idnits 2.17.1 draft-ietf-tcpm-alternativebackoff-ecn-10.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 (August 31, 2018) is 2062 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Unused Reference: 'RFC7713' is defined on line 537, but no explicit reference was found in the text == Outdated reference: A later version (-28) exists of draft-ietf-tcpm-accurate-ecn-06 -- Obsolete informational reference (is this intentional?): RFC 8312 (Obsoleted by RFC 9438) Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 2 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group N. Khademi 3 Internet-Draft M. Welzl 4 Intended status: Experimental University of Oslo 5 Expires: March 4, 2019 G. Armitage 6 Netflix 7 G. Fairhurst 8 University of Aberdeen 9 August 31, 2018 11 TCP Alternative Backoff with ECN (ABE) 12 draft-ietf-tcpm-alternativebackoff-ecn-10 14 Abstract 16 Active Queue Management (AQM) mechanisms allow for burst tolerance 17 while enforcing short queues to minimise the time that packets spend 18 enqueued at a bottleneck. This can cause noticeable performance 19 degradation for TCP connections traversing such a bottleneck, 20 especially if there are only a few flows or their bandwidth-delay- 21 product is large. The reception of a Congestion Experienced (CE) ECN 22 mark indicates that an AQM mechanism is used at the bottleneck, and 23 therefore the bottleneck network queue is likely to be short. 24 Feedback of this signal allows the TCP sender-side ECN reaction in 25 congestion avoidance to reduce the Congestion Window (cwnd) by a 26 smaller amount than the congestion control algorithm's reaction to 27 inferred packet loss. This specification therefore defines an 28 experimental change to the TCP reaction specified in RFC3168, as 29 permitted by RFC 8311. 31 Status of This Memo 33 This Internet-Draft is submitted in full conformance with the 34 provisions of BCP 78 and BCP 79. 36 Internet-Drafts are working documents of the Internet Engineering 37 Task Force (IETF). Note that other groups may also distribute 38 working documents as Internet-Drafts. The list of current Internet- 39 Drafts is at https://datatracker.ietf.org/drafts/current/. 41 Internet-Drafts are draft documents valid for a maximum of six months 42 and may be updated, replaced, or obsoleted by other documents at any 43 time. It is inappropriate to use Internet-Drafts as reference 44 material or to cite them other than as "work in progress." 46 This Internet-Draft will expire on March 4, 2019. 48 Copyright Notice 50 Copyright (c) 2018 IETF Trust and the persons identified as the 51 document authors. All rights reserved. 53 This document is subject to BCP 78 and the IETF Trust's Legal 54 Provisions Relating to IETF Documents 55 (https://trustee.ietf.org/license-info) in effect on the date of 56 publication of this document. Please review these documents 57 carefully, as they describe your rights and restrictions with respect 58 to this document. Code Components extracted from this document must 59 include Simplified BSD License text as described in Section 4.e of 60 the Trust Legal Provisions and are provided without warranty as 61 described in the Simplified BSD License. 63 Table of Contents 65 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 66 2. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . 3 67 3. Specification . . . . . . . . . . . . . . . . . . . . . . . . 4 68 3.1. Choice of ABE Multiplier . . . . . . . . . . . . . . . . 4 69 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . 6 70 4.1. Why Use ECN to Vary the Degree of Backoff? . . . . . . . 6 71 4.2. An RTT-based response to indicated congestion . . . . . . 7 72 5. ABE Deployment Requirements . . . . . . . . . . . . . . . . . 7 73 6. ABE Experiment Goals . . . . . . . . . . . . . . . . . . . . 8 74 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 8 75 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 9 76 9. Implementation Status . . . . . . . . . . . . . . . . . . . . 9 77 10. Security Considerations . . . . . . . . . . . . . . . . . . . 9 78 11. Revision Information . . . . . . . . . . . . . . . . . . . . 9 79 12. References . . . . . . . . . . . . . . . . . . . . . . . . . 11 80 12.1. Normative References . . . . . . . . . . . . . . . . . . 11 81 12.2. Informative References . . . . . . . . . . . . . . . . . 11 82 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 13 84 1. Introduction 86 Explicit Congestion Notification (ECN) [RFC3168] makes it possible 87 for an Active Queue Management (AQM) mechanism to signal the presence 88 of incipient congestion without necessarily incurring packet loss. 89 This lets the network deliver some packets to an application that 90 would have been dropped if the application or transport did not 91 support ECN. This packet loss reduction is the most obvious benefit 92 of ECN, but it is often relatively modest. Other benefits of 93 deploying ECN have been documented in RFC8087 [RFC8087]. 95 The rules for ECN were originally written to be very conservative, 96 and required the congestion control algorithms of ECN-Capable 97 transport protocols to treat indications of congestion signalled by 98 ECN exactly the same as they would treat an inferred packet loss 99 [RFC3168]. Research has demonstrated the benefits of reducing 100 network delays that are caused by interaction of loss-based TCP 101 congestion control and excessive buffering [BUFFERBLOAT]. This has 102 led to the creation of AQM mechanisms like PIE [RFC8033] and CoDel 103 [CODEL2012][RFC8289], which prevent bloated queues that are common 104 with unmanaged and excessively large buffers deployed across the 105 Internet [BUFFERBLOAT]. 107 The AQM mechanisms mentioned above aim to keep a sustained queue 108 short while tolerating transient (short-term) packet bursts. 109 However, currently used loss-based congestion control mechanisms are 110 not always able to effectively utilise a bottleneck link where there 111 are short queues. For example, a TCP sender using the Reno 112 congestion control needs to be able to store at least an end-to-end 113 bandwidth-delay product (BDP) worth of data at the bottleneck buffer 114 if it is to maintain full path utilisation in the face of loss- 115 induced reduction of the congestion window (cwnd) [RFC5681], which 116 effectively doubles the amount of data that can be in flight, the 117 maximum round-trip time (RTT) experience, and the path's effective 118 RTT using the network path. 120 Modern AQM mechanisms can use ECN to signal the early signs of 121 impending queue buildup long before a tail-drop queue would be forced 122 to resort to dropping packets. It is therefore appropriate for the 123 transport protocol congestion control algorithm to have a more 124 measured response when it receives an indication with an early- 125 warning of congestion after the remote endpoint receives an ECN CE- 126 marked packet. Recognizing these changes in modern AQM practices, 127 the strict requirement that ECN CE signals be treated identically to 128 inferred packet loss have been relaxed [RFC8311]. This document 129 therefore defines a new sender-side-only congestion control response, 130 called "ABE" (Alternative Backoff with ECN). ABE improves TCP's 131 average throughput when routers use AQM controlled buffers that allow 132 only for short queues. 134 2. Definitions 136 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 137 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 138 document are to be interpreted as described in RFC 2119 139 [RFC2119][RFC8174] . 141 3. Specification 143 This specification changes the congestion control algorithm of an 144 ECN-Capable TCP transport protocol by changing the TCP sender 145 response to feedback from the TCP receiver that indicates reception 146 of a CE-marked packet, i.e., receipt of a packet with the ECN-Echo 147 flag (defined in [RFC3168]) set, following the process defined in 148 [RFC8311]. 150 The TCP sender response is currently specified in section 6.1.2 of 151 the ECN specification [RFC3168], updated by [RFC8311]: 153 The indication of congestion should be treated just as a 154 congestion loss in non-ECN-Capable TCP. That is, the TCP source 155 halves the congestion window "cwnd" and reduces the slow start 156 threshold "ssthresh", unless otherwise specified by an 157 Experimental RFC in the IETF document stream. 159 Following publication of RFC 8311, this document specifies a sender- 160 side change to TCP: 162 Receipt of a packet with the ECN-Echo flag SHOULD trigger the TCP 163 source to set the slow start threshold (ssthresh) to 0.8 times the 164 FlightSize, with a lower bound of 2 * SMSS applied to the result. 165 As in [RFC5681], the TCP sender also reduces the cwnd value to no 166 more than the new ssthresh value. RFC 3168 section 6.1.2 provides 167 guidance on setting a cwnd less than 2 * SMSS. 169 3.1. Choice of ABE Multiplier 171 ABE decouples the reaction of a TCP sender to inferred packet loss 172 and indication of ECN-signalled congestion in the congestion 173 avoidance phase. To achieve this, ABE uses a different scaling 174 factor in Equation 4 in Section 3.1 of [RFC5681]. The description 175 respectively uses beta_{loss} and beta_{ecn} to refer to the 176 multiplicative decrease factors applied in response to inferred 177 packet loss, and in response to a receiver indicating ECN-signalled 178 congestion. For non-ECN-enabled TCP connections, only beta_{loss} 179 applies. 181 In other words, in response to inferred packet loss: 183 ssthresh = max (FlightSize * beta_{loss}, 2 * SMSS) 185 and in response to an indication of an ECN-signalled congestion: 187 ssthresh = max (FlightSize * beta_{ecn}, 2 * SMSS) 188 and 190 cwnd = ssthresh 192 (If ssthresh == 2 * SMSS, RFC 3168 section 6.1.2 provides guidance 193 on setting a cwnd lower than 2 * SMSS.) 195 where FlightSize is the amount of outstanding data in the network, 196 upper-bounded by the smaller of the sender's cwnd and the receiver's 197 advertised window (rwnd) [RFC5681]. The higher the values of 198 beta_{loss} and beta_{ecn}, the less aggressive the response of any 199 individual backoff event. 201 The appropriate choice for beta_{loss} and beta_{ecn} values is a 202 balancing act between path utilisation and draining the bottleneck 203 queue. More aggressive backoff (smaller beta_*) risks underutilising 204 the path, while less aggressive backoff (larger beta_*) can result in 205 slower draining of the bottleneck queue. 207 The Internet has already been running with at least two different 208 beta_{loss} values for several years: the standard value is 0.5 209 [RFC5681], and the Linux implementation of CUBIC [RFC8312] has used a 210 multiplier of 0.7 since kernel version 2.6.25 released in 2008. ABE 211 does not change the value of beta_{loss} used by current TCP 212 implementations. 214 The recommendation in this document specifies a value of 215 beta_{ecn}=0.8. This recommended beta_{ecn} value is only applicable 216 for the standard TCP congestion control [RFC5681]. The selection of 217 beta_{ecn} enables tuning the response of a TCP connection to shallow 218 AQM marking thresholds. beta_{loss} characterizes the response of a 219 congestion control algorithm to packet loss, i.e., exhaustion of 220 buffers (of unknown depth). Different values for beta_{loss} have 221 been suggested for TCP congestion control algorithms. Consequently, 222 beta_{ecn} is likely to be an algorithm-specific parameter rather 223 than a constant multiple of the algorithm's existing beta_{loss}. 225 A range of tests (section IV, [ABE2017]) with NewReno and CUBIC over 226 CoDel and PIE in lightly-multiplexed scenarios have explored this 227 choice of parameter. The results of these tests indicate that CUBIC 228 connections benefit from beta_{ecn} of 0.85 (cf. beta_{loss} = 0.7), 229 and NewReno connections see improvements with beta_{ecn} in the range 230 0.7 to 0.85 (cf. beta_{loss} = 0.5). 232 4. Discussion 234 Much of the technical background to ABE can be found in a research 235 paper [ABE2017]. This paper used a mix of experiments, theory and 236 simulations with NewReno [RFC5681] and CUBIC [RFC8312] to evaluate 237 the technique. The technique was shown to present "...significant 238 performance gains in lightly-multiplexed [few concurrent flows] 239 scenarios, without losing the delay-reduction benefits of deploying 240 CoDel or PIE". The performance improvement is achieved when reacting 241 to ECN-Echo in congestion avoidance (when ssthresh > cwnd) by 242 multiplying cwnd and ssthresh with a value in the range [0.7,0.85]. 243 Applying ABE when cwnd <= ssthresh is not currently recommended, but 244 may benefit from additional attention, experimentation and 245 specification. 247 4.1. Why Use ECN to Vary the Degree of Backoff? 249 AQM mechanisms such as CoDel [RFC8289] and PIE [RFC8033] set a delay 250 target in routers and use congestion notifications to constrain the 251 queuing delays experienced by packets, rather than in response to 252 impending or actual bottleneck buffer exhaustion. With current 253 default delay targets, CoDel and PIE both effectively emulate a 254 bottleneck with a short queue (section II, [ABE2017]) while also 255 allowing short traffic bursts into the queue. This provides 256 acceptable performance for TCP connections over a path with a low 257 BDP, or in highly multiplexed scenarios (many concurrent transport 258 flows). However, in a lightly-multiplexed case over a path with a 259 large BDP, conventional TCP backoff leads to gaps in packet 260 transmission and under-utilisation of the path. 262 Instead of discarding packets, an AQM mechanism is allowed to mark 263 ECN-Capable packets with an ECN CE-mark. The reception of a CE-mark 264 feedback not only indicates congestion on the network path, it also 265 indicates that an AQM mechanism exists at the bottleneck along the 266 path, and hence the CE-mark likely came from a bottleneck with a 267 controlled short queue. Reacting differently to an ECN-signalled 268 congestion than to an inferred packet loss can then yield the benefit 269 of a reduced back-off when queues are short. Using ECN can also be 270 advantageous for several other reasons [RFC8087]. 272 The idea of reacting differently to inferred packet loss and 273 detection of an ECN-signalled congestion pre-dates this 274 specification. For example, previous research proposed using ECN CE- 275 marked feedback to modify TCP congestion control behaviour via a 276 larger multiplicative decrease factor in conjunction with a smaller 277 additive increase factor [ICC2002]. The goal of this former work was 278 to operate across AQM bottlenecks using Random Early Detection (RED) 279 that were not necessarily configured to emulate a short queue (The 280 current usage of RED as an Internet AQM method is limited [RFC7567]). 282 4.2. An RTT-based response to indicated congestion 284 This specification applies to the use of ECN feedback as defined in 285 [RFC3168], which specifies a response to indicated congestion that is 286 no more frequent that once per path round trip time. Since ABE 287 responds to indicated congestion once per RTT, it therefore does not 288 respond to any further loss within the same RTT, because an ABE 289 sender has already reduced the congestion window. If congestion 290 persists after such reduction, ABE continues to reduce the congestion 291 window in each consecutive RTT. This consecutive reduction can 292 protect the network against long-standing unfairness in the case of 293 AQM algorithms that do not keep a small average queue length. The 294 mechanism does not rely on Accurate ECN 295 ([I-D.ietf-tcpm-accurate-ecn]). 297 In contrast, transport protocol mechanisms can also be designed to 298 utilise more frequent and detailed ECN feedback (e.g., Accurate ECN 299 [I-D.ietf-tcpm-accurate-ecn]), which then permit a congestion control 300 response that adjusts the sending rate more frequently. Datacenter 301 TCP (DCTCP) [RFC8257] is an example of this approach. 303 5. ABE Deployment Requirements 305 This update is a sender-side only change. Like other changes to 306 congestion control algorithms, it does not require any change to the 307 TCP receiver or to network devices. It does not require any ABE- 308 specific changes in routers or the use of Accurate ECN feedback 309 [I-D.ietf-tcpm-accurate-ecn] by a receiver. 311 If the method is only deployed by some senders, and not by others, 312 the senders that use this method can gain some advantage, possibly at 313 the expense of other flows that do not use this updated method. 314 Because this advantage applies only to ECN-marked packets and not to 315 packet loss indications, an ECN-Capable bottleneck will still fall 316 back to dropping packets if an TCP sender using ABE is too 317 aggressive, and the result is no different than if the TCP sender was 318 using traditional loss-based congestion control. 320 When used with bottlenecks that do not support ECN-marking the 321 specification does not modify the transport protocol. 323 6. ABE Experiment Goals 325 [RFC3168] states that the congestion control response following an 326 indication of ECN-signalled congestion is the same as the response to 327 a dropped packet. [RFC8311] updates this specification to allow 328 systems to provide a different behaviour when they experience ECN- 329 signalled congestion rather than packet loss. The present 330 specification defines such an experiment and has thus been assigned 331 an Experimental status before being proposed as a Standards-Track 332 update. 334 The purpose of the Internet experiment is to collect experience with 335 deployment of ABE, and confirm acceptable safety in deployed networks 336 that use this update to TCP congestion control. To evaluate ABE, 337 this experiment therefore requires support in AQM routers for ECN- 338 marking of packets carrying the ECN-Capable Transport, ECT(0), 339 codepoint [RFC3168]. 341 The result of this Internet experiment ought to include an 342 investigation of the implications of experiencing an ECN-CE mark 343 followed by loss within the same RTT. At the end of the experiment, 344 this will be reported to the TCPM WG or the IESG. 346 7. Acknowledgements 348 Authors N. Khademi, M. Welzl and G. Fairhurst were part-funded by 349 the European Community under its Seventh Framework Programme through 350 the Reducing Internet Transport Latency (RITE) project (ICT-317700). 351 The views expressed are solely those of the authors. 353 Author G. Armitage performed most of his work on this document while 354 employed by Swinburne University of Technology, Melbourne, Australia. 356 The authors would like to thank Stuart Cheshire for many suggestions 357 when revising the draft, and the following people for their 358 contributions to [ABE2017]: Chamil Kulatunga, David Ros, Stein 359 Gjessing, Sebastian Zander. Thanks also to (in alphabetical order) 360 Roland Bless, Bob Briscoe, David Black, Markku Kojo, John Leslie, 361 Lawrence Stewart, Dave Taht and the TCPM Working Group for providing 362 valuable feedback on this document. 364 The authors would finally like to thank everyone who provided 365 feedback on the congestion control behaviour specified in this update 366 received from the IRTF Internet Congestion Control Research Group 367 (ICCRG). 369 8. IANA Considerations 371 XX RFC ED - PLEASE REMOVE THIS SECTION XXX 373 This document includes no request to IANA. 375 9. Implementation Status 377 ABE is implemented as a patch for Linux and FreeBSD. This is meant 378 for research and available for download from 379 http://heim.ifi.uio.no/michawe/research/abe/. This code was used to 380 produce the test results that are reported in [ABE2017]. The FreeBSD 381 code has been committed to the mainline kernel on March 19, 2018 382 [ABE-FreeBSD]. 384 10. Security Considerations 386 The described method is a sender-side only transport change, and does 387 not change the protocol messages exchanged. The security 388 considerations for ECN [RFC3168] therefore still apply. 390 This is a change to TCP congestion control with ECN that will 391 typically lead to a change in the capacity achieved when flows share 392 a network bottleneck. This could result in some flows receiving more 393 than their fair share of capacity. Similar unfairness in the way 394 that capacity is shared is also exhibited by other congestion control 395 mechanisms that have been in use in the Internet for many years 396 (e.g., CUBIC [RFC8312]). Unfairness may also be a result of other 397 factors, including the round trip time experienced by a flow. ABE 398 applies only when ECN-marked packets are received, not when packets 399 are lost, hence use of ABE cannot lead to congestion collapse. 401 11. Revision Information 403 XX RFC ED - PLEASE REMOVE THIS SECTION XXX 405 -10. Incorported changes following the Gen-ART review by Russ 406 Housley. Correction to URL. 408 -09. Changed to "Following publication of RFC 8311, this document 409 specifies a sender-side change to TCP:" 411 -08. Addressed comments from AD review on the document structure, 412 and relationship to existing RFCs. 414 -07. Addressed comments following WGLC. 416 o Updated Reference citations. 418 o Removed paragraph containing a wrong statement related to timeout 419 in section 4.1. 421 o Discuss what happens when cwnd <= ssthresh. 423 o Added text on Concern about lower bound of 2*SMSS. 425 -06. Addressed Michael Scharf's comments. 427 -05. Refined the description of the experiment based on feedback at 428 IETF-100. Incorporated comments from David Black. 430 -04. Incorporates review comments from Lawrence Stewart and the 431 remaining comments from Roland Bless. References are updated. 433 -03. Several review comments from Roland Bless are addressed. 434 Consistent terminology and equations. Clarification on the scope of 435 recommended beta_{ecn} value. 437 -02. Corrected the equations in Section 3.1. Updated the 438 affiliations. Lower bound for cwnd is defined. A recommendation for 439 window-based transport protocols is changed to cover all transport 440 protocols that implement a congestion control reduction to an ECN 441 congestion signal. Added text about ABE's FreeBSD mainline kernel 442 status including a reference to the FreeBSD code review page. 443 References are updated. 445 -01. Text improved, mainly incorporating comments from Stuart 446 Cheshire. The reference to a technical report has been updated to a 447 published version of the tests [ABE2017]. Used "AQM Mechanism" 448 throughout in place of other alternatives, and more consistent use of 449 technical language and clarification on the intended purpose of the 450 experiments required by EXP status. There was no change to the 451 technical content. 453 -00. draft-ietf-tcpm-alternativebackoff-ecn-00 replaces draft- 454 khademi-tcpm-alternativebackoff-ecn-01. Text describing the nature 455 of the experiment was added. 457 Individual draft -01. This I-D now refers to draft-black-tsvwg-ecn- 458 experimentation-02, which replaces draft-khademi-tsvwg-ecn- 459 response-00 to make a broader update to RFC 3168 for the sake of 460 allowing experiments. As a result, some of the motivating and 461 discussing text that was moved from draft-khademi-alternativebackoff- 462 ecn-03 to draft-khademi-tsvwg-ecn-response-00 has now been re- 463 inserted here. 465 Individual draft -00. draft-khademi-tsvwg-ecn-response-00 and draft- 466 khademi-tcpm-alternativebackoff-ecn-00 replace draft-khademi- 467 alternativebackoff-ecn-03, following discussion in the TSVWG and TCPM 468 working groups. 470 12. References 472 12.1. Normative References 474 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 475 Requirement Levels", BCP 14, RFC 2119, 476 DOI 10.17487/RFC2119, March 1997, 477 . 479 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 480 of Explicit Congestion Notification (ECN) to IP", 481 RFC 3168, DOI 10.17487/RFC3168, September 2001, 482 . 484 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 485 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 486 . 488 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 489 Recommendations Regarding Active Queue Management", 490 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 491 . 493 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 494 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 495 May 2017, . 497 [RFC8257] Bensley, S., Thaler, D., Balasubramanian, P., Eggert, L., 498 and G. Judd, "Data Center TCP (DCTCP): TCP Congestion 499 Control for Data Centers", RFC 8257, DOI 10.17487/RFC8257, 500 October 2017, . 502 [RFC8311] Black, D., "Relaxing Restrictions on Explicit Congestion 503 Notification (ECN) Experimentation", RFC 8311, 504 DOI 10.17487/RFC8311, January 2018, 505 . 507 12.2. Informative References 509 [ABE-FreeBSD] 510 "ABE patch review in FreeBSD", 511 . 514 [ABE2017] Khademi, N., Armitage, G., Welzl, M., Fairhurst, G., 515 Zander, S., and D. Ros, "Alternative Backoff: Achieving 516 Low Latency and High Throughput with ECN and AQM", IFIP 517 NETWORKING 2017, Stockholm, Sweden, June 2017. 519 [BUFFERBLOAT] 520 Gettys, J. and K. Nichols, "Bufferbloat: Dark Buffers in 521 the Internet", November 2011. 523 [CODEL2012] 524 Nichols, K. and V. Jacobson, "Controlling Queue Delay", 525 July 2012, . 527 [I-D.ietf-tcpm-accurate-ecn] 528 Briscoe, B., Kuehlewind, M., and R. Scheffenegger, "More 529 Accurate ECN Feedback in TCP", draft-ietf-tcpm-accurate- 530 ecn-06 (work in progress), March 2018. 532 [ICC2002] Kwon, M. and S. Fahmy, "TCP Increase/Decrease Behavior 533 with Explicit Congestion Notification (ECN)", IEEE 534 ICC 2002, New York, New York, USA, May 2002, 535 . 537 [RFC7713] Mathis, M. and B. Briscoe, "Congestion Exposure (ConEx) 538 Concepts, Abstract Mechanism, and Requirements", RFC 7713, 539 DOI 10.17487/RFC7713, December 2015, 540 . 542 [RFC8033] Pan, R., Natarajan, P., Baker, F., and G. White, 543 "Proportional Integral Controller Enhanced (PIE): A 544 Lightweight Control Scheme to Address the Bufferbloat 545 Problem", RFC 8033, DOI 10.17487/RFC8033, February 2017, 546 . 548 [RFC8087] Fairhurst, G. and M. Welzl, "The Benefits of Using 549 Explicit Congestion Notification (ECN)", RFC 8087, 550 DOI 10.17487/RFC8087, March 2017, 551 . 553 [RFC8289] Nichols, K., Jacobson, V., McGregor, A., Ed., and J. 554 Iyengar, Ed., "Controlled Delay Active Queue Management", 555 RFC 8289, DOI 10.17487/RFC8289, January 2018, 556 . 558 [RFC8312] Rhee, I., Xu, L., Ha, S., Zimmermann, A., Eggert, L., and 559 R. Scheffenegger, "CUBIC for Fast Long-Distance Networks", 560 RFC 8312, DOI 10.17487/RFC8312, February 2018, 561 . 563 Authors' Addresses 565 Naeem Khademi 566 University of Oslo 567 PO Box 1080 Blindern 568 Oslo N-0316 569 Norway 571 Email: naeemk@ifi.uio.no 573 Michael Welzl 574 University of Oslo 575 PO Box 1080 Blindern 576 Oslo N-0316 577 Norway 579 Email: michawe@ifi.uio.no 581 Grenville Armitage 582 Netflix Inc. 584 Email: garmitage@netflix.com 586 Godred Fairhurst 587 University of Aberdeen 588 School of Engineering, Fraser Noble Building 589 Aberdeen AB24 3UE 590 UK 592 Email: gorry@erg.abdn.ac.uk