idnits 2.17.1 draft-ietf-rmcat-eval-criteria-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 document seems to lack a both a reference to RFC 2119 and the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords. RFC 2119 keyword, line 705: '... 1. It MUST emit a packet at least ...' Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (January 31, 2014) is 3731 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'RFC5681' is defined on line 597, but no explicit reference was found in the text == Unused Reference: 'SA4-EVAL' is defined on line 600, but no explicit reference was found in the text == Unused Reference: 'SA4-LR' is defined on line 604, but no explicit reference was found in the text == Unused Reference: 'TCP-eval-suite' is defined on line 607, but no explicit reference was found in the text == Outdated reference: A later version (-09) exists of draft-ietf-rmcat-cc-requirements-00 == Outdated reference: A later version (-18) exists of draft-ietf-avtcore-rtp-circuit-breakers-01 Summary: 1 error (**), 0 flaws (~~), 7 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 RMCAT WG V. Singh 3 Internet-Draft J. Ott 4 Intended status: Informational Aalto University 5 Expires: August 04, 2014 January 31, 2014 7 Evaluating Congestion Control for Interactive Real-time Media 8 draft-ietf-rmcat-eval-criteria-00 10 Abstract 12 The Real-time Transport Protocol (RTP) is used to transmit media in 13 telephony and video conferencing applications. This document 14 describes the guidelines to evaluate new congestion control 15 algorithms for interactive point-to-point real-time media. 17 Status of This Memo 19 This Internet-Draft is submitted in full conformance with the 20 provisions of BCP 78 and BCP 79. 22 Internet-Drafts are working documents of the Internet Engineering 23 Task Force (IETF). Note that other groups may also distribute 24 working documents as Internet-Drafts. The list of current Internet- 25 Drafts is at http://datatracker.ietf.org/drafts/current/. 27 Internet-Drafts are draft documents valid for a maximum of six months 28 and may be updated, replaced, or obsoleted by other documents at any 29 time. It is inappropriate to use Internet-Drafts as reference 30 material or to cite them other than as "work in progress." 32 This Internet-Draft will expire on August 04, 2014. 34 Copyright Notice 36 Copyright (c) 2014 IETF Trust and the persons identified as the 37 document authors. All rights reserved. 39 This document is subject to BCP 78 and the IETF Trust's Legal 40 Provisions Relating to IETF Documents 41 (http://trustee.ietf.org/license-info) in effect on the date of 42 publication of this document. Please review these documents 43 carefully, as they describe your rights and restrictions with respect 44 to this document. Code Components extracted from this document must 45 include Simplified BSD License text as described in Section 4.e of 46 the Trust Legal Provisions and are provided without warranty as 47 described in the Simplified BSD License. 49 Table of Contents 51 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 52 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 53 3. Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 54 3.1. RTP Log Format . . . . . . . . . . . . . . . . . . . . . 5 55 4. Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . 5 56 4.1. Avoiding Congestion Collapse . . . . . . . . . . . . . . 5 57 4.2. Stability . . . . . . . . . . . . . . . . . . . . . . . . 5 58 4.3. Media Traffic . . . . . . . . . . . . . . . . . . . . . . 6 59 4.4. Start-up Behaviour . . . . . . . . . . . . . . . . . . . 6 60 4.5. Diverse Environments . . . . . . . . . . . . . . . . . . 6 61 4.6. Varying Path Characteristics . . . . . . . . . . . . . . 7 62 4.7. Reacting to Transient Events or Interruptions . . . . . . 7 63 4.8. Fairness With Similar Cross-Traffic . . . . . . . . . . . 7 64 4.9. Impact on Cross-Traffic . . . . . . . . . . . . . . . . . 7 65 4.10. Extensions to RTP/RTCP . . . . . . . . . . . . . . . . . 8 66 5. Minimum Requirements for Evaluation . . . . . . . . . . . . . 8 67 6. Evaluation Parameters . . . . . . . . . . . . . . . . . . . . 8 68 6.1. Bottleneck Traffic Flows . . . . . . . . . . . . . . . . 8 69 6.2. Access Links . . . . . . . . . . . . . . . . . . . . . . 9 70 6.3. Example Bottleneck Link Parameters . . . . . . . . . . . 9 71 6.4. DropTail Router Queue Parameters . . . . . . . . . . . . 10 72 6.5. Media Flow Parameters . . . . . . . . . . . . . . . . . . 11 73 6.6. Cross-traffic Parameters . . . . . . . . . . . . . . . . 11 74 7. Status of Proposals . . . . . . . . . . . . . . . . . . . . . 11 75 8. Security Considerations . . . . . . . . . . . . . . . . . . . 12 76 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 12 77 10. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 12 78 11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 12 79 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 12 80 12.1. Normative References . . . . . . . . . . . . . . . . . . 12 81 12.2. Informative References . . . . . . . . . . . . . . . . . 13 82 Appendix A. Application Trade-off . . . . . . . . . . . . . . . 14 83 A.1. Measuring Quality . . . . . . . . . . . . . . . . . . . . 14 84 Appendix B. Proposal to evaluate Self-fairness of RMCAT 85 congestion control algorithm . . . . . . . . . . . . 14 86 B.1. Evaluation Parameters . . . . . . . . . . . . . . . . . . 15 87 B.1.1. Media Traffic Generator . . . . . . . . . . . . . . . 15 88 B.1.2. Bottleneck Link Bandwidth . . . . . . . . . . . . . . 16 89 B.1.3. Bottleneck Link Queue Type and Length . . . . . . . . 16 90 B.1.4. RMCAT flows and delay legs . . . . . . . . . . . . . 16 91 B.1.5. Impairment Generator . . . . . . . . . . . . . . . . 17 92 B.2. Proposed Passing Criteria . . . . . . . . . . . . . . . . 17 93 B.3. Extensibility of the Experiment . . . . . . . . . . . . . 17 94 Appendix C. Change Log . . . . . . . . . . . . . . . . . . . . . 18 95 C.1. Changes in draft-ietf-rmcat-eval-criteria-00 . . . . . . 18 96 C.2. Changes in draft-singh-rmcat-cc-eval-04 . . . . . . . . . 18 97 C.3. Changes in draft-singh-rmcat-cc-eval-03 . . . . . . . . . 18 98 C.4. Changes in draft-singh-rmcat-cc-eval-02 . . . . . . . . . 19 99 C.5. Changes in draft-singh-rmcat-cc-eval-01 . . . . . . . . . 19 100 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 19 102 1. Introduction 104 This memo describes the guidelines to help with evaluating new 105 congestion control algorithms for interactive point-to-point real 106 time media. The requirements for the congestion control algorithm 107 are outlined in [I-D.ietf-rmcat-cc-requirements]). This document 108 builds upon previous work at the IETF: Specifying New Congestion 109 Control Algorithms [RFC5033] and Metrics for the Evaluation of 110 Congestion Control Algorithms [RFC5166]. 112 The guidelines proposed in the document are intended to help prevent 113 a congestion collapse, promote fair capacity usage and optimize the 114 media flow's throughput. Furthermore, the proposed algorithms are 115 expected to operate within the envelope of the circuit breakers 116 defined in [I-D.ietf-avtcore-rtp-circuit-breakers]. 118 This document only provides broad-level criteria for evaluating a new 119 congestion control algorithm and the working group should expect a 120 thorough scientific study to make its decision. The results of the 121 evaluation are not expected to be included within the internet-draft 122 but should be cited in the document. 124 2. Terminology 126 The terminology defined in RTP [RFC3550], RTP Profile for Audio and 127 Video Conferences with Minimal Control [RFC3551], RTCP Extended 128 Report (XR) [RFC3611], Extended RTP Profile for RTCP-based Feedback 129 (RTP/AVPF) [RFC4585] and Support for Reduced-Size RTCP [RFC5506] 130 apply. 132 3. Metrics 134 [RFC5166] describes the basic metrics for congestion control. 135 Metrics that are of interest for interactive multimedia are: 137 o Throughput. 139 o Minimizing oscillations in the transmission rate (stability) when 140 the end-to-end capacity varies slowly. 142 o Delay. 144 o Reactivity to transient events. 146 o Packet losses and discards. 148 o Section 2.1 of [RFC5166] discusses the tradeoff between 149 throughput, delay and loss. 151 Each experiment is expected to log every incoming and outgoing packet 152 (the RTP logging format is described in Section 3.1). The logging 153 can be done inside the application or at the endpoints using pcap 154 (packet capture, e.g., tcpdump, wireshark). The following are 155 calculated based on the information in the packet logs: 157 1. Sending rate, Receiver rate, Goodput 159 2. Packet delay 161 3. Packet loss 163 4. If using, retransmission or FEC: residual loss 165 5. Packets discarded from the playout or de-jitter buffer 167 [Open issue (1): The "unfairness" test is (measured at 1s intervals): 168 1. Do not trigger the circuit breaker. 169 2. Over 3 times or less than 1/3 times the throughput for an RMCAT 170 media stream compared to identical RMCAT streams competing on a 171 bottleneck, for a case when the competing streams have similar RTTs. 172 3. Over 3 times delay compared to RTT measurements performed before 173 starting the RMCAT flow or for the case when competing with identical 174 RMCAT streams having similar RTTs. 175 ] 177 [Open issue (2): Possibly using Jain-fairness index.] 179 Convergence time: the time taken to reach a stable rate at startup, 180 after the available link capacity changes, or when new flows get 181 added to the bottleneck link. 183 Bandwidth Utilization, defined as ratio of the instantaneous sending 184 rate to the instantaneous bottleneck capacity. This metric is useful 185 when an RMCAT flow is by itself or competing with similar cross- 186 traffic. 188 From the logs the statistical measures (min, max, mean, standard 189 deviation and variance) for the whole duration or any specific part 190 of the session can be calculated. Also the metrics (sending rate, 191 receiver rate, goodput, latency) can be visualized in graphs as 192 variation over time, the measurements in the plot are at 1 second 193 intervals. Additionally, from the logs it is possible to plot the 194 histogram or CDF of packet delay. 196 3.1. RTP Log Format 198 The log file is tab or comma separated containing the following 199 details: 201 Send or receive timestamp (unix) 202 RTP payload type 203 SSRC 204 RTP sequence no 205 RTP timestamp 206 marker bit 207 payload size 209 If the congestion control implements, retransmissions or FEC, the 210 evaluation should report both packet loss (before applying error- 211 resilience) and residual packet loss (after applying error- 212 resilience). 214 4. Guidelines 216 A congestion control algorithm should be tested in simulation or a 217 testbed environment, and the experiments should be repeated multiple 218 times to infer statistical significance. The following guidelines 219 are considered for evaluation: 221 4.1. Avoiding Congestion Collapse 223 The congestion control algorithm is expected to take an action, such 224 as reducing the sending rate, when it detects congestion. Typically, 225 it should intervene before the circuit breaker 226 [I-D.ietf-avtcore-rtp-circuit-breakers] is engaged. 228 Does the congestion control propose any changes to (or diverge from) 229 the circuit breaker conditions defined in 230 [I-D.ietf-avtcore-rtp-circuit-breakers]. 232 4.2. Stability 233 The congestion control should be assessed for its stability when the 234 path characteristics do not change over time. Changing the media 235 encoding rate estimate too often or by too much may adversely affect 236 the application layer performance. 238 4.3. Media Traffic 240 The congestion control algorithm should be assessed with different 241 types of media behavior, i.e., the media should contain idle and 242 data-limited periods. For example, periods of silence for audio, 243 varying amount of motion for video, or bursty nature of I-frames. 245 The evaluation may be done in two stages. In the first stage, the 246 endpoint generates traffic at the rate calculated by the congestion 247 controller. In the second stage, real codecs or models of video 248 codecs are used to mimic application-limited data periods and varying 249 video frame sizes. 251 4.4. Start-up Behaviour 253 The congestion control algorithm should be assessed with different 254 start-rates. The main reason is to observe the behavior of the 255 congestion control in different evaluation scenarios, such as when 256 competing with varying amount of cross-traffic or how quickly does 257 the congestion control algorithm achieve a stable sending rate. 259 [Editor's note: requires a robust definition for unfriendliness and 260 convergence time.] 262 4.5. Diverse Environments 264 The congestion control algorithm should be assessed in heterogeneous 265 environments, containing both wired and wireless paths. Examples of 266 wireless access technologies are: 802.11, GPRS, HSPA, or LTE. One of 267 the main challenges of the wireless environments for the congestion 268 control algorithm is to distinguish between congestion induced loss 269 and transmission (bit-error corruption) loss. Congestion control 270 algorithms may incorrectly identify transmission loss as congestion 271 loss and reduce the media encoding rate by too much, which may cause 272 oscillatory behavior and deteriorate the users' quality of 273 experience. Furthermore, packet loss may induce additional delay in 274 networks with wireless paths due to link-layer retransmissions. 276 4.6. Varying Path Characteristics 278 The congestion control algorithm should be evaluated for a range of 279 path characteristics such as, different end-to-end capacity and 280 latency, varying amount of cross traffic on a bottleneck link and a 281 router's queue length. For the moment, only DropTail queues are 282 used. However, if new Active Queue Management (AQM) schemes become 283 available, the performance of the congestion control algorithm should 284 be again evaluated. 286 In an experiment, if the media only flows in a single direction, the 287 feedback path should also be tested with varying amounts of 288 impairments. 290 The main motivation for the previous and current criteria is to 291 identify situations in which the proposed congestion control is less 292 performant. 294 4.7. Reacting to Transient Events or Interruptions 296 The congestion control algorithm should be able to handle changes in 297 end-to-end capacity and latency. Latency may change due to route 298 updates, link failures, handovers etc. In mobile environment the 299 end-to-end capacity may vary due to the interference, fading, 300 handovers, etc. In wired networks the end-to-end capacity may vary 301 due to changes in resource reservation. 303 4.8. Fairness With Similar Cross-Traffic 305 The congestion control algorithm should be evaluated when competing 306 with other RTP flows using the same or another candidate congestion 307 control algorithm. The proposal should highlight the bottleneck 308 capacity share of each RTP flow. 310 [Editor's note: If we define Unfriendliness then that criteria should 311 be applied here.] 313 4.9. Impact on Cross-Traffic 315 The congestion control algorithm should be evaluated when competing 316 with standard TCP. Short TCP flows may be considered as transient 317 events and the RTP flow may give way to the short TCP flow to 318 complete quickly. However, long-lived TCP flows may starve out the 319 RTP flow depending on router queue length. 321 The proposal should also measure the impact on varied number of 322 cross-traffic sources, i.e., few and many competing flows, or mixing 323 various amounts of TCP and similar cross-traffic. 325 4.10. Extensions to RTP/RTCP 327 The congestion control algorithm should indicate if any protocol 328 extensions are required to implement it and should carefully describe 329 the impact of the extension. 331 5. Minimum Requirements for Evaluation 333 [Editor's Note: If needed, a minimum evaluation criteria can be based 334 on the above guidelines or defined tests/scenarios.] 336 6. Evaluation Parameters 338 An evaluation scenario is created from a list of network, link and 339 flow characteristics. The example parameters discussed in the 340 following subsections are meant to aid in creating evaluation 341 scenarios and do not describe an evaluation scenario. The scenario 342 discussed in Appendix B takes into account all these parameters. 344 6.1. Bottleneck Traffic Flows 346 The network scenario describes the types of flows sharing the common 347 bottleneck with a single RMCAT flow, they are: 349 1. A single RMCAT flow by itself. 351 2. Competing with similar RMCAT flows. These competing flows may 352 use the same algorithm or another candidate RMCAT algorithm. 354 3. Compete with long-lived TCP. 356 4. Compete with bursty TCP. 358 5. Compete with LEDBAT flows. 360 6. Compete with unresponsive interactive media flows (i.e., not only 361 CBR). 363 Figure 1 shows an example evaluation topology, where S1..Sn are 364 traffic sources, these sources are either RMCAT or a mixture of 365 traffic flows listed above. R1..Rn are the corresponding receivers. 366 A and B are routers that can be configured to introduce impairments. 367 Access links are in between the sender/receiver and the router, while 368 the bottleneck link is between the Routers A and B. 370 +---+ Access Access +---+ 371 |S1 |======= \ / =======|R1 | 372 +---+ link \\ // link +---+ 373 \\ // 374 +---+ +-----+ Bottleneck +-----+ +---+ 375 |S2 |=======| A |------------------------------>| B |=======|R2 | 376 +---+ | |<------------------------------| | +---+ 377 +-----+ Link +-----+ 378 (...) // \\ (...) 379 // \\ 380 +---+ // \\ +---+ 381 |Sn |====== / \ ======|Rn | 382 +---+ +---+ 384 Figure 1: Simple Topology 386 [Open Issue: Discuss more complex topologies] 388 6.2. Access Links 390 The media senders and receivers are typically connected to the 391 bottleneck link, common access links are: 393 1. Ethernet (LAN) 395 2. Wireless LAN (WLAN) 397 3. 3G/LTE 399 [Open issue: point to a reference containing parameters or traces to 400 model WLAN and 3G/LTE.] 402 A real-world network typically consists of a mixture of links, the 403 most important aspect is to identify the location of the bottleneck 404 link. The bottleneck link can move from one node to another 405 depending on the amount of cross-traffic or due to the varying link 406 capacity. The design of the experiments should take this into 407 account. In the simplest case the access link may not be the 408 bottleneck link but an intermediate node. 410 6.3. Example Bottleneck Link Parameters 412 The bottleneck link carries multiple flows, these flows may be other 413 RMCAT flows or other types of cross-traffic. The experiments should 414 dimension the bottleneck link based on the number of flows and the 415 expected behavior. For example, if 5 media flows are expected to 416 share the bottleneck link equally, the bottleneck link is set to 5 417 times the desired transmission rate. 419 If the experiment carries only media in one direction, then the 420 upstream (sender to receiver) bottleneck link carries media packets 421 while the downstream (receiver to sender) bottleneck carries the 422 feedback packets. The bottleneck link parameters discussed in this 423 section apply only to a single direction, hence the bottleneck link 424 in the reverse direction can choose the same or have different 425 parameters. 427 The link latency corresponds to the propagation delay of the link, 428 i.e., the time it takes for a packet to traverse the bottleneck link, 429 it does not include queuing delay. In an experiment with several 430 links the experiment should describe if the links add latency or not. 431 It is possible for experiments to have multiple hops with different 432 link latencies. Experiments are expected to verify that the 433 congestion control is able to work in challenging situations, for 434 example over trans-continental and/or satellite links. The 435 experiment should pick link latency values from the following: 437 1. Very low latency: 0-1ms 439 2. Low latency: 50ms 441 3. High latency: 150ms 443 4. Extreme latency: 300ms 445 Similarly, to model lossy links, the experiments can choose one of 446 the following loss rates, the fractional loss is the ratio of packets 447 lost and packets sent. 449 1. no loss: 0% 451 2. 1% 453 3. 5% 455 4. 10% 457 5. 20% 459 These fractional losses can be generated using traces, Gilbert-Elliot 460 model, randomly (uncorrelated) loss. 462 6.4. DropTail Router Queue Parameters 464 The router queue length is measured as the time taken to drain the 465 FIFO queue, they are: 467 1. QoS-aware (or short): 70ms 468 2. Nominal: 500ms 470 3. Buffer-bloated: 2000ms 472 However, the size of the queue is typically measured in bytes or 473 packets and to convert the queue length measured in seconds to queue 474 length in bytes: 476 QueueSize (in bytes) = QueueSize (in sec) x Throughput (in bps)/8 478 6.5. Media Flow Parameters 480 The media sources can be modeled in two ways. In the first, the 481 sources always have data to send, i.e., have no data limited 482 intervals and are able to generate the media rate requested by the 483 RMCAT congestion control algorithm. In the second, the traffic 484 generator models the behavior of a media codec, mainly the burstiness 485 (time-varying data produced by a video GOP). 487 At the beginning of the session, the media sources are configured to 488 start at a given start rate, they are: 490 1. 200 kbps 492 2. 800 kbps 494 3. 1300 kbps 496 4. 4000 kbps 498 6.6. Cross-traffic Parameters 500 Long-lived TCP flows will download data throughout the session and 501 are expected to have infinite amount of data to send or receive. 503 [Open issue: short-lived/bursty TCP cross-traffic parameters are 504 still TBD. 506 7. Status of Proposals 508 Congestion control algorithms are expected to be published as 509 "Experimental" documents until they are shown to be safe to deploy. 510 An algorithm published as a draft should be experimented in 511 simulation, or a controlled environment (testbed) to show its 512 applicability. Every congestion control algorithm should include a 513 note describing the environments in which the algorithm is tested and 514 safe to deploy. It is possible that an algorithm is not recommended 515 for certain environments or perform sub-optimally for the user. 517 [Editor's Note: Should there be a distinction between "Informational" 518 and "Experimental" drafts for congestion control algorithms in RMCAT. 519 [RFC5033] describes Informational proposals as algorithms that are 520 not safe for deployment but are proposals to experiment with in 521 simulation/testbeds. While Experimental algorithms are ones that are 522 deemed safe in some environments but require a more thorough 523 evaluation (from the community).] 525 8. Security Considerations 527 Security issues have not been discussed in this memo. 529 9. IANA Considerations 531 There are no IANA impacts in this memo. 533 10. Contributors 535 The content and concepts within this document are a product of the 536 discussion carried out in the Design Team. 538 Michael Ramalho provided the text for the scenario discussed in 539 Appendix B. 541 11. Acknowledgements 543 Much of this document is derived from previous work on congestion 544 control at the IETF. 546 The authors would like to thank Harald Alvestrand, Luca De Cicco, 547 Wesley Eddy, Lars Eggert, Kevin Gross, Vinayak Hegde, Stefan Holmer, 548 Randell Jesup, Piers O'Hanlon, Colin Perkins, Michael Ramalho, 549 Zaheduzzaman Sarker, Timothy B. Terriberry, Michael Welzl, and Mo 550 Zanaty for providing valuable feedback on earlier versions of this 551 draft. Additionally, also thank the participants of the design team 552 for their comments and discussion related to the evaluation criteria. 554 12. References 556 12.1. Normative References 558 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 559 Jacobson, "RTP: A Transport Protocol for Real-Time 560 Applications", STD 64, RFC 3550, July 2003. 562 [RFC3551] Schulzrinne, H. and S. Casner, "RTP Profile for Audio and 563 Video Conferences with Minimal Control", STD 65, RFC 3551, 564 July 2003. 566 [RFC3611] Friedman, T., Caceres, R., and A. Clark, "RTP Control 567 Protocol Extended Reports (RTCP XR)", RFC 3611, November 568 2003. 570 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 571 "Extended RTP Profile for Real-time Transport Control 572 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 573 2006. 575 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 576 Real-Time Transport Control Protocol (RTCP): Opportunities 577 and Consequences", RFC 5506, April 2009. 579 [I-D.ietf-rmcat-cc-requirements] 580 Jesup, R., "Congestion Control Requirements For RMCAT", 581 draft-ietf-rmcat-cc-requirements-00 (work in progress), 582 July 2013. 584 [I-D.ietf-avtcore-rtp-circuit-breakers] 585 Perkins, C. and V. Singh, "RTP Congestion Control: Circuit 586 Breakers for Unicast Sessions", draft-ietf-avtcore-rtp- 587 circuit-breakers-01 (work in progress), October 2012. 589 12.2. Informative References 591 [RFC5033] Floyd, S. and M. Allman, "Specifying New Congestion 592 Control Algorithms", BCP 133, RFC 5033, August 2007. 594 [RFC5166] Floyd, S., "Metrics for the Evaluation of Congestion 595 Control Mechanisms", RFC 5166, March 2008. 597 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 598 Control", RFC 5681, September 2009. 600 [SA4-EVAL] 601 R1-081955, 3GPP., "LTE Link Level Throughput Data for SA4 602 Evaluation Framework", 3GPP R1-081955, 5 2008. 604 [SA4-LR] S4-050560, 3GPP., "Error Patterns for MBMS Streaming over 605 UTRAN and GERAN", 3GPP S4-050560, 5 2008. 607 [TCP-eval-suite] 608 Lachlan, A., Marcondes, C., Floyd, S., Dunn, L., Guillier, 609 R., Gang, W., Eggert, L., Ha, S., and I. Rhee, "Towards a 610 Common TCP Evaluation Suite", Proc. PFLDnet. 2008, August 611 2008. 613 Appendix A. Application Trade-off 615 Application trade-off is yet to be defined. see RMCAT requirements 616 [I-D.ietf-rmcat-cc-requirements] document. Perhaps each experiment 617 should define the application's expectation or trade-off. 619 A.1. Measuring Quality 621 No quality metric is defined for performance evaluation, it is 622 currently an open issue. However, there is consensus that congestion 623 control algorithm should be able to show that it is useful for 624 interactive video by performing analysis using a real codec and video 625 sequences. 627 Appendix B. Proposal to evaluate Self-fairness of RMCAT congestion 628 control algorithm 630 The goal of the experiment discussed in this section is to initially 631 take out as many unknowns from the scenario. Later experiments can 632 define more complex environments, topologies and media behavior. 633 This experiment evaluates the performance of the RMCAT sender 634 competing with other similar RMCAT flows (running the same algorithm 635 or other RMCAT proposals) on the bottleneck link. There are up to 20 636 RMCAT flows competing for capacity, but the media only flows in one 637 direction, from senders (S1..S20) to receivers (R1..R20) and the 638 feedback packets flow in the reverse direction. 640 Figure 2 shows the experiment setup and it has subtle differences 641 compared to the simple topology in Figure 1. Groups of 10 receivers 642 are connected to the bottleneck link through two different routers 643 (Router C and D). The rationale for adding these additional routers 644 is to create two delay legs, i.e., two groups of endpoints with 645 different network latencies and measure the performance of the RMCAT 646 congestion control algorithm. If fewer than 10 sources are 647 initialized, all traffic flows experience the same delay because they 648 share the same delay leg. 650 Router A has a single forward direction bottleneck link (i.e., the 651 bottleneck capacity and delay constraints applies only to the media 652 packets going from the sender to the receiver, the feedback packets 653 are unaffected). Hence, the Round-Trip Time (RTT) is primarily 654 composed of the bottleneck queue delay and any forward path 655 (propagation) latency. The main reason for not applying any 656 constraints on the return path is to provide the best-case 657 performance scenario for the congestion control algorithm. In later 658 experiments, it is possible to add similar capacity and delay 659 constraints on the return path. 661 +---+ 662 / === |R1 | 663 +---+ +-----+ // +---+ 664 |S1 |======= \ / =| C | // 665 +---+ \\ // +-----+ \\ (...) 666 \\ // \\ 667 +---+ +-----+ Bottleneck +-----+ \\ +---+ 668 |S2 |=======| A |-------------------->| B | \ ===|R10| 669 +---+ | |<--------------------| | +---+ 670 +-----+ Link +-----+ 671 (...) // \\ +---+ 672 // \\ / === |R11| 673 +---+ // \\ +-----+ // +---+ 674 |S20|====== / \ =| D |// 675 +---+ +-----+\\ (...) 676 \\ 677 \\ +---+ 678 \ ===|R20| 679 +---+ 681 Figure 2: Self-fairness Evaluation Setup 683 Loss impairments are applied at Router C and Router D, but only to 684 the feedback flows. If the losses are set to 0%, it represents a 685 case where the return path is over-provisioned for all traffic. In 686 later experiments the loss impairments can be added to the media path 687 as well. 689 The media sources are configured to send infinite amount of data, 690 i.e., the sources always have data to send and have no data limited 691 intervals. Additionally, the media sources are always successful in 692 generating the media rate requested by the RMCAT congestion control 693 algorithm. In this experiment, we avoid the potentially complicated 694 scenario of using media traffic generators that try to model the 695 behavior of media codecs (mainly the burstiness). 697 B.1. Evaluation Parameters 699 B.1.1. Media Traffic Generator 701 The media source always generates at the rate requested by the 702 congestion control and has infinite data to send. Furthermore, the 703 media packet generator is subject to the following constraints: 705 1. It MUST emit a packet at least once per 100 ms time interval. 707 2. For low media rate source: when generating data at a rate less 708 than a maximum length MTU every 100 ms would allow (e.g., 120 709 kbps = 1500 bytes/packet * 10 packets/sec * 8 bits/byte), the 710 RMCAT source must modulate the packet size (RTP payload size) of 711 RTP packets that are sent every 100 ms to attain the desired 712 rate. 714 3. For high media rate sources: when generating data at a rate 715 greater than a maximum length MTU every 100 ms would allow, the 716 source must do so by sending (approximately) maximum MTU sized 717 packets and adjusting the inter-departure interval to be 718 approximately equal. The intent of this to ensure the data is 719 sent relatively smoothly independent of the bit rate, subject to 720 the first constraint. 722 B.1.2. Bottleneck Link Bandwidth 724 The bottleneck link capacity is dimensioned such that each RMCAT flow 725 in an ideal situation with perfectly equal capacity sharing for all 726 the flows on the bottleneck obtains the following throughputs: 200 727 kbps, 800 kbps, 1.3 Mbps and 4 Mbps. 728 For example, experiments with five RMCAT flows with an 800 kbps/flow 729 target rate should set the bottleneck link capacity to 4 Mbps. 731 B.1.3. Bottleneck Link Queue Type and Length 733 The bottleneck link queue (Router A) is a simple FIFO queue having a 734 buffer length corresponding to 70 ms, 500 ms or 2000 ms (defined in 735 Section 6.4) of delay at the bottleneck link rate (i.e., actual 736 buffer lengths in bytes are dependent on bottleneck link bandwidth). 738 B.1.4. RMCAT flows and delay legs 740 Experiments run with 1, 3, 5, 10 and 20 RMCAT sources, they are 741 outlined as follows: 743 1. Experiments with 1, 3, and 5 RMCAT flows, all RMCAT flows 744 commence simultaneously. A single delay leg is used and the link 745 latency is set to one of the following : 0 ms, 50 ms and 150 ms. 747 2. For 10 and 20 source experiments where all RMCAT flows begin 748 simultaneously the sources are split evenly into two different 749 bulk delay legs. One leg is set to 0 ms bulk delay leg and the 750 other is set to 150 ms. 752 3. For 10 and 20 source experiments where the first set will use 0 753 ms of bulk delay and the second set will use 150 ms bulk delay. 755 1. Random starts within interval [0 ms, 500 ms]. 757 2. One "early-coming" flow (i.e., the 1st flow starting and 758 achieving steady-state before the next N-1 simultaneously 759 begin). 761 3. One "late-coming" flow (i.e., the Nth flow starting after 762 steady-state has occurred for the existing N-1 flows). 764 These cases assess if there are any early or late-comer 765 advantages or disadvantages for a particular algorithm and to see 766 if any unfairness is reproducible or unpredictable. 768 [Open issue (A.1): which group does the early and late flow belong 769 to?] 771 [Open issue (A.2): Start rate for the media flows] 773 B.1.5. Impairment Generator 775 Packet loss is created in the reverse path (affects only feedback 776 packets). Cases of 0%, 1%, 5% and 10% are studied for the 1, 3, and 777 5 RMCAT flow experiments, losses are not applied to flows with 10 or 778 20 RMCAT flows. 780 B.2. Proposed Passing Criteria 782 [Editor's note: there has been little or no discussion on the below 783 criteria, however, they are listed here for the sake of completeness. 785 No unfairness is observed, i.e., at steady state each flow attains a 786 throughput between [ B/(3*N), (3*B)/N ], where B is the link 787 bandwidth and N is the number of flows. 789 No flow experiences packet loss when queue length is set to 500 ms or 790 greater. 792 All individual sources must be in their steady state within twenty 793 LRTTs (where LRTT is defined as the RTT associated with the flow with 794 the Largest RTT in the experiment). ] 796 B.3. Extensibility of the Experiment 798 The above scenario describes only RMCAT sources competing for 799 capacity on the bottleneck link, however, future experiments can use 800 different types of cross-traffic (as described in Section 6.1). 802 Currently, the forward path (carrying media packets) is characterized 803 to add delay and a fixed bottleneck link capacity, in the future 804 packet losses and capacity changes can be applied to mimic a wireless 805 link layer (for e.g., WiFi, 3G, LTE). Additionally, only losses are 806 applied to the reverse path (carrying feedback packets), later 807 experiments can apply the same forward path (carrying media packets) 808 impairments to the reverse path. 810 Appendix C. Change Log 812 Note to the RFC-Editor: please remove this section prior to 813 publication as an RFC. 815 C.1. Changes in draft-ietf-rmcat-eval-criteria-00 817 o Updated references. 819 o Resubmitted as WG draft. 821 C.2. Changes in draft-singh-rmcat-cc-eval-04 823 o Incorporate feedback from IETF 87, Berlin. 825 o Clarified metrics: convergence time, bandwidth utilization. 827 o Changed fairness criteria to fairness test. 829 o Added measuring pre- and post-repair loss. 831 o Added open issue of measuring video quality to appendix. 833 o clarified use of DropTail and AQM. 835 o Updated text in "Minimum Requirements for Evaluation" 837 C.3. Changes in draft-singh-rmcat-cc-eval-03 839 o Incorporate the discussion within the design team. 841 o Added a section on evaluation parameters, it describes the flow 842 and network characteristics. 844 o Added Appendix with self-fairness experiment. 846 o Changed bottleneck parameters from a proposal to an example set. 848 C.4. Changes in draft-singh-rmcat-cc-eval-02 850 o Added scenario descriptions. 852 C.5. Changes in draft-singh-rmcat-cc-eval-01 854 o Removed QoE metrics. 856 o Changed stability to steady-state. 858 o Added measuring impact against few and many flows. 860 o Added guideline for idle and data-limited periods. 862 o Added reference to TCP evaluation suite in example evaluation 863 scenarios. 865 Authors' Addresses 867 Varun Singh 868 Aalto University 869 School of Electrical Engineering 870 Otakaari 5 A 871 Espoo, FIN 02150 872 Finland 874 Email: varun@comnet.tkk.fi 875 URI: http://www.netlab.tkk.fi/~varun/ 877 Joerg Ott 878 Aalto University 879 School of Electrical Engineering 880 Otakaari 5 A 881 Espoo, FIN 02150 882 Finland 884 Email: jo@comnet.tkk.fi