idnits 2.17.1 draft-singh-rmcat-cc-eval-03.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 680: '... 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 (July 15, 2013) is 3938 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'RFC5681' is defined on line 594, but no explicit reference was found in the text == Unused Reference: 'SA4-EVAL' is defined on line 597, but no explicit reference was found in the text == Unused Reference: 'SA4-LR' is defined on line 601, but no explicit reference was found in the text == Unused Reference: 'TCP-eval-suite' is defined on line 604, but no explicit reference was found in the text == Outdated reference: A later version (-18) exists of draft-ietf-avtcore-rtp-circuit-breakers-01 Summary: 1 error (**), 0 flaws (~~), 6 warnings (==), 1 comment (--).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: January 16, 2014 July 15, 2013 7 Evaluating Congestion Control for Interactive Real-time Media 8 draft-singh-rmcat-cc-eval-03.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 January 16, 2014. 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 . . . . . . . . . . . . . . . . . . . . . . . . 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 . . . . . . . . . . . . . . . . . . . . . . 5 59 4.4. Start-up Behaviour . . . . . . . . . . . . . . . . . . . 6 60 4.5. Diverse Environments . . . . . . . . . . . . . . . . . . 6 61 4.6. Varying Path Characteristics . . . . . . . . . . . . . . 6 62 4.7. Reacting to Transient Events or Interruptions . . . . . . 6 63 4.8. Fairness With Similar Cross-Traffic . . . . . . . . . . . 7 64 4.9. Impact on Cross-Traffic . . . . . . . . . . . . . . . . . 7 65 4.10. Extensions to RTP/RTCP . . . . . . . . . . . . . . . . . 7 66 5. Minimum Requirements for Evaluation . . . . . . . . . . . . . 7 67 6. Evaluation Parameters . . . . . . . . . . . . . . . . . . . . 7 68 6.1. Bottleneck Traffic Flows . . . . . . . . . . . . . . . . 8 69 6.2. Access Links . . . . . . . . . . . . . . . . . . . . . . 8 70 6.3. Bottleneck Link Parameters . . . . . . . . . . . . . . . 9 71 6.4. Router Queue Parameters . . . . . . . . . . . . . . . . . 10 72 6.5. Media Flow Parameters . . . . . . . . . . . . . . . . . . 10 73 6.6. Cross-traffic Parameters . . . . . . . . . . . . . . . . 11 74 7. Status of Proposals . . . . . . . . . . . . . . . . . . . . . 11 75 8. Security Considerations . . . . . . . . . . . . . . . . . . . 11 76 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 11 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. Proposal to evaluate Self-fairness of RMCAT 83 congestion control algorithm . . . . . . . . . . . . 13 84 A.1. Evaluation Parameters . . . . . . . . . . . . . . . . . . 15 85 A.1.1. Media Traffic Generator . . . . . . . . . . . . . . . 15 86 A.1.2. Bottleneck Link Bandwidth . . . . . . . . . . . . . . 15 87 A.1.3. Bottleneck Link Queue Type and Length . . . . . . . . 15 88 A.1.4. RMCAT flows and delay legs . . . . . . . . . . . . . 15 89 A.1.5. Impairment Generator . . . . . . . . . . . . . . . . 16 90 A.2. Proposed Passing Criteria . . . . . . . . . . . . . . . . 16 91 A.3. Extensability of the Experiment . . . . . . . . . . . . . 17 92 Appendix B. Change Log . . . . . . . . . . . . . . . . . . . . . 17 93 B.1. Changes in draft-singh-rmcat-cc-eval-03 . . . . . . . . . 17 94 B.2. Changes in draft-singh-rmcat-cc-eval-02 . . . . . . . . . 17 95 B.3. Changes in draft-singh-rmcat-cc-eval-01 . . . . . . . . . 17 96 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 18 98 1. Introduction 100 This memo describes the guidelines to help with evaluating new 101 congestion control algorithms for interactive point-to-point real 102 time media. The requirements for the congestion control algorithm 103 are outlined in [I-D.jesup-rmcat-reqs]). This document builds upon 104 previous work at the IETF: Specifying New Congestion Control 105 Algorithms [RFC5033] and Metrics for the Evaluation of Congestion 106 Control Algorithms [RFC5166]. 108 The guidelines proposed in the document are intended to help prevent 109 a congestion collapse, promote fair capacity usage and optimize the 110 media flow's throughput. Furthermore, the proposed algorithms are 111 expected to operate within the envelope of the circuit breakers 112 defined in [I-D.ietf-avtcore-rtp-circuit-breakers]. 114 This document only provides broad-level criteria for evaluating a new 115 congestion control algorithm and the working group should expect a 116 thorough scientific study to make its decision. The results of the 117 evaluation are not expected to be included within the internet-draft 118 but should be cited in the document. 120 2. Terminology 122 The terminology defined in RTP [RFC3550], RTP Profile for Audio and 123 Video Conferences with Minimal Control [RFC3551], RTCP Extended 124 Report (XR) [RFC3611], Extended RTP Profile for RTCP-based Feedback 125 (RTP/AVPF) [RFC4585] and Support for Reduced-Size RTCP [RFC5506] 126 apply. 128 3. Metrics 130 [RFC5166] describes the basic metrics for congestion control. 131 Metrics that are important to interactive multimedia are: 133 o Throughput. 135 o Minimizing oscillations in the transmission rate (stability) when 136 the end-to-end capacity varies slowly. 138 o Delay. 140 o Reactivity to transient events. 142 o Packet losses and discards. 144 Each experiment logs every incoming and outgoing packet (the RTP 145 logging format is described in Section 3.1). The logging can be done 146 inside the application or at the endpoints using pcap (packet 147 capture, e.g., tcpdump, wireshark). The following are calculated 148 based on the information in the packet logs: 150 1. Sending rate, Receiver rate, Goodput 152 2. Packet delay 154 3. Packet loss 156 4. Packets discarded from the playout or de-jitter buffer 158 [Editor's note: How to handle packet re-transmissions? loss before 159 retransmission, after retransmission?] 161 [Open issue (1): Instead of defining fairness, there has been 162 discussion on defining "unfairness". The criteria are: 163 1. Do not trigger the circuit breaker. 164 2. Over 3 times or less than 1/3 times the throughput for an RMCAT 165 media stream compared to identical RMCAT streams competing on a 166 bottleneck, for a case when the competing streams have similar RTTs. 167 3. Over 3 times delay compared to RTT measurements performed before 168 starting the RMCAT flow or for the case when competing with identical 169 RMCAT streams having similar RTTs. 170 Here, rather than discussing the number '3'? Does the criteria 171 capture Unfairness adequately?] 173 [Open issue (2): Convergence time was discussed briefly in the design 174 meetings. It is defined as: the time it takes the congestion control 175 to reach a stable rate (at startup or after new RMCAT flows are 176 added). What is a stable rate?] 178 [Open issue (3): previous versions of the document had Bandwidth 179 Utilization, defined as ratio of sending rate to the available 180 bottleneck capacity. This is useful when the RMCAT flow is by itself 181 or competing with similar flows (where the assumption would be that 182 all flows get an equal share). Remove this?] 184 From the logs the statistical measures (min, max, mean, standard 185 deviation and variance) for the whole duration or any specific part 186 of the session can be calculated. Also the metrics (sending rate, 187 receiver rate, goodput, latency) can be visualized in graphs as 188 variation over time, the measurements in the plot are at 1 second 189 intervals. Additionally, from the logs it is possible to plot the 190 histogram or CDF of packet delay. 192 Section 2.1 of [RFC5166] discusses the tradeoff between throughput, 193 delay and loss. 195 [Open issue (4): Application trade-off is yet to be defined. see 196 RMCAT requirements [I-D.jesup-rmcat-reqs] document. Perhaps each 197 experiment should define the application's expectation or trade-off.] 199 3.1. RTP Log Format 201 The log file is tab or comma separated containing the following 202 details: 204 Send or receive timestamp (unix) 205 RTP payload type 206 SSRC 207 RTP sequence no 208 RTP timestamp 209 marker bit 210 payload size 212 [Open issue (5): Should the retransmissions for post-repair loss 213 metric be logged in a separate file? the repair streams have 214 different payload type and/or SSRC.] 216 4. Guidelines 218 A congestion control algorithm should be tested in simulation or a 219 testbed environment, and the experiments should be repeated multiple 220 times to infer statistical significance. The following guidelines 221 are considered for evaluation: 223 4.1. Avoiding Congestion Collapse 225 Does the congestion control propose any changes to (or diverge from) 226 the circuit breaker conditions defined in 227 [I-D.ietf-avtcore-rtp-circuit-breakers]. 229 4.2. Stability 231 The congestion control should be assessed for its stability when the 232 path characteristics do not change over time. Changing the media 233 encoding rate estimate too often or by too much may adversely affect 234 the application layer performance. 236 4.3. Media Traffic 238 The congestion control algorithm should be assessed with different 239 types of media behavior, i.e., the media should contain idle and 240 data-limited periods. For example, periods of silence for audio or 241 varying amount of motion for video. However, the evaluation can be 242 done in two stages. In the first stage, media stream can generate 243 traffic at the rate calculated by the congestion controller. In the 244 second stage, real codecs or models of video codecs should be used to 245 mimic real-world cases. 247 4.4. Start-up Behaviour 249 The congestion control algorithm should be assessed with different 250 start-rates. The main reason is to observe the behavior of the 251 congestion control in different evaluation scenarios, such as when 252 competing with varying amount of cross-traffic or how quickly does 253 the congestion control algorithm achieve a stable sending rate. 255 [Editor's note: requires a robust definition for unfriendliness and 256 convergence time.] 258 4.5. Diverse Environments 260 The congestion control algorithm should be assessed in heterogeneous 261 environments, containing both wired and wireless paths. Examples of 262 wireless access technologies are: 802.11, GPRS, HSPA, or LTE. One of 263 the main challenges of the wireless environments for the congestion 264 control algorithm is to distinguish between congestion induced loss 265 and transmission (bit-error corruption) loss. Congestion control 266 algorithms may incorrectly identify transmission loss as congestion 267 loss and reduce the media encoding rate by too much, which may cause 268 oscillatory behavior and deteriorate the users' quality of 269 experience. Furthermore, packet loss may induce additional delay in 270 networks with wireless paths due to link-layer retransmissions. 272 4.6. Varying Path Characteristics 274 The congestion control algorithm should be evaluated for a range of 275 path characteristics such as, different end-to-end capacity and 276 latency, varying amount of cross traffic on a bottle-neck link and a 277 router's queue length. In an experiment, if the media only flows in 278 a single direction, the feedback path should also be tested with 279 varying amounts of impairments. 281 The main motivation for the previous and current criteria is to 282 identify situations in which the proposed congestion control is less 283 performant. 285 [Open issue (6): Different types of queueing mechanisms? Random 286 Early Detection or only DropTail?]. 288 4.7. Reacting to Transient Events or Interruptions 289 The congestion control algorithm should be able to handle changes in 290 end-to-end capacity and latency. Latency may change due to route 291 updates, link failures, handovers etc. In mobile environment the 292 end-to-end capacity may vary due to the interference, fading, 293 handovers, etc. In wired networks the end-to-end capacity may vary 294 due to changes in resource reservation. 296 4.8. Fairness With Similar Cross-Traffic 298 The congestion control algorithm should be evaluated when competing 299 with other RTP flows using the same or another candidate congestion 300 control algorithm. The proposal should highlight the bottleneck 301 capacity share of each RTP flow. 303 [Editor's note: If we define Unfriendliness then that criteria should 304 be applied here.] 306 4.9. Impact on Cross-Traffic 308 The congestion control algorithm should be evaluated when competing 309 with standard TCP. Short TCP flows may be considered as transient 310 events and the RTP flow may give way to the short TCP flow to 311 complete quickly. However, long-lived TCP flows may starve out the 312 RTP flow depending on router queue length. 314 The proposal should also measure the impact on varied number of 315 cross-traffic sources, i.e., few and many competing flows, or mixing 316 various amounts of TCP and similar cross-traffic. 318 4.10. Extensions to RTP/RTCP 320 The congestion control algorithm should indicate if any protocol 321 extensions are required to implement it and should carefully describe 322 the impact of the extension. 324 5. Minimum Requirements for Evaluation 326 [Editor's Note: If needed, a minimum evaluation criteria can be based 327 on the above guidelines] 329 6. Evaluation Parameters 331 An evaluation scenario is created from a list of network, link and 332 flow characteristics. The parameters discussed in the following 333 subsections are meant to aid in creating evaluation scenarios and do 334 not describe an evaluation scenario. The scenario discussed in 335 Appendix A takes into account all these parameters. 337 6.1. Bottleneck Traffic Flows 339 The network scenario describes the types of flows sharing the common 340 bottleneck with a single RMCAT flow, they are: 342 1. A single RMCAT flow by itself. 344 2. Competing with similar RMCAT flows. These competing flows may 345 use the same algorithm or another candidate RMCAT algorithm. 347 3. Compete with long-lived TCP. 349 4. Compete with bursty TCP. 351 5. Compete with LEDBAT flows. 353 6. Compete with unresponsive interactive media flows (i.e., not only 354 CBR). 356 Figure 1 shows an example evaluation topology, where S1..Sn are 357 traffic sources, these sources are either RMCAT or a mixture of 358 traffic flows listed above. R1..Rn are the corresponding receivers. 359 A and B are routers that can be configured to introduce impairments. 360 Access links are in between the sender/receiver and the router, while 361 the bottleneck link is between the Routers A and B. 363 +---+ Access Access +---+ 364 |S1 |======= \ / =======|R1 | 365 +---+ link \\ // link +---+ 366 \\ // 367 +---+ +-----+ Bottleneck +-----+ +---+ 368 |S2 |=======| A |------------------------------>| B |=======|R2 | 369 +---+ | |<------------------------------| | +---+ 370 +-----+ Link +-----+ 371 (...) // \\ (...) 372 // \\ 373 +---+ // \\ +---+ 374 |Sn |====== / \ ======|Rn | 375 +---+ +---+ 377 Figure 1: Simple Topology 379 [Open Issue (7): Discuss more complex topologies] 381 6.2. Access Links 383 The media senders and receivers are typically connected to the 384 bottleneck link, common access links are: 386 1. Ethernet (LAN) 388 2. Wireless LAN (WLAN) 390 3. 3G/LTE 392 [Open issue (8): need to describe parameters or traces to model WLAN 393 and 3G/LTE.] 395 A real-world network typically consists of a mixture of links, the 396 most important aspect is to identify the location of the bottleneck 397 link. The bottleneck link can move from one node to another 398 depending on the amount of cross-traffic or due to the varying link 399 capacity. The design of the experiments should take this into 400 account. In the simplest case the access link may not be the 401 bottleneck link but an intermediate node. 403 6.3. Bottleneck Link Parameters 405 The bottleneck link carries multiple flows, these flows may be other 406 RMCAT flows or other types of cross-traffic. The experiments should 407 dimension the bottleneck link based on the number of flows and the 408 expected behavior. For example, if 5 media flows are expected to 409 share the bottleneck link equally, the bottleneck link is set to 5 410 times the desired transmission rate. 412 If the experiment carries only media in one direction, then the 413 upstream (sender to receiver) bottleneck link carries media packets 414 while the downstream (receiver to sender) bottleneck carries the 415 feedback packets. The bottleneck link parameters discussed in this 416 section apply only to a single direction, hence the bottleneck link 417 in the reverse direction can choose the same or have different 418 parameters. 420 The link latency corresponds to the propagation delay of the link, 421 i.e., the time it takes for a packet to traverse the bottleneck link, 422 it does not include queuing delay. In an experiment with several 423 links the experiment should describe if the links add latency or not. 424 It is possible for experiments to have multiple hops with different 425 link latencies. Experiments are expected to verify that the 426 congestion control is able to work in challenging situations, for 427 example over trans-continental and/or satellite links. The 428 experiment should pick link latency values from the following: 430 1. Very low latency: 0-1ms 432 2. Low latency: 50ms 433 3. High latency: 150ms 435 4. Extreme latency: 300ms 437 [Editor's note: currently describes the latency for a single link, 438 instead of end-to-end delay. Which is preferred? or both?] 440 Similarly, to model lossy links, the experiments can choose one of 441 the following loss rates, the fractional loss is the ratio of packets 442 lost and packets sent. 444 1. no loss: 0% 446 2. 1% 448 3. 5% 450 4. 10% 452 5. 20% 454 [Open issue (10): how is the loss generated? using traces, Gilbert- 455 Elliot model, randomly (uncorrelated) loss.] 457 6.4. Router Queue Parameters 459 The router queue length is measured as the time taken to drain the 460 FIFO queue, they are: 462 1. QoS-aware (or short): 70ms 464 2. Nominal: 500ms 466 3. Buffer-bloated: 2000ms 468 However, the size of the queue is typically measured in bytes or 469 packets and to convert the queue length measured in seconds to queue 470 length in bytes: 472 QueueSize (in bytes) = QueueSize (in sec) x Throughput (in bps)/8 474 [Open issue (11): Confirm the above values, do we need to define 475 parameters for other types of queues?] 477 6.5. Media Flow Parameters 479 The media sources can be modeled in two ways. In the first, the 480 sources always have data to send, i.e., have no data limited 481 intervals and are able to generate the media rate requested by the 482 RMCAT congestion control algorithm. In the second, the traffic 483 generator models the behavior of a media codec, mainly the burstiness 484 (time-varying data produced by a video GOP). 486 At the beginning of the session, the media sources are configured to 487 start at a given start rate, they are: 489 1. 200 kbps 491 2. 800 kbps 493 3. 1300 kbps 495 4. 4000 kbps 497 6.6. Cross-traffic Parameters 499 [Open issue(12): TCP cross-traffic parameters are still TBD, mainly 500 the bursty TCP. Long-lived TCP flows will download data throughout 501 the session and are expected to have infinite amount of data to send 502 or receive.] 504 7. Status of Proposals 506 Congestion control algorithms are expected to be published as 507 "Experimental" documents until they are shown to be safe to deploy. 508 An algorithm published as a draft should be experimented in 509 simulation, or a controlled environment (testbed) to show its 510 applicability. Every congestion control algorithm should include a 511 note describing the environments in which the algorithm is tested and 512 safe to deploy. It is possible that an algorithm is not recommended 513 for certain environments or perform sub-optimally for the user. 515 [Editor's Note: Should there be a distinction between "Informational" 516 and "Experimental" drafts for congestion control algorithms in RMCAT. 517 [RFC5033] describes Informational proposals as algorithms that are 518 not safe for deployment but are proposals to experiment with in 519 simulation/testbeds. While Experimental algorithms are ones that are 520 deemed safe in some environments but require a more thorough 521 evaluation (from the community).] 523 8. Security Considerations 525 Security issues have not been discussed in this memo. 527 9. IANA Considerations 528 There are no IANA impacts in this memo. 530 10. Contributors 532 The content and concepts within this document are a product of the 533 discussion carried out in the Design Team. 535 Michael Ramalho provided the text for the scenario discussed in 536 Appendix A. 538 11. Acknowledgements 540 Much of this document is derived from previous work on congestion 541 control at the IETF. 543 The authors would like to thank Harald Alvestrand, Luca De Cicco, 544 Wesley Eddy, Lars Eggert, Kevin Gross, Vinayak Hegde, Stefan Holmer, 545 Randell Jesup, Piers O'Hanlon, Colin Perkins, Michael Ramalho, 546 Zaheduzzaman Sarker, Timothy B. Terriberry, Michael Welzl, and Mo 547 Zanaty for providing valuable feedback on earlier versions of this 548 draft. Additionally, also thank the participants of the design team 549 for their comments and discussion related to the evaluation criteria. 551 12. References 553 12.1. Normative References 555 [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. 556 Jacobson, "RTP: A Transport Protocol for Real-Time 557 Applications", STD 64, RFC 3550, July 2003. 559 [RFC3551] Schulzrinne, H. and S. Casner, "RTP Profile for Audio and 560 Video Conferences with Minimal Control", STD 65, RFC 3551, 561 July 2003. 563 [RFC3611] Friedman, T., Caceres, R., and A. Clark, "RTP Control 564 Protocol Extended Reports (RTCP XR)", RFC 3611, November 565 2003. 567 [RFC4585] Ott, J., Wenger, S., Sato, N., Burmeister, C., and J. Rey, 568 "Extended RTP Profile for Real-time Transport Control 569 Protocol (RTCP)-Based Feedback (RTP/AVPF)", RFC 4585, July 570 2006. 572 [RFC5506] Johansson, I. and M. Westerlund, "Support for Reduced-Size 573 Real-Time Transport Control Protocol (RTCP): Opportunities 574 and Consequences", RFC 5506, April 2009. 576 [I-D.jesup-rmcat-reqs] 577 Jesup, R., "Congestion Control Requirements For RMCAT", 578 draft-jesup-rmcat-reqs-01 (work in progress), February 579 2013. 581 [I-D.ietf-avtcore-rtp-circuit-breakers] 582 Perkins, C. and V. Singh, "RTP Congestion Control: Circuit 583 Breakers for Unicast Sessions", draft-ietf-avtcore-rtp- 584 circuit-breakers-01 (work in progress), October 2012. 586 12.2. Informative References 588 [RFC5033] Floyd, S. and M. Allman, "Specifying New Congestion 589 Control Algorithms", BCP 133, RFC 5033, August 2007. 591 [RFC5166] Floyd, S., "Metrics for the Evaluation of Congestion 592 Control Mechanisms", RFC 5166, March 2008. 594 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 595 Control", RFC 5681, September 2009. 597 [SA4-EVAL] 598 R1-081955, 3GPP., "LTE Link Level Throughput Data for SA4 599 Evaluation Framework", 3GPP R1-081955, 5 2008. 601 [SA4-LR] S4-050560, 3GPP., "Error Patterns for MBMS Streaming over 602 UTRAN and GERAN", 3GPP S4-050560, 5 2008. 604 [TCP-eval-suite] 605 Lachlan, A., Marcondes, C., Floyd, S., Dunn, L., Guillier, 606 R., Gang, W., Eggert, L., Ha, S., and I. Rhee, "Towards a 607 Common TCP Evaluation Suite", Proc. PFLDnet. 2008, August 608 2008. 610 Appendix A. Proposal to evaluate Self-fairness of RMCAT congestion 611 control algorithm 613 The goal of the experiment discussed in this section is to initially 614 take out as many unknowns from the scenario. Later experiments can 615 define more complex environments, topologies and media behavior. 616 This experiment evaluates the performance of the RMCAT sender 617 competing with other similar RMCAT flows (running the same algorithm 618 or other RMCAT proposals) on the bottleneck link. There are up to 20 619 RMCAT flows competing for capacity, but the media only flows in one 620 direction, from senders (S1..S20) to receivers (R1..R20) and the 621 feedback packets flow in the reverse direction. 623 Figure 2 shows the experiment setup and it has subtle differences 624 compared to the simple topology in Figure 1. Groups of 10 receivers 625 are connected to the bottleneck link through two different routers 626 (Router C and D). The rationale for adding these additional routers 627 is to create two delay legs, i.e., two groups of endpoints with 628 different network latencies and measure the performance of the RMCAT 629 congestion control algorithm. If fewer than 10 sources are 630 initialized, all traffic flows experience the same delay because they 631 share the same delay leg. 633 Router A has a single forward direction bottleneck link (i.e., the 634 bottleneck capacity and delay constraints applies only to the media 635 packets going from the sender to the receiver, the feedback packets 636 are unaffected). Hence, the Round-Trip Time (RTT) is primarily 637 composed of the bottleneck queue delay and any forward path 638 (propagation) latency. The main reason for not applying any 639 constraints on the return path is to provide the best-case 640 performance scenario for the congestion control algorithm. In later 641 experiments, it is possible to add similar capacity and delay 642 constraints on the return path. 644 +---+ 645 / === |R1 | 646 +---+ +-----+ // +---+ 647 |S1 |======= \ / =| C | // 648 +---+ \\ // +-----+ \\ (...) 649 \\ // \\ 650 +---+ +-----+ Bottleneck +-----+ \\ +---+ 651 |S2 |=======| A |-------------------->| B | \ ===|R10| 652 +---+ | |<--------------------| | +---+ 653 +-----+ Link +-----+ 654 (...) // \\ +---+ 655 // \\ / === |R11| 656 +---+ // \\ +-----+ // +---+ 657 |S20|====== / \ =| D |// 658 +---+ +-----+\\ (...) 659 \\ 660 \\ +---+ 661 \ ===|R20| 662 +---+ 664 Figure 2: Self-fairness Evaluation Setup 666 Loss impairments are applied at Router C and Router D, but only to 667 the feedback flows. If the losses are set to 0%, it represents a 668 case where the return path is over-provisioned for all traffic. In 669 later experiments the loss impairments can be added to the media path 670 as well. 672 A.1. Evaluation Parameters 674 A.1.1. Media Traffic Generator 676 The media source always generates at the rate requested by the 677 congestion control and has infinite data to send. Furthermore, the 678 media packet generator is subject to the following constraints: 680 1. It MUST emit a packet at least once per 100 ms time interval. 682 2. For low media rate source: when generating data at a rate less 683 than a maximum length MTU every 100 ms would allow (e.g., 120 684 kbps = 1500 bytes/packet * 10 packets/sec * 8 bits/byte), the 685 RMCAT source must modulate the packet size (RTP payload size) of 686 RTP packets that are sent every 100 ms to attain the desired 687 rate. 689 3. For high media rate sources: when generating data at a rate 690 greater than a maximum length MTU every 100 ms would allow, the 691 source must do so by sending (approximately) maximum MTU sized 692 packets and adjusting the inter-departure interval to be 693 approximately equal. The intent of this to ensure the data is 694 sent relatively smoothly independent of the bit rate, subject to 695 the first constraint. 697 A.1.2. Bottleneck Link Bandwidth 699 The bottleneck link capacity is dimensioned such that each RMCAT flow 700 in an ideal situation with perfectly equal capacity sharing for all 701 the flows on the bottleneck obtains the following throughputs: 200 702 kbps, 800 kbps, 1.3 Mbps and 4 Mbps. 703 For example, experiments with five RMCAT flows with an 800 kbps/flow 704 target rate should set the bottleneck link capacity to 4 Mbps. 706 A.1.3. Bottleneck Link Queue Type and Length 708 The bottleneck link queue (Router A) is a simple FIFO queue having a 709 buffer length corresponding to 70 ms, 500 ms or 2000 ms (defined in 710 Section 6.4) of delay at the bottleneck link rate (i.e., actual 711 buffer lengths in bytes are dependent on bottleneck link bandwidth). 713 A.1.4. RMCAT flows and delay legs 715 Experiments run with 1, 3, 5, 10 and 20 RMCAT sources, they are 716 outlined as follows: 718 1. Experiments with 1, 3, and 5 RMCAT flows, all RMCAT flows 719 commence simultaneously. A single delay leg is used and the link 720 latency is set to one of the following : 0 ms, 50 ms and 150 ms. 722 2. For 10 and 20 source experiments where all RMCAT flows begin 723 simultaneously the sources are split evenly into two different 724 bulk delay legs. One leg is set to 0 ms bulk delay leg and the 725 other is set to 150 ms. 727 3. For 10 and 20 source experiments where the first set will use 0 728 ms of bulk delay and the second set will use 150 ms bulk delay. 730 1. Random starts within interval [0 ms, 500 ms]. 732 2. One "early-coming" flow (i.e., the 1st flow starting and 733 achieving steady-state before the next N-1 simultaneously 734 begin). 736 3. One "late-coming" flow (i.e., the Nth flow starting after 737 steady-state has occurred for the existing N-1 flows). 739 These cases assess if there are any early or late-comer 740 advantages or disadvantages for a particular algorithm and to see 741 if any unfairness is reproducible or unpredictable. 743 [Open issue (A.1): which group does the early and late flow belong 744 to?] 746 [Open issue (A.2): Start rate for the media flows] 748 A.1.5. Impairment Generator 750 Packet loss is created in the reverse path (affects only feedback 751 packets). Cases of 0%, 1%, 5% and 10% are studied for the 1, 3, and 752 5 RMCAT flow experiments, losses are not applied to flows with 10 or 753 20 RMCAT flows. 755 A.2. Proposed Passing Criteria 757 [Editor's note: there has been little or no discussion on the below 758 criteria, however, they are listed here for the sake of completeness. 760 No unfairness is observed, i.e., at steady state each flow attains a 761 throughput between [ B/(3*N), (3*B)/N ], where B is the link 762 bandwidth and N is the number of flows. 764 No flow experiences packet loss when queue length is set to 500 ms or 765 greater. 767 All individual sources must be in their steady state within twenty 768 LRTTs (where LRTT is defined as the RTT associated with the flow with 769 the Largest RTT in the experiment). ] 771 A.3. Extensability of the Experiment 773 The above scenario describes only RMCAT sources competing for 774 capacity on the bottleneck link, however, future experiments can use 775 different types of cross-traffic (as described in Section 6.1). 777 Currently, the forward path (carrying media packets) is characterized 778 to add delay and a fixed bottleneck link capacity, in the future 779 packet losses and capacity changes can be applied to mimic a wireless 780 link layer (for e.g., WiFi, 3G, LTE). Additionally, only losses are 781 applied to the reverse path (carrying feedback packets), later 782 experiments can apply the same forward path (carrying media packets) 783 impairments to the reverse path. 785 Appendix B. Change Log 787 Note to the RFC-Editor: please remove this section prior to 788 publication as an RFC. 790 B.1. Changes in draft-singh-rmcat-cc-eval-03 792 o Incorporate the discussion within the design team. 794 o Added a section on evaluation parameters, it describes the flow 795 and network characteristics. 797 o Added Appendix with self-fairness experiment. 799 B.2. Changes in draft-singh-rmcat-cc-eval-02 801 o Added scenario descriptions. 803 B.3. Changes in draft-singh-rmcat-cc-eval-01 805 o Removed QoE metrics. 807 o Changed stability to steady-state. 809 o Added measuring impact against few and many flows. 811 o Added guideline for idle and data-limited periods. 813 o Added reference to TCP evaluation suite in example evaluation 814 scenarios. 816 Authors' Addresses 818 Varun Singh 819 Aalto University 820 School of Electrical Engineering 821 Otakaari 5 A 822 Espoo, FIN 02150 823 Finland 825 Email: varun@comnet.tkk.fi 826 URI: http://www.netlab.tkk.fi/~varun/ 828 Joerg Ott 829 Aalto University 830 School of Electrical Engineering 831 Otakaari 5 A 832 Espoo, FIN 02150 833 Finland 835 Email: jo@comnet.tkk.fi