idnits 2.17.1 draft-singh-rmcat-cc-eval-02.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 date (February 26, 2013) is 4070 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'S3' is mentioned on line 333, but not defined == Missing Reference: 'S1' is mentioned on line 343, but not defined == Missing Reference: 'S2' is mentioned on line 318, but not defined == Missing Reference: 'See S1' is mentioned on line 347, but not defined == Missing Reference: 'S4' is mentioned on line 361, but not defined == Missing Reference: 'S5' is mentioned on line 366, but not defined == Outdated reference: A later version (-18) exists of draft-ietf-avtcore-rtp-circuit-breakers-01 Summary: 0 errors (**), 0 flaws (~~), 8 warnings (==), 1 comment (--).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 30, 2013 February 26, 2013 7 Evaluating Congestion Control for Interactive Real-time Media 8 draft-singh-rmcat-cc-eval-02.txt 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 30, 2013. 34 Copyright Notice 36 Copyright (c) 2013 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 . . . . . . . . . . . . . . . . . . . . . . . . 2 52 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 53 3. Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 54 4. Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . 4 55 4.1. Avoiding Congestion Collapse . . . . . . . . . . . . . . 4 56 4.2. Stability . . . . . . . . . . . . . . . . . . . . . . . . 4 57 4.3. Media Traffic . . . . . . . . . . . . . . . . . . . . . . 5 58 4.4. Diverse Environments . . . . . . . . . . . . . . . . . . 5 59 4.5. Varying Path Characteristics . . . . . . . . . . . . . . 5 60 4.6. Reacting to Transient Events or Interruptions . . . . . . 5 61 4.7. Fairness With Similar Cross-Traffic . . . . . . . . . . . 6 62 4.8. Impact on Cross-Traffic . . . . . . . . . . . . . . . . . 6 63 4.9. Extensions to RTP/RTCP . . . . . . . . . . . . . . . . . 6 64 5. Minimum Requirements for Evaluation . . . . . . . . . . . . . 6 65 6. Example Evaluation Scenarios . . . . . . . . . . . . . . . . 6 66 6.1. [S1] RTP flow on a fixed link . . . . . . . . . . . . . . 7 67 6.2. [S2] RTP flow on a variable capacity link . . . . . . . . 7 68 6.3. [S3] Fairness to RTP flows running the same congestion 69 control algorithm (self-fairness) . . . . . . . . . . . . 8 70 6.4. [S4 and S5] Competing with short and long TCP flows . . . 8 71 7. Status of Proposals . . . . . . . . . . . . . . . . . . . . . 9 72 8. Security Considerations . . . . . . . . . . . . . . . . . . . 9 73 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 9 74 10. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 9 75 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 9 76 11.1. Normative References . . . . . . . . . . . . . . . . . . 9 77 11.2. Informative References . . . . . . . . . . . . . . . . . 10 78 Appendix A. Change Log . . . . . . . . . . . . . . . . . . . . . 11 79 A.1. Changes in draft-singh-rmcat-cc-eval-02 . . . . . . . . . 11 80 A.2. Changes in draft-singh-rmcat-cc-eval-01 . . . . . . . . . 11 81 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 11 83 1. Introduction 85 This memo describes the guidelines to help with evaluating new 86 congestion control algorithms for interactive point-to-point real 87 time media. The requirements for the congestion control algorithm 88 are outlined in [I-D.jesup-rtp-congestion-reqs]). This document 89 builds upon previous work at the IETF: Specifying New Congestion 90 Control Algorithms [RFC5033] and Metrics for the Evaluation of 91 Congestion Control Algorithms [RFC5166]. 93 The guidelines proposed in the document are intended to prevent a 94 congestion collapse, promote fair capacity usage and optimize the 95 media flow's throughput, and quality. Furthermore, the proposed 96 algorithms are expected to operate within the envelope of the circuit 97 breakers defined in [I-D.ietf-avtcore-rtp-circuit-breakers]. 99 This document only provides broad-level criteria for evaluating a new 100 congestion control algorithm and the working group should expect a 101 thorough scientific study to make its decision. The results of the 102 evaluation are not expected to be included within the internet-draft 103 but should be cited in the document. 105 2. Terminology 107 The terminology defined in RTP [RFC3550], RTP Profile for Audio and 108 Video Conferences with Minimal Control [RFC3551], RTCP Extended 109 Report (XR) [RFC3611], Extended RTP Profile for RTCP-based Feedback 110 (RTP/AVPF) [RFC4585] and Support for Reduced-Size RTCP [RFC5506] 111 apply. 113 3. Metrics 115 [RFC5166] describes the basic metrics for congestion control. 116 Metrics that are important to interactive multimedia are: 118 o Throughput: (Sending Rate, Receiving Rate, Goodput) 120 o Minimizing oscillations in encoding rate (stability) 122 o Reactivity to transient events 124 o Packet loss and discard rate 126 o Users' quality of experience 128 [Editor's Note: measurement interval and statistical measures (min, 129 max, mean, median, standard deviation and variance) are yet to be 130 specified.] 132 Section 2.1 of [RFC5166] discusses the tradeoff between throughput, 133 delay and loss. 135 (i) Bandwidth Utilization: is the ratio of the encoding rate to 136 the (available) end-to-end path capacity. 138 * Under-utilization: is the period of time when the endpoint's 139 encoding rate is lower than the end-to-end capacity, i.e., the 140 bandwidth utilization is less than 1. 142 * Overuse: is the period of time when the endpoint's encoding 143 rate is higher than the end-to-end capacity, i.e., the 144 bandwidth utilization is greater than 1. 146 * Steady-state: is the period of time when the endpoint's 147 encoding rate is relatively stable, i.e., the bandwidth 148 utilization is constant. 150 (ii) Packet Loss and Discard Rate. 152 (iii) Fair Share. 154 [Editor's Note: This metric should match the ones defined in the 155 RMCAT requirements [I-D.jesup-rtp-congestion-reqs] document.] 157 (iv) Quality: There are many different types of quality metrics 158 for audio and video. Audio quality is often expressed by a MOS 159 ("Mean Opinion Score") and can be calculated using an objective 160 algorithm (E-model/R-model). Section 4.7 of [RFC3611] can also be 161 used for VoIP metrics. Similarly, there exist several metrics to 162 measure video quality, for example Peak Signal to Noise Ratio 163 (PSNR). 165 [Editor's Note: Should the algorithm compare average PSNR of test 166 video sequences or what other video quality metric can be used? 167 If Quality is used as a metric, it should not be the only metric 168 used to compare rate-control schemes. Also, algorithms using 169 different codecs cannot be compared]. 171 4. Guidelines 173 A congestion control algorithm should be tested in simulation or a 174 testbed environment, and the experiments should be repeated multiple 175 times to infer statistical significance. The following guidelines 176 are considered for evaluation: 178 4.1. Avoiding Congestion Collapse 180 Does the congestion control propose any changes to (or diverge from) 181 the circuit breaker conditions defined in 182 [I-D.ietf-avtcore-rtp-circuit-breakers]. 184 4.2. Stability 186 The congestion control should be assessed for its stability when the 187 path characteristics do not change over time. Changing the media 188 encoding rate too often or by too much may adversely affect the 189 users' quality of experience. 191 4.3. Media Traffic 193 The congestion control algorithm should be assessed with different 194 types of media behavior, i.e., the media should contain idle and 195 data-limited periods. For example, periods of silence for audio or 196 varying amount of motion for video. 198 4.4. Diverse Environments 200 The congestion control algorithm should be assessed in heterogeneous 201 environments, containing both wired and wireless paths. Examples of 202 wireless access technologies are: 802.11x, GPRS, HSPA, or LTE. One 203 of the main challenges of the wireless environments is the inability 204 to distinguish congestion induced loss from transmission (bit-error) 205 loss. Congestion control algorithms may incorrectly identify 206 transmission loss as congestion loss and reduce the media encoding 207 rate too much, which may cause oscillatory behavior and deteriorate 208 the users' quality of experience. Furthermore, packet loss may 209 induce additional delay in networks with wireless paths due to link- 210 layer retransmissions. 212 4.5. Varying Path Characteristics 214 The congestion control algorithm should be evaluated for a range of 215 path characteristics such as, different end-to-end capacity and 216 latency, varying amount of cross traffic on a bottle-neck link and a 217 router's queue length. The main motivation for the previous and 218 current criteria is to determine under which circumstances will the 219 proposed congestion control algorithm break down and also determine 220 the operational range of the algorithm. 222 [Editor's Note: Different types of queueing mechanisms? Random Early 223 Detection or only DropTail?]. 225 4.6. Reacting to Transient Events or Interruptions 227 The congestion control algorithm should be able to handle changes in 228 end-to-end capacity and latency. Latency may change due to route 229 updates, link failures, handovers etc. In mobile environment the 230 end-to-end capacity may vary due to the interference, fading, 231 handovers, etc. In wired networks the end-to-end capacity may vary 232 due to changes in resource reservation. 234 4.7. Fairness With Similar Cross-Traffic 236 The congestion control algorithm should be evaluated when competing 237 with other RTP flows using the same congestion control algorithm. 238 The proposal should highlight the bottleneck capacity share of each 239 RTP flow. 241 4.8. Impact on Cross-Traffic 243 [Editor's Note: There was discussion about removing this guideline, 244 however, no decision was made [I-D.jesup-rtp-congestion-reqs].] 246 The congestion control algorithm should be evaluated when competing 247 with standard TCP. Short TCP flows may be considered as transient 248 events and the RTP flow may give way to the short TCP flow to 249 complete quickly. However, long-lived TCP flows may starve out the 250 RTP flow depending on router queue length. In the latter case the 251 proposed congestion control for RTP should be as aggressive as 252 standard TCP [RFC5681]. 254 The proposal should also measure the impact on varied number of 255 cross-traffic sources, i.e., few and many competing flows, or mixing 256 various amounts of TCP and similar cross-traffic. 258 4.9. Extensions to RTP/RTCP 260 The congestion control algorithm should indicate if any protocol 261 extensions are required to implement it and should carefully describe 262 the impact of the extension. 264 5. Minimum Requirements for Evaluation 266 [Editor's Note: If needed, a minimum evaluation criteria can be based 267 on the above guidelines] 269 6. Example Evaluation Scenarios 271 In the scenarios listed below, all RTP flows are bi-directional and 272 point-to-point. 274 Unless specified, the following parameters are used in each scenario: 276 o Video Start Rate: 128 kbps 278 o Maximum end-to-end delay: 300ms, packets arriving after this are 279 discarded 281 o Video Frame rate: 15 282 o Audio packetization interval: 20ms 284 o MTU: 1450 bytes 286 o [Editor's Note: the numbers in this section are TBD] 288 Topology: 290 o Dumbbell, the endpoint is connected to the bottleneck link via an 291 access links. The bottleneck may be shared by multiple endpoints. 293 o Parking lot: there are three bottleneck links arranged 294 horizontally, these links are connected by access links. In this 295 case, flows may share different bottleneck links. 297 [Editor's note: Should the queue-size be specified as well?]. 299 6.1. [S1] RTP flow on a fixed link 301 This scenario evaluates the ramp-up to the bottleneck capacity and 302 the stability of the proposed congestion control algorithm. 304 This scenario uses the dumbbell topology and both the access link can 305 be ADSL (500kbps uplink, 256 downlink, 2ms one-way delay) or WLAN 306 (54Mbps, 2ms one-way delay, 2-5% packet loss rate and link layer re- 307 transmissions). 309 The bottleneck link can have one of the following capacities: 310 500kbps, 1Mbps, 5Mbps and link delay: 10ms, 50ms, 120ms. 312 Each congestion control algorithm should plot the variation of the 313 sending rate against time, also plot the instances of packets losses. 314 Additionally, measure the time taken for the sending rate to reach 315 the end-to-end capacity (average and standard deviation over 10 316 simulation runs). 318 6.2. [S2] RTP flow on a variable capacity link 320 This scenario evaluates the reactivity of the proposed congestion 321 control algorithm to transient network events due to interference and 322 handovers in mobile environments. 324 This scenario uses the dumbbell topology, and both end-points use 3G/ 325 LTE access. Sample 3G/LTE (uplink and downlink) bandwidth traces are 326 available at [SA4-EVAL], loss patterns at [SA4-LR] and the link 327 delay: 30ms, 80ms. The bottleneck link can have one of the following 328 capacities: 500kbps, 5Mbps and link delay: 20ms. 330 Each congestion control algorithm should plot the variation of the 331 sending rate against time, also plot the instances of packets losses. 333 6.3. [S3] Fairness to RTP flows running the same congestion control 334 algorithm (self-fairness) 336 This scenario shows if the proposed algorithm can share the 337 bottleneck link equitably, irrespective of number of flows. 339 In this scenario there is more than one endpoint connected to the 340 bottleneck link. 342 (a) All the access links have the same link characteristics and 343 start at the same time (see [S1]). The bottleneck link can have 344 one of the following link capacity: 500kbpsm 5Mbpps and link delay 345 20ms. 347 (b) The access links have different link characteristics [See S1] 348 but start at the same time. 350 (c) An RTP flow is added at 10s intervals (upto 5 flows), the late 351 arriving flows have increasing access link delay (0, 5, 10, 20, 352 50ms). The bottleneck link can have one of the following 353 capacities: 1Mbps, 10Mbps and link delay: 10ms, 50ms, 120ms. 355 [Parking lot topology simulation: TBD] 357 6.4. [S4 and S5] Competing with short and long TCP flows 359 [Editor's Note: Remove these scenarios?] 361 [S4] Competing with long-lived TCP flows: In this scenario the 362 proposed algorithm is expected to be TCP-friendly, i.e., it should 363 neither starve out the competing TCP flows (causing a congestion 364 collapse) nor should it be starved out by TCP. 366 [S5] Competing with short TCP flows: Depending on the level of 367 statistical multiplexing on the bottleneck link, the proposed 368 algorithm may behave differently. If there are a few short TCP flows 369 then the proposed algorithm may observe these flows as transient 370 events and let them complete quickly. Alternatively, if there are 371 many short flows then the proposed algorithm may have to compete with 372 the flows as if they were long lived TCP flows. 374 [TCP-eval-suite] contains examples of TCP traffic load and scenario 375 settings. 377 [Editor's Note: definition of many and few short TCP flows may depend 378 on the bottleneck link capacity.] 380 [Editor's Note: clarify if media packets are generated using a 381 traffic generator.] 383 7. Status of Proposals 385 Congestion control algorithms are expected to be published as 386 "Experimental" documents until they are shown to be safe to deploy. 387 An algorithm published as a draft should be experimented in 388 simulation, or a controlled environment (testbed) to show its 389 applicability. Every congestion control algorithm should include a 390 note describing the environments in which the algorithm is tested and 391 safe to deploy. It is possible that an algorithm is not recommended 392 for certain environments or perform sub-optimally for the user. 394 [Editor's Note: Should there be a distinction between "Informational" 395 and "Experimental" drafts for congestion control algorithms in RMCAT. 396 [RFC5033] describes Informational proposals as algorithms that are 397 not safe for deployment but are proposals to experiment with in 398 simulation/testbeds. While Experimental algorithms are ones that are 399 deemed safe in some environments but require a more thorough 400 evaluation (from the community).] 402 8. Security Considerations 404 Security issues have not been discussed in this memo. 406 9. IANA Considerations 408 There are no IANA impacts in this memo. 410 10. Acknowledgements 412 Much of this document is derived from previous work on congestion 413 control at the IETF. 415 The authors would like to thank Harald Alvestrand, Luca De Cicco, 416 Wesley Eddy, Lars Eggert, Vinayak Hegde, Stefan Holmer, Randell 417 Jesup, Piers O'Hanlon, Colin Perkins, Timothy B. Terriberry, Michael 418 Welzl, and Sarker Zaheduzzaman for providing valuable feedback on 419 earlier versions of this draft. 421 11. References 423 11.1. Normative References 425 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 426 Jacobson, "RTP: A Transport Protocol for Real-Time 427 Applications", STD 64, RFC 3550, July 2003. 429 [RFC3551] Schulzrinne, H. and S. Casner, "RTP Profile for Audio and 430 Video Conferences with Minimal Control", STD 65, RFC 3551, 431 July 2003. 433 [RFC3611] Friedman, T., Caceres, R., and A. Clark, "RTP Control 434 Protocol Extended Reports (RTCP XR)", RFC 3611, November 435 2003. 437 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 438 "Extended RTP Profile for Real-time Transport Control 439 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 440 2006. 442 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 443 Real-Time Transport Control Protocol (RTCP): Opportunities 444 and Consequences", RFC 5506, April 2009. 446 [I-D.jesup-rtp-congestion-reqs] 447 Jesup, R. and H. Alvestrand, "Congestion Control 448 Requirements For Real Time Media", draft-jesup-rtp- 449 congestion-reqs-00 (work in progress), March 2012. 451 [I-D.ietf-avtcore-rtp-circuit-breakers] 452 Perkins, C. and V. Singh, "RTP Congestion Control: Circuit 453 Breakers for Unicast Sessions", draft-ietf-avtcore-rtp- 454 circuit-breakers-01 (work in progress), October 2012. 456 11.2. Informative References 458 [RFC5033] Floyd, S. and M. Allman, "Specifying New Congestion 459 Control Algorithms", BCP 133, RFC 5033, August 2007. 461 [RFC5166] Floyd, S., "Metrics for the Evaluation of Congestion 462 Control Mechanisms", RFC 5166, March 2008. 464 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 465 Control", RFC 5681, September 2009. 467 [SA4-EVAL] 468 R1-081955, 3GPP., "LTE Link Level Throughput Data for SA4 469 Evaluation Framework", 3GPP R1-081955, 5 2008. 471 [SA4-LR] S4-050560, 3GPP., "Error Patterns for MBMS Streaming over 472 UTRAN and GERAN", 3GPP S4-050560, 5 2008. 474 [TCP-eval-suite] 475 Lachlan, A., Marcondes, C., Floyd, S., Dunn, L., Guillier, 476 R., Gang, W., Eggert, L., Ha, S., and I. Rhee, "Towards a 477 Common TCP Evaluation Suite", Proc. PFLDnet. 2008, August 478 2008. 480 Appendix A. Change Log 482 Note to the RFC-Editor: please remove this section prior to 483 publication as an RFC. 485 A.1. Changes in draft-singh-rmcat-cc-eval-02 487 o Added scenario descriptions. 489 A.2. Changes in draft-singh-rmcat-cc-eval-01 491 o Removed QoE metrics. 493 o Changed stability to steady-state. 495 o Added measuring impact against few and many flows. 497 o Added guideline for idle and data-limited periods. 499 o Added reference to TCP evaluation suite in example evaluation 500 scenarios. 502 Authors' Addresses 504 Varun Singh 505 Aalto University 506 School of Electrical Engineering 507 Otakaari 5 A 508 Espoo, FIN 02150 509 Finland 511 Email: varun@comnet.tkk.fi 512 URI: http://www.netlab.tkk.fi/~varun/ 513 Joerg Ott 514 Aalto University 515 School of Electrical Engineering 516 Otakaari 5 A 517 Espoo, FIN 02150 518 Finland 520 Email: jo@comnet.tkk.fi