| RFC 8837 | WebRTC QoS | January 2021 |
| Jones, et al. | Standards Track | [Page] |
Networks can provide different forwarding treatments for individual packets based on Differentiated Services Code Point (DSCP) values on a per-hop basis. This document provides the recommended DSCP values for web browsers to use for various classes of Web Real-Time Communication (WebRTC) traffic.¶
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 7841.¶
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc8837.¶
Copyright (c) 2021 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 (https://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.¶
Differentiated Services Code Point (DSCP) [RFC2474] packet marking can help provide QoS in some environments. This specification provides default packet marking for browsers that support WebRTC applications, but does not change any advice or requirements in other RFCs. The contents of this specification are intended to be a simple set of implementation recommendations based on previous RFCs.¶
Networks in which these DSCP markings are beneficial (likely to improve QoS for WebRTC traffic) include:¶
There are cases where these DSCP markings do not help but, aside from possible priority inversion for "Less-than-Best-Effort traffic" (see Section 5), they seldom make things worse if packets are marked appropriately.¶
DSCP values are, in principle, site specific with each site selecting its own code points for controlling per-hop behavior to influence the QoS for transport-layer flows. However, in the WebRTC use cases, the browsers need to set them to something when there is no site-specific information. This document describes a subset of DSCP code point values drawn from existing RFCs and common usage for use with WebRTC applications. These code points are intended to be the default values used by a WebRTC application. While other values could be used, using a non-default value may result in unexpected per-hop behavior. It is RECOMMENDED that WebRTC applications use non-default values only in private networks that are configured to use different values.¶
This specification defines inputs that are provided by the WebRTC application hosted in the browser that aid the browser in determining how to set the various packet markings. The specification also defines the mapping from abstract QoS policies (flow type, priority level) to those packet markings.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
The terms "browser" and "non-browser" are defined in [RFC7742] and carry the same meaning in this document.¶
This document is a complement to [RFC7657], which describes the interaction between DSCP and real-time communications. That RFC covers the implications of using various DSCP values, particularly focusing on the Real-time Transport Protocol (RTP) [RFC3550] streams that are multiplexed onto a single transport-layer flow.¶
There are a number of guidelines specified in [RFC7657] that apply to marking traffic sent by WebRTC applications, as it is common for multiple RTP streams to be multiplexed on the same transport-layer flow. Generally, the RTP streams would be marked with a value as appropriate from Table 1. A WebRTC application might also multiplex data channel [RFC8831] traffic over the same 5-tuple as RTP streams, which would also be marked per that table. The guidance in [RFC7657] says that all data channel traffic would be marked with a single value that is typically different from the value(s) used for RTP streams multiplexed with the data channel traffic over the same 5-tuple, assuming RTP streams are marked with a value other than Default Forwarding (DF). This is expanded upon further in the next section.¶
This specification does not change or override the advice in any other RFCs about setting packet markings. Rather, it simply selects a subset of DSCP values that is relevant in the WebRTC context.¶
The DSCP value set by the endpoint is not trusted by the network. In addition, the DSCP value may be remarked at any place in the network for a variety of reasons to any other DSCP value, including the DF value to provide basic best-effort service. Even so, there is a benefit to marking traffic even if it only benefits the first few hops. The implications are discussed in Section 3.2 of [RFC7657]. Further, a mitigation for such action is through an authorization mechanism. Such an authorization mechanism is outside the scope of this document.¶
This document recommends DSCP values for two classes of WebRTC flows:¶
Each of the RTP streams and distinct data channels consist of all of the packets associated with an independent media entity, so an RTP stream or distinct data channel is not always equivalent to a transport-layer flow defined by a 5-tuple (source address, destination address, source port, destination port, and protocol). There may be multiple RTP streams and data channels multiplexed over the same 5-tuple, with each having a different level of importance to the application and, therefore, potentially marked using different DSCP values than another RTP stream or data channel within the same transport-layer flow. (Note that there are restrictions with respect to marking different data channels carried within the same Stream Control Transmission Protocol (SCTP) association as outlined in Section 5.)¶
The following are the inputs provided by the WebRTC application to the browser:¶
[RFC8835] defines in more detail what an individual flow is within the WebRTC context and priorities for media and data flows.¶
Currently in WebRTC, media sent over RTP is assumed to be interactive [RFC8835] and browser APIs do not exist to allow an application to differentiate between interactive and non-interactive video.¶
The DSCP values for each flow type of interest to WebRTC based on application priority are shown in Table 1. These values are based on the framework and recommended values in [RFC4594]. A web browser SHOULD use these values to mark the appropriate media packets. More information on Expedited Forwarding (EF) and Assured Forwarding (AF) can be found in [RFC3246] and [RFC2597], respectively. DF is Default Forwarding, which provides the basic best-effort service [RFC2474].¶
WebRTC's use of multiple DSCP values may result in packets with certain DSCP values being blocked by a network. See Section 4.2 of [RFC8835] for further discussion, including how WebRTC implementations establish and maintain connectivity when such blocking is encountered.¶
| Flow Type | Very Low | Low | Medium | High |
|---|---|---|---|---|
| Audio | LE (1) | DF (0) | EF (46) | EF (46) |
| Interactive Video with or without Audio | LE (1) | DF (0) | AF42, AF43 (36, 38) | AF41, AF42 (34, 36) |
| Non-Interactive Video with or without Audio | LE (1) | DF (0) | AF32, AF33 (28, 30) | AF31, AF32 (26, 28) |
| Data | LE (1) | DF (0) | AF11 | AF21 |
The application priority, indicated by the columns "Very Low", "Low", "Medium", and "High", signifies the relative importance of the flow within the application. It is an input that the browser receives to assist in selecting the DSCP value and adjusting the network transport behavior.¶
The above table assumes that packets marked with LE are treated as lower effort (i.e., "less than best effort"), such as the LE behavior described in [RFC8622]. However, the treatment of LE is implementation dependent. If an implementation treats LE as other than "less than best effort", then the actual priority (or, more precisely, the per-hop behavior) of the packets may be changed from what is intended. It is common for LE to be treated the same as DF, so applications and browsers using LE cannot assume that LE will be treated differently than DF [RFC7657]. During development of this document, the CS1 DSCP was recommended for "very low" application priority traffic; implementations that followed that recommendation SHOULD be updated to use the LE DSCP instead of the CS1 DSCP.¶
Implementers should also note that excess EF traffic is dropped. This could mean that a packet marked as EF may not get through, although the same packet marked with a different DSCP value would have gotten through. This is not a flaw, but how excess EF traffic is intended to be treated.¶
The browser SHOULD first select the flow type of the flow. Within the flow type, the relative importance of the flow SHOULD be used to select the appropriate DSCP value.¶
Currently, all WebRTC video is assumed to be interactive [RFC8835], for which the interactive video DSCP values in Table 1 SHOULD be used. Browsers MUST NOT use the AF3x DSCP values (for non-interactive video in Table 1) for WebRTC applications. Non-browser implementations of WebRTC MAY use the AF3x DSCP values for video that is known not to be interactive, e.g., all video in a WebRTC video playback application that is not implemented in a browser.¶
The combination of flow type and application priority provides specificity and helps in selecting the right DSCP value for the flow. All packets within a flow SHOULD have the same application priority. In some cases, the selected application priority cell may have multiple DSCP values, such as AF41 and AF42. These offer different drop precedences. The different drop precedence values provide additional granularity in classifying packets within a flow. For example, in a video conference, the video flow may have medium application priority, thus either AF42 or AF43 may be selected. More important video packets (e.g., a video picture or frame encoded without any dependency on any prior pictures or frames) might be marked with AF42 and less important packets (e.g., a video picture or frame encoded based on the content of one or more prior pictures or frames) might be marked with AF43 (e.g., receipt of the more important packets enables a video renderer to continue after one or more packets are lost).¶
It is worth noting that the application priority is utilized by the coupled congestion control mechanism for media flows per [RFC8699] and the SCTP scheduler for data channel traffic per [RFC8831].¶
For reasons discussed in Section 6 of [RFC7657], if multiple flows are multiplexed using a reliable transport (e.g., TCP), then all of the packets for all flows multiplexed over that transport-layer flow MUST be marked using the same DSCP value. Likewise, all WebRTC data channel packets transmitted over an SCTP association MUST be marked using the same DSCP value, regardless of how many data channels (streams) exist or what kind of traffic is carried over the various SCTP streams. In the event that the browser wishes to change the DSCP value in use for an SCTP association, it MUST reset the SCTP congestion controller after changing values. However, frequent changes in the DSCP value used for an SCTP association are discouraged, as this would defeat any attempts at effectively managing congestion. It should also be noted that any change in DSCP value that results in a reset of the congestion controller puts the SCTP association back into slow start, which may have undesirable effects on application performance.¶
For the data channel traffic multiplexed over an SCTP association, it is RECOMMENDED that the DSCP value selected be the one associated with the highest priority requested for all data channels multiplexed over the SCTP association. Likewise, when multiplexing multiple flows over a TCP connection, the DSCP value selected SHOULD be the one associated with the highest priority requested for all multiplexed flows.¶
If a packet enters a network that has no support for a flow-type-application priority combination specified in Table 1, then the network node at the edge will remark the DSCP value based on policies. This could result in the flow not getting the network treatment it expects based on the original DSCP value in the packet. Subsequently, if the packet enters a network that supports a larger number of these combinations, there may not be sufficient information in the packet to restore the original markings. Mechanisms for restoring such original DSCP is outside the scope of this document.¶
In summary, DSCP marking provides neither guarantees nor promised levels of service. However, DSCP marking is expected to provide a statistical improvement in real-time service as a whole. The service provided to a packet is dependent upon the network design along the path, as well as the network conditions at every hop.¶
Since the JavaScript application specifies the flow type and application priority that determine the media flow DSCP values used by the browser, the browser could consider application use of a large number of higher priority flows to be suspicious. If the server hosting the JavaScript application is compromised, many browsers within the network might simultaneously transmit flows with the same DSCP marking. The Diffserv architecture requires ingress traffic conditioning for reasons that include protecting the network from this sort of attack.¶
Otherwise, this specification does not add any additional security implications beyond those addressed in the following DSCP-related specifications. For security implications on use of DSCP, please refer to Section 7 of [RFC7657] and Section 6 of [RFC4594]. Please also see [RFC8826] as an additional reference.¶
This document has no IANA actions.¶
This specification contains downwards references to [RFC4594] and [RFC7657]. However, the parts of the former RFCs used by this specification are sufficiently stable for these downward references. The guidance in the latter RFC is necessary to understand the Diffserv technology used in this document and the motivation for the recommended DSCP values and procedures.¶
Thanks to David Black, Magnus Westerlund, Paolo Severini, Jim Hasselbrook, Joe Marcus, Erik Nordmark, Michael Tüxen, and Brian Carpenter for their invaluable input.¶
This document is dedicated to the memory of James Polk, a long-time friend and colleague. James made important contributions to this specification, including serving initially as one of the primary authors. The IETF global community mourns his loss and he will be missed dearly.¶