idnits 2.17.1 draft-ietf-rmcat-cc-requirements-00.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The abstract seems to contain references ([I-D.jesup-rtp-congestion-reqs]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 15, 2013) is 3937 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'RFC5506' is defined on line 351, but no explicit reference was found in the text == Outdated reference: A later version (-19) exists of draft-ietf-rtcweb-overview-06 == Outdated reference: A later version (-02) exists of draft-nichols-tsvwg-codel-01 -- Obsolete informational reference (is this intentional?): RFC 2309 (Obsoleted by RFC 7567) Summary: 1 error (**), 0 flaws (~~), 4 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group R. Jesup 3 Internet-Draft Mozilla 4 Intended status: Informational July 15, 2013 5 Expires: January 16, 2014 7 Congestion Control Requirements For RMCAT 8 draft-ietf-rmcat-cc-requirements-00 10 Abstract 12 Congestion control is needed for all data transported across the 13 Internet, in order to promote fair usage and prevent congestion 14 collapse. The requirements for interactive, point-to-point real time 15 multimedia, which needs by low-delay, semi-reliable data delivery, 16 are different from the requirements for bulk transfer like FTP or 17 bursty transfers like Web pages, and the TCP algorithms are not 18 suitable for this traffic. 20 This document attempts to describe a set of requirements that can be 21 used to evaluate other congestion control mechanisms in order to 22 figure out their fitness for this purpose, and in particular to 23 provide a set of possible requirements for proposals coming out of 24 the RMCAT Working Group. 26 This document is derived from draft-jesup-rtp-congestion-reqs 27 [I-D.jesup-rtp-congestion-reqs]. 29 Requirements Language 31 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 32 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 33 document are to be interpreted as described in RFC 2119 [RFC2119]. 35 Status of This Memo 37 This Internet-Draft is submitted in full conformance with the 38 provisions of BCP 78 and BCP 79. 40 Internet-Drafts are working documents of the Internet Engineering 41 Task Force (IETF). Note that other groups may also distribute 42 working documents as Internet-Drafts. The list of current Internet- 43 Drafts is at http://datatracker.ietf.org/drafts/current/. 45 Internet-Drafts are draft documents valid for a maximum of six months 46 and may be updated, replaced, or obsoleted by other documents at any 47 time. It is inappropriate to use Internet-Drafts as reference 48 material or to cite them other than as "work in progress." 49 This Internet-Draft will expire on January 16, 2014. 51 Copyright Notice 53 Copyright (c) 2013 IETF Trust and the persons identified as the 54 document authors. All rights reserved. 56 This document is subject to BCP 78 and the IETF Trust's Legal 57 Provisions Relating to IETF Documents 58 (http://trustee.ietf.org/license-info) in effect on the date of 59 publication of this document. Please review these documents 60 carefully, as they describe your rights and restrictions with respect 61 to this document. Code Components extracted from this document must 62 include Simplified BSD License text as described in Section 4.e of 63 the Trust Legal Provisions and are provided without warranty as 64 described in the Simplified BSD License. 66 Table of Contents 68 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 69 2. Requirements . . . . . . . . . . . . . . . . . . . . . . . . 3 70 3. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 6 71 4. Security Considerations . . . . . . . . . . . . . . . . . . . 7 72 5. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 7 73 6. References . . . . . . . . . . . . . . . . . . . . . . . . . 7 74 6.1. Normative References . . . . . . . . . . . . . . . . . . 7 75 6.2. Informative References . . . . . . . . . . . . . . . . . 7 76 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 8 78 1. Introduction 80 The traditional TCP congestion control requirements were developed in 81 order to promote efficient use of the Internet for reliable bulk 82 transfer of non-time-critical data, such as transfer of large files. 83 They have also been used successfully to govern the reliable transfer 84 of smaller chunks of data in "as fast as possible" mode, such as when 85 fetching Web pages. 87 These algorithms have also been used for transfer of media streams 88 that are viewed in a non-interactive manner, such as "streaming" 89 video, where having the data ready when the viewer wants it is 90 important, but the exact timing of the delivery is not. 92 When doing real time interactive media, the requirements are 93 different; one needs to provide the data continuously, within a very 94 limited time window (no more than 100s of milliseconds end-to-end 95 delay), the sources of data may be able to adapt the amount of data 96 that needs sending within fairly wide margins, and may tolerate some 97 amount of packet loss, but since the data is generated in real time, 98 sending "future" data is impossible, and since it's consumed in real 99 time, data delivered late is useless. 101 One particular protocol portfolio being developed for this use case 102 is WebRTC [I-D.ietf-rtcweb-overview], which envisions sending 103 multiple RTP-based flows between two peers, in conjunction with data 104 flows, all at the same time, without having special arrangements with 105 the intervening service providers. 107 Given that this use case is the focus of this document, use cases 108 involving noninteractive media such as YouTube-like video streaming, 109 and use cases using multicast/broadcast-type technologies, are out of 110 scope. 112 The terminology defined in [I-D.ietf-rtcweb-overview] is used in this 113 memo. 115 2. Requirements 117 1. The congestion control algorithm must attempt to provide as-low- 118 as-possible-delay transit for real-time traffic while still 119 providing a useful amount of bandwidth, even when faced with 120 intermediate bottlenecks and competing flows. There may be 121 lower limits on the amount of bandwidth that is useful, but this 122 is largely application-specific and the application may be able 123 to modify or remove flows in order allow some useful flows to 124 get enough bandwidth. (Example: not enough bandwidth for low- 125 latency video+audio, but enough for audio-only.) 127 a. It should also deal well with routing changes and interface 128 changes (WiFi to 3G data, etc) which may radically change 129 the bandwidth available. 131 2. The algorithm must be fair to other flows, both realtime flows 132 (such as other instances of itself), and TCP flows, both long- 133 lived and bursts such as the traffic generated by a typical web 134 browsing session. Note that 'fair' is a rather hard-to-define 135 term. 137 a. The algorithm must not overreact to short-term bursts (such 138 as web-browsing) which can quickly saturate a local- 139 bottleneck router or link, but also clear quickly, and 140 should recover quickly when the burst ends. This is 141 inherently at odds with the need to react quickly-enough to 142 avoid queue buildup. 144 b. We will need make some evaluation of fairness, but deciding 145 what is "fair" is a tough question and likely to be 146 partially subjective, but we should specify some of the 147 inputs needed in order to select among algorithms and 148 tunings presented as options. 150 c. The critical issue here is to have enough information for 151 the WG members to decide if an algorithm is "fair", and how 152 "unfair" it is (to other flows or to itself) in various edge 153 and corner cases. 155 3. The algorithm should where possible merge information across 156 multiple RTP streams between the same endpoints, whether or not 157 they're multiplexed on the same ports, in order to allow 158 congestion control of the set of streams together instead of as 159 multiple independent streams. This allows better overall 160 bandwidth management, faster response to changing conditions, 161 and fairer sharing of bandwidth with other network users. 162 Alternatively, it should work with an external bandwidth control 163 framework to coordinate bandwidth usage. 165 a. If possible, it should also share information and adaptation 166 with other non-RTP flows between the same endpoints, such as 167 a WebRTC data channel 169 b. The most correlated bandwidth usage would be with other 170 flows on the same 5-tuple, but there may be use in 171 coordinating measurement and control of the local link(s). 173 4. The algorithm should not require any special support from 174 network elements (ECN, etc). As much as possible, it should 175 leverage existing information about the incoming flows to 176 provide feedback to the sender. Examples of this information 177 are the packet arrival times, acknowledgments and feedback, 178 packet timestamps, packet sizes, packet losses. Extra 179 information could be added to the packets to provide more 180 detailed information on actual send times (as opposed to 181 sampling times), but should not be required. 183 a. When additional input signals such as ECN are available, 184 they should be utilized if possible. 186 5. Since the assumption here is a set of RTP streams, the 187 backchannel typically should be done via RTCP; the alternative 188 would be to include it in a reverse RTP channel using header 189 extensions. 191 a. In order to react sufficiently quickly, the AVPF/SAVPF RTP 192 profile[RFC4585] MUST be used 194 b. Note that in some cases, backchannel messages may be delayed 195 until the RTCP channel can be allocated enough bandwidth, 196 even under AVPF rules. This may also imply negotiating a 197 higher maximum percentage for RTCP data or allowing RMCAT 198 solutions to violate or modify the rules specified for AVPF. 200 c. Note that RTCP is of course unreliable 202 d. Bandwidth for the feedback messages should be minimized 203 (such as via RFC 5506 [RFC5506]to allow RTCP without SR/RR) 205 e. Header extensions would avoid the RTCP timing rules issues, 206 and allow the application to allocate bandwidth as needed 207 for the congestion algorithm. 209 f. Backchannel data should be minimized to avoid taking too 210 much reverse-channel bandwidth (since this will often be 211 used in a bidirectional set of flows). In areas of 212 stability, backchannel data may be sent more infrequently so 213 long as algorithm stability and fairness are maintained. 214 When the channel is unstable or has not yet reached 215 equilibrium after a change, backchannel feedback may be more 216 frequent and use more reverse-channel bandwidth. This is an 217 area with considerable flexibility of design, and different 218 approaches to backchannel messages and frequency are 219 expected to be evaluated. 221 6. Where possible and helpful, the algorithm should leverage and 222 piggyback on other RTP/RTCP communications, such as SR/RR, rctp- 223 fb PLI, RPSI, SLI or application-specific NACK messages (such as 224 for loss information), and also reverse-direction RTP. 226 7. The algorithm should sense the unexpected lack of backchannel 227 information as a possible indication of a channel overuse 228 problem and react accordingly to avoid burst events causing a 229 congestion collapse. 231 8. It should attempt to avoid bandwidth 'collapse' when facing a 232 long-lived saturating TCP flow or flows. (I.e. a classic delay- 233 sensitive algorithm will reduce bandwidth to keep delay down 234 until the TCP flow has all the bandwidth). See the Cx-TCP 235 algorithm discussed in a recent Transactions On Networking 236 [cx-tcp] for an example of a delay-sensitive congestion-control 237 algorithm that transitions to a loss-based mode when competing 238 with TCP flows - at the cost of increased delay. 240 9. The algorithm should be stable and low-delay when faced with 241 active queue management (AQM) such as RED [RFC2309] or CoDel 242 [I-D.nichols-tsvwg-codel] or fq-codel in the channel. 244 10. The algorithm should quickly adapt to initial network conditions 245 at the start of a flow. This should occur both if the initial 246 bandwidth is above or below the bottleneck bandwidth. 248 a. The startup adaptation may be faster than adaptation later 249 in a flow. It should allow for both slow-start operation 250 (adapt up) and history-based startup (start at a point 251 expected to be at or below channel bandwidth from historical 252 information, which may need to adapt down quickly if the 253 initial guess is wrong). Starting too low and/or adapting 254 up too slowly can cause a critical point in a personal 255 communication to be poor ("Hello!"). 257 b. Starting over-bandwidth causes other problems for user 258 experience, so there's a tension here. 260 c. Alternative methods to help startup like probing during 261 setup with dummy data may be useful in some applications. 263 d. A flow may need to change adaptation rates due to network 264 conditions or changes in the provided flows (such as un- 265 muting or sending data after a gap). 267 11. It should be evaluated in how it works both with backbone-router 268 bottlenecks, (asymmetric) local-loop bottlenecks, and local-lan 269 (WiFi/etc) bottlenecks, and in competition with varying numbers 270 and types of streams (TCP, TCP variants in use, LEDBAT 271 [I-D.ietf-ledbat-congestion], inflexible VoIP UDP flows). 273 12. It should be stable if the RTP streams are halted or 274 discontinuous (VAD/DTX). 276 a. After a resumption of RTP data it may adapt more quickly 277 (similar to the start of a flow), and previous bandwidth 278 estimates may need to be aged or thrown away. 280 3. IANA Considerations 282 This document makes no request of IANA. 284 Note to RFC Editor: this section may be removed on publication as an 285 RFC. 287 4. Security Considerations 289 An attacker with the ability to delete, delay or insert messages in 290 the flow can fake congestion signals, unless they are passed on a 291 tamper-proof path. Since some possible algorithms depend on the 292 timing of packet arrival, even a traditional protected channel does 293 not fully mitigate such attacks. 295 An attack that reduces bandwidth is not necessarily significant, 296 since an on-path attacker could break the connection by discarding 297 all packets. Attacks that increase the percieved available bandwidth 298 are concievable, and need to be evaluated. 300 Algorithm designers SHOULD consider the possibility of malicious on- 301 path attackers. 303 5. Acknowledgements 305 This document is the result of discussions in various fora of the 306 WebRTC effort, in particular on the rtp-congestion@alvestrand.no 307 mailing list. Many people contributed their thoughts to this. 309 6. References 311 6.1. Normative References 313 [I-D.ietf-rtcweb-overview] 314 Alvestrand, H., "Overview: Real Time Protocols for Brower- 315 based Applications", draft-ietf-rtcweb-overview-06 (work 316 in progress), February 2013. 318 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 319 Requirement Levels", BCP 14, RFC 2119, March 1997. 321 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 322 "Extended RTP Profile for Real-time Transport Control 323 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 324 2006. 326 6.2. Informative References 328 [I-D.ietf-ledbat-congestion] 329 Shalunov, S., Hazel, G., Iyengar, J., and M. Kuehlewind, 330 "Low Extra Delay Background Transport (LEDBAT)", draft- 331 ietf-ledbat-congestion-10 (work in progress), September 332 2012. 334 [I-D.jesup-rtp-congestion-reqs] 335 Jesup, R. and H. Alvestrand, "Congestion Control 336 Requirements For Real Time Media", draft-jesup-rtp- 337 congestion-reqs-00 (work in progress), March 2012. 339 [I-D.nichols-tsvwg-codel] 340 Nichols, K. and V. Jacobson, "Controlled Delay Active 341 Queue Management", draft-nichols-tsvwg-codel-01 (work in 342 progress), February 2013. 344 [RFC2309] Braden, B., Clark, D., Crowcroft, J., Davie, B., Deering, 345 S., Estrin, D., Floyd, S., Jacobson, V., Minshall, G., 346 Partridge, C., Peterson, L., Ramakrishnan, K., Shenker, 347 S., Wroclawski, J., and L. Zhang, "Recommendations on 348 Queue Management and Congestion Avoidance in the 349 Internet", RFC 2309, April 1998. 351 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 352 Real-Time Transport Control Protocol (RTCP): Opportunities 353 and Consequences", RFC 5506, April 2009. 355 [cx-tcp] Budzisz, L., Stanojevic, R., Schlote, A., Baker, F., and 356 R. Shorten, "On the Fair Coexistence of Loss- and Delay- 357 Based TCP", December 2011. 359 Author's Address 361 Randell Jesup 362 Mozilla 363 USA 365 Email: randell-ietf@jesup.org