idnits 2.17.1 draft-ietf-rmcat-cc-requirements-09.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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document doesn't use any RFC 2119 keywords, yet seems to have RFC 2119 boilerplate text. -- The document date (December 12, 2014) is 3422 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'RFC5506' is defined on line 507, but no explicit reference was found in the text == Outdated reference: A later version (-19) exists of draft-ietf-rtcweb-overview-13 == Outdated reference: A later version (-18) exists of draft-ietf-avtcore-rtp-circuit-breakers-08 == Outdated reference: A later version (-13) exists of draft-ietf-rtcweb-data-channel-12 Summary: 0 errors (**), 0 flaws (~~), 6 warnings (==), 1 comment (--). 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 Z. Sarker, Ed. 5 Expires: June 15, 2015 Ericsson 6 December 12, 2014 8 Congestion Control Requirements for Interactive Real-Time Media 9 draft-ietf-rmcat-cc-requirements-09 11 Abstract 13 Congestion control is needed for all data transported across the 14 Internet, in order to promote fair usage and prevent congestion 15 collapse. The requirements for interactive, point-to-point real-time 16 multimedia, which needs low-delay, semi-reliable data delivery, are 17 different from the requirements for bulk transfer like FTP or bursty 18 transfers like Web pages. Due to an increasing amount of RTP-based 19 real-time media traffic on the Internet (e.g. with the introduction 20 of the Web Real-Time Communication (WebRTC)), it is especially 21 important to ensure that this kind of traffic is congestion 22 controlled. 24 This document describes a set of requirements that can be used to 25 evaluate other congestion control mechanisms in order to figure out 26 their fitness for this purpose, and in particular to provide a set of 27 possible requirements for real-time media congestion avoidance 28 technique. 30 Requirements Language 32 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 33 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 34 document are to be interpreted as described in RFC 2119 [RFC2119]. 35 The terms are presented in many cases using lowercase for 36 readability. 38 Status of This Memo 40 This Internet-Draft is submitted in full conformance with the 41 provisions of BCP 78 and BCP 79. 43 Internet-Drafts are working documents of the Internet Engineering 44 Task Force (IETF). Note that other groups may also distribute 45 working documents as Internet-Drafts. The list of current Internet- 46 Drafts is at http://datatracker.ietf.org/drafts/current/. 48 Internet-Drafts are draft documents valid for a maximum of six months 49 and may be updated, replaced, or obsoleted by other documents at any 50 time. It is inappropriate to use Internet-Drafts as reference 51 material or to cite them other than as "work in progress." 53 This Internet-Draft will expire on June 15, 2015. 55 Copyright Notice 57 Copyright (c) 2014 IETF Trust and the persons identified as the 58 document authors. All rights reserved. 60 This document is subject to BCP 78 and the IETF Trust's Legal 61 Provisions Relating to IETF Documents 62 (http://trustee.ietf.org/license-info) in effect on the date of 63 publication of this document. Please review these documents 64 carefully, as they describe your rights and restrictions with respect 65 to this document. Code Components extracted from this document must 66 include Simplified BSD License text as described in Section 4.e of 67 the Trust Legal Provisions and are provided without warranty as 68 described in the Simplified BSD License. 70 Table of Contents 72 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 73 2. Requirements . . . . . . . . . . . . . . . . . . . . . . . . 3 74 3. Deficiencies of existing mechanisms . . . . . . . . . . . . . 8 75 4. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 9 76 5. Security Considerations . . . . . . . . . . . . . . . . . . . 9 77 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 10 78 7. References . . . . . . . . . . . . . . . . . . . . . . . . . 10 79 7.1. Normative References . . . . . . . . . . . . . . . . . . 10 80 7.2. Informative References . . . . . . . . . . . . . . . . . 10 81 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 11 83 1. Introduction 85 Most of today's TCP congestion control schemes were developed with a 86 focus on an use of the Internet for reliable bulk transfer of non- 87 time-critical data, such as transfer of large files. They have also 88 been used successfully to govern the reliable transfer of smaller 89 chunks of data in as short a time as possible, such as when fetching 90 Web pages. 92 These algorithms have also been used for transfer of media streams 93 that are viewed in a non-interactive manner, such as "streaming" 94 video, where having the data ready when the viewer wants it is 95 important, but the exact timing of the delivery is not. 97 When doing real-time interactive media, the requirements are 98 different; one needs to provide the data continuously, within a very 99 limited time window (no more than 100s of milliseconds end-to-end 100 delay), the sources of data may be able to adapt the amount of data 101 that needs sending within fairly wide margins but can be rate limited 102 by the application- even not always have data to send, and may 103 tolerate some amount of packet loss, but since the data is generated 104 in real-time, sending "future" data is impossible, and since it's 105 consumed in real-time, data delivered late is commonly useless. 107 While the requirements for real-time interactive media differ from 108 the requirements for the other flow types, these other flow types 109 will be present in the network. The congestion control algorithm for 110 real-time interactive media must work properly when these other flow 111 types are present as cross traffic on the network. 113 One particular protocol portfolio being developed for this use case 114 is WebRTC [I-D.ietf-rtcweb-overview], where one envisions sending 115 multiple flows using the Real-time Transport Protocol (RTP) [RFC3550] 116 between two peers, in conjunction with data flows, all at the same 117 time, without having special arrangements with the intervening 118 service providers. As RTP does not provide any congestion control 119 mechanism; a set of circuit breakers, such as 120 [I-D.ietf-avtcore-rtp-circuit-breakers], are required to protect the 121 network from excessive congestion caused by the non-congestion 122 controlled flows. When the real-time interactive media is congestion 123 controlled, it is recommended that the congestion control mechanism 124 operates within the constraints defined by these circuit breakers 125 when circuit breaker is present and that it should not cause 126 congestion collapse when circuit breaker is not implemented. 128 Given that this use case is the focus of this document, use cases 129 involving non-interactive media such as video streaming, and use 130 cases using multicast/broadcast-type technologies, are out of scope. 132 The terminology defined in [I-D.ietf-rtcweb-overview] is used in this 133 memo. 135 2. Requirements 137 1. The congestion control algorithm must attempt to provide as-low- 138 as-possible-delay transit for interactive real-time traffic 139 while still providing a useful amount of bandwidth. There may 140 be lower limits on the amount of bandwidth that is useful, but 141 this is largely application-specific and the application may be 142 able to modify or remove flows in order allow some useful flows 143 to get enough bandwidth. (Example: not enough bandwidth for 144 low-latency video+audio, but enough for audio-only.) 145 A. Jitter (variation in the bitrate over short time scales) 146 also is relevant, though moderate amounts of jitter will be 147 absorbed by jitter buffers. Transit delay should be 148 considered to track the short-term maximums of delay 149 including jitter. 151 B. It should provide this as-low-as-possible-delay transit and 152 minimize self-induced latency even when faced with 153 intermediate bottlenecks and competing flows. Competing 154 flows may limit what's possible to achieve. 156 C. It should be resilience to the effects of the events, such 157 as routing changes, which may alter or remove bottlenecks or 158 change the bandwidth available especially if there is a 159 reduction in available bandwidth or increase in observed 160 delay. It is expected that the mechanism reacts quickly to 161 the such events to avoid delay buildup. In the context of 162 this memo, a 'quick' reaction is on the order of a few RTTs, 163 subject to the constraints of the media codec, but is likely 164 within a second. Reaction on the next RTT is explicitly not 165 required, since many codecs cannot adapt their sending rate 166 that quickly, but equally response cannot be arbitrarily 167 delayed. 169 D. It should react quickly to handle both local and remote 170 interface changes (WLAN to 3G data, etc) which may radically 171 change the bandwidth available or bottlenecks, especially if 172 there is a reduction in available bandwidth or increase in 173 bottleneck delay. It is assumed that an interface change 174 can generate a notification to the algorithm. 176 E. The real-time interactive media applications can be rate 177 limited. This means the offered loads can be less than the 178 available bandwidth at any given moment, and may vary 179 dramatically over time, including dropping to no load and 180 then resuming a high load, such as in a mute/unmute 181 operation. Hence, the algorithm must be designed to handle 182 such behavior from media source or application. Note that 183 the reaction time between a change in the bandwidth 184 available from the algorithm and a change in the offered 185 load is variable, and may be different when increasing 186 versus decreasing. 188 F. The algorithm requires to avoid building up queues when 189 competing with short-term bursts of traffic (for example, 190 traffic generated by web-browsing) which can quickly 191 saturate a local-bottleneck router or link, but also clear 192 quickly. The algorithm should also react quickly to regain 193 its previous share of the bandwidth when the local- 194 bottleneck or link is cleared. 196 G. Similarly periodic bursty flows such as MPEG DASH 197 [MPEG_DASH] or proprietary media streaming algorithms may 198 compete in bursts with the algorithm, and may not be 199 adaptive within a burst. They are often layered on top of 200 TCP but use TCP in a bursty manner that can interact poorly 201 with competing flows during the bursts. The algorithm must 202 not increase the already existing delay buildup during those 203 bursts. Note that this competing traffic may be on a shared 204 access link, or the traffic burst may cause a shift in the 205 location of the bottleneck for the duration of the burst. 207 2. The algorithm must be fair to other flows, both real-time flows 208 (such as other instances of itself), and TCP flows, both long- 209 lived and bursts such as the traffic generated by a typical web 210 browsing session. Note that 'fair' is a rather hard-to-define 211 term. It should be fair with itself, giving fair share of the 212 bandwidth to multiple flows with similar RTTs, and if possible 213 to multiple flows with different RTTs. 215 A. Existing flows at a bottleneck must also be fair to new 216 flows to that bottleneck, and must allow new flows to ramp 217 up to a useful share of the bottleneck bandwidth as quickly 218 as possible. A useful share will depend on the media types 219 involved, total bandwidth available and the user experience 220 requirements of a particular service. Note that relative 221 RTTs may affect the rate new flows can ramp up to a 222 reasonable share. 224 3. The algorithm should not starve competing TCP flows, and should 225 as best as possible avoid starvation by TCP flows. 227 A. The congestion control should prioritise achieving a useful 228 share of the bandwidth depending on the media types and 229 total available bandwidth over achieving as low as possible 230 transit delay, when these two requirements are in conflict. 232 4. The algorithm should as quickly as possible adapt to initial 233 network conditions at the start of a flow. This should occur 234 both if the initial bandwidth is above or below the bottleneck 235 bandwidth. 237 A. The algorithm should allow different modes of adaptation for 238 example,the startup adaptation may be faster than adaptation 239 later in a flow. It should allow for both slow-start 240 operation (adapt up) and history-based startup (start at a 241 point expected to be at or below channel bandwidth from 242 historical information, which may need to adapt down quickly 243 if the initial guess is wrong). Starting too low and/or 244 adapting up too slowly can cause a critical point in a 245 personal communication to be poor ("Hello!"). Starting 246 over-bandwidth causes other problems for user experience, so 247 there's a tension here. Alternative methods to help startup 248 like probing during setup with dummy data may be useful in 249 some applications; in some cases there will be a 250 considerable gap in time between flow creation and the 251 initial flow of data. Again, A flow may need to change 252 adaptation rates due to network conditions or changes in the 253 provided flows (such as un-muting or sending data after a 254 gap). 256 5. The algorithm should be stable if the RTP streams are halted or 257 discontinuous (for example - Voice Activity Detection). 259 A. After stream resumption, the algorithm should attempt to 260 rapidly regain its previous share of the bandwidth; the 261 aggressiveness with which this is done will decay with the 262 length of the pause. 264 6. The algorithm should where possible merge information across 265 multiple RTP streams sent between two endpoints, when those RTP 266 streams share a common bottleneck, whether or not those streams 267 are multiplexed onto the same ports, in order to allow 268 congestion control of the set of streams together instead of as 269 multiple independent streams. This allows better overall 270 bandwidth management, faster response to changing conditions, 271 and fairer sharing of bandwidth with other network users. 273 A. The algorithm should also share information and adaptation 274 with other non-RTP flows between the same endpoints, such as 275 a WebRTC DataChannel [I-D.ietf-rtcweb-data-channel], when 276 possible. 278 B. When there are multiple streams across the same 5-tuple 279 coordinating their bandwidth use and congestion control, the 280 algorithm should allow the application to control the 281 relative split of available bandwidth. The most correlated 282 bandwidth usage would be with other flows on the same 283 5-tuple, but there may be use in coordinating measurement 284 and control of the local link(s). Use of information about 285 previous flows, especially on the same 5-tuple, may be 286 useful input to the algorithm, especially to startup 287 performance of a new flow. 289 7. The algorithm should not require any special support from 290 network elements to convey congestion related information to be 291 functional. As much as possible, it should leverage available 292 information about the incoming flow to provide feedback to the 293 sender. Examples of this information are the packet arrival 294 times, acknowledgements and feedback, packet timestamps, and 295 packet losses, Explicit Congestion Notification (ECN) [RFC3168]; 296 all of these can provide information about the state of the path 297 and any bottlenecks. However, the use of available information 298 is algorithm dependent. 300 A. Extra information could be added to the packets to provide 301 more detailed information on actual send times (as opposed 302 to sampling times), but should not be required. 304 8. Since the assumption here is a set of RTP streams, the 305 backchannel typically should be done via RTCP[RFC3550]; one 306 alternative would be to include it instead in a reverse RTP 307 channel using header extensions. 309 A. In order to react sufficiently quickly when using RTCP for a 310 backchannel, an RTP profile such as RTP/AVPF [RFC4585] or 311 RTP/SAVPF [RFC5124] that allows sufficiently frequent 312 feedback must be used. Note that in some cases, backchannel 313 messages may be delayed until the RTCP channel can be 314 allocated enough bandwidth, even under AVPF rules. This may 315 also imply negotiating a higher maximum percentage for RTCP 316 data or allowing solutions to violate or modify the rules 317 specified for AVPF. 319 B. Bandwidth for the feedback messages should be minimized 320 (such as via RFC 5506 [RFC5506]to allow RTCP without Sender 321 Reports/Receiver Reports) 323 C. Backchannel data should be minimized to avoid taking too 324 much reverse-channel bandwidth (since this will often be 325 used in a bidirectional set of flows). In areas of 326 stability, backchannel data may be sent more infrequently so 327 long as algorithm stability and fairness are maintained. 328 When the channel is unstable or has not yet reached 329 equilibrium after a change, backchannel feedback may be more 330 frequent and use more reverse-channel bandwidth. This is an 331 area with considerable flexibility of design, and different 332 approaches to backchannel messages and frequency are 333 expected to be evaluated. 335 9. Flows managed by this algorithm and flows competing against at a 336 bottleneck may have different DSCP[RFC5865] markings depending 337 on the type of traffic, or may be subject to flow-based QoS. A 338 particular bottleneck or section of the network path may or may 339 not honor DSCP markings. The algorithm should attempt to 340 leverage DSCP markings when they're available. 342 A. In WebRTC, a division of packets into 4 classes is 343 envisioned in order of priority: faster-than-audio, audio, 344 video, best-effort, and bulk-transfer. Typically the flows 345 managed by this algorithm would be audio or video in that 346 hierarchy, and feedback flows would be faster-than-audio. 348 10. The algorithm should sense the unexpected lack of backchannel 349 information as a possible indication of a channel overuse 350 problem and react accordingly to avoid burst events causing a 351 congestion collapse. 353 11. The algorithm should be stable and maintain low-delay when faced 354 with Active Queue Management (AQM) algorithms. Also note that 355 these algorithms may apply across multiple queues in the 356 bottleneck, or to a single queue 358 3. Deficiencies of existing mechanisms 360 Among the existing congestion control mechanisms TCP Friendly Rate 361 Control (TFRC) [RFC5348] is the one which claims to be suitable for 362 real-time interactive media. TFRC is, an equation based, congestion 363 control mechanism which provides reasonably fair share of the 364 bandwidth when competing with TCP flows and offers much lower 365 throughput variations than TCP. This is achieved by a slower 366 response to the available bandwidth change than TCP. TFRC is 367 designed to perform best with applications which has fixed packet 368 size and does not have fixed period between sending packets. 370 TFRC operates on detecting loss events and reacts to loss caused by 371 congestion by reducing its sending rate. It allows applications to 372 increase the sending rate until loss is observed in the flows. As it 373 is noted in IAB/IRTF report [RFC7295] large buffers are available in 374 the network elements which introduces additional delay in the 375 communication, it becomes important to take all possible congestion 376 indications into considerations. Looking at the current Internet 377 deployment, TFRC's only consideration of loss events as congestion 378 indication can be considered as biggest lacking. 380 A typical real-time interactive communication includes live encoded 381 audio and video flow(s). In such a communication scenario an audio 382 source typically needs fixed interval between packets, needs to vary 383 their segment size instead of their packet rate in response to 384 congestion and sends smaller packets, a variant of TFRC , Small- 385 Packet TFRC (TFRC-SP) [RFC4828] addresses the issues related to such 386 kind of sources ; a video source generally varies video frame sizes, 387 can produce large frames which need to be further fragmented to fit 388 into path Maximum Transmission Unit (MTU) size, and have almost fixed 389 interval between producing frames under a certain frame rate, TFRC is 390 known to be less optimal when using with such video sources. 392 There are also some mismatches between TFRC's design assumptions and 393 how the media sources in a typical real-time interactive application 394 works. TFRC is design to maintain smooth sending rate however media 395 sources can change rates in steps for both rate increase and rate 396 decrease. TFRC can operate in two modes - i) Bytes per second and 397 ii) packets per second, where typical real-time interactive media 398 sources operates on bit per second. There are also limitations on 399 how quickly the media sources can adapt to specific sending rates. 400 The modern video encoders can operate on a mode where they can vary 401 the output bitrate a lot depending on the way there are configured, 402 the current scene it is encoding and more. Therefore, it is possible 403 that the video source does not always output at a bitrate they are 404 allowed to. TFRC tries to raise its sending rate when transmitting 405 at maximum allowed rate and increases only twice the current 406 transmission rate hence it may create issues when the video source 407 vary their bitrates. 409 Moreover, there are number of studies on TFRC which shows it's 410 limitations which includes TFRC's unfairness on low statistically 411 multiplexed links, oscillatory behavior, performance issue in highly 412 dynamic loss rates conditions and more [CH09]. 414 Looking at all these deficiencies it can be concluded that the 415 requirements of congestion control mechanism for real-time 416 interactive media cannot be met by TFRC as defined in the standard. 418 4. IANA Considerations 420 This document makes no request of IANA. 422 Note to RFC Editor: this section may be removed on publication as an 423 RFC. 425 5. Security Considerations 427 An attacker with the ability to delete, delay or insert messages in 428 the flow can fake congestion signals, unless they are passed on a 429 tamper-proof path. Since some possible algorithms depend on the 430 timing of packet arrival, even a traditional protected channel does 431 not fully mitigate such attacks. 433 An attack that reduces bandwidth is not necessarily significant, 434 since an on-path attacker could break the connection by discarding 435 all packets. Attacks that increase the perceived available bandwidth 436 are conceivable, and need to be evaluated. Such attacks could result 437 in starvation of competing flows and permit amplification attacks. 439 Algorithm designers should consider the possibility of malicious on- 440 path attackers. 442 6. Acknowledgements 444 This document is the result of discussions in various fora of the 445 WebRTC effort, in particular on the rtp-congestion@alvestrand.no 446 mailing list. Many people contributed their thoughts to this. 448 7. References 450 7.1. Normative References 452 [I-D.ietf-rtcweb-overview] 453 Alvestrand, H., "Overview: Real Time Protocols for 454 Browser-based Applications", draft-ietf-rtcweb-overview-13 455 (work in progress), November 2014. 457 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 458 Requirement Levels", BCP 14, RFC 2119, March 1997. 460 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 461 Jacobson, "RTP: A Transport Protocol for Real-Time 462 Applications", STD 64, RFC 3550, July 2003. 464 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 465 "Extended RTP Profile for Real-time Transport Control 466 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 467 2006. 469 [RFC5124] Ott, J. and E. Carrara, "Extended Secure RTP Profile for 470 Real-time Transport Control Protocol (RTCP)-Based Feedback 471 (RTP/SAVPF)", RFC 5124, February 2008. 473 7.2. Informative References 475 [CH09] Choi, S. and M. Handley, "Designing TCP-Friendly Window- 476 based Congestion Control for Real-time Multimedia 477 Applications", PFLDNeT 2009 Workshop , May 2009. 479 [I-D.ietf-avtcore-rtp-circuit-breakers] 480 Perkins, C. and V. Singh, "Multimedia Congestion Control: 481 Circuit Breakers for Unicast RTP Sessions", draft-ietf- 482 avtcore-rtp-circuit-breakers-08 (work in progress), 483 December 2014. 485 [I-D.ietf-rtcweb-data-channel] 486 Jesup, R., Loreto, S., and M. Tuexen, "WebRTC Data 487 Channels", draft-ietf-rtcweb-data-channel-12 (work in 488 progress), September 2014. 490 [MPEG_DASH] 491 "Dynamic adaptive streaming over HTTP (DASH) -- Part 1: 492 Media presentation description and segment formats", April 493 2012. 495 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 496 of Explicit Congestion Notification (ECN) to IP", RFC 497 3168, September 2001. 499 [RFC4828] Floyd, S. and E. Kohler, "TCP Friendly Rate Control 500 (TFRC): The Small-Packet (SP) Variant", RFC 4828, April 501 2007. 503 [RFC5348] Floyd, S., Handley, M., Padhye, J., and J. Widmer, "TCP 504 Friendly Rate Control (TFRC): Protocol Specification", RFC 505 5348, September 2008. 507 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 508 Real-Time Transport Control Protocol (RTCP): Opportunities 509 and Consequences", RFC 5506, April 2009. 511 [RFC5865] Baker, F., Polk, J., and M. Dolly, "A Differentiated 512 Services Code Point (DSCP) for Capacity-Admitted Traffic", 513 RFC 5865, May 2010. 515 [RFC7295] Tschofenig, H., Eggert, L., and Z. Sarker, "Report from 516 the IAB/IRTF Workshop on Congestion Control for 517 Interactive Real-Time Communication", RFC 7295, July 2014. 519 Authors' Addresses 521 Randell Jesup 522 Mozilla 523 USA 525 Email: randell-ietf@jesup.org 526 Zaheduzzaman Sarker (editor) 527 Ericsson 528 Sweden 530 Email: zaheduzzaman.sarker@ericsson.com