)", RFC 5764, May 2010.
[RFC5888] Camarillo, G. and H. Schulzrinne, "The Session Description Protocol (SDP) Grouping Framework", RFC 5888, June 2010.
[RFC6051] Perkins, C. and T. Schierl, "Rapid Synchronisation of RTP Flows", RFC 6051, November 2010.
[RFC6128] Begen, A., "RTP Control Protocol (RTCP) Port for Source- Specific Multicast (SSM) Sessions", RFC 6128, February 2011.
[RFC6284] Begen, A., Wing, D., and T. Van Caenegem, "Port Mapping Between Unicast and Multicast RTP Sessions", RFC 6284, June 2011.