idnits 2.17.1 draft-ietf-tcpm-alternativebackoff-ecn-11.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 seems to lack the recommended RFC 2119 boilerplate, even if it appears to use RFC 2119 keywords -- however, there's a paragraph with a matching beginning. Boilerplate error? (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document date (August 31, 2018) is 2058 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Unused Reference: 'RFC7713' is defined on line 538, but no explicit reference was found in the text == Outdated reference: A later version (-28) exists of draft-ietf-tcpm-accurate-ecn-06 -- Obsolete informational reference (is this intentional?): RFC 8312 (Obsoleted by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 (~~), 4 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-11 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", "NOT RECOMMENDED", "MAY", and 138 "OPTIONAL" in this document are to be interpreted as described in BCP 139 14 RFC 2119 [RFC2119] [RFC8174] when, and only when, they appear in 140 all capitals, as shown here. 142 3. Specification 144 This specification changes the congestion control algorithm of an 145 ECN-Capable TCP transport protocol by changing the TCP sender 146 response to feedback from the TCP receiver that indicates reception 147 of a CE-marked packet, i.e., receipt of a packet with the ECN-Echo 148 flag (defined in [RFC3168]) set, following the process defined in 149 [RFC8311]. 151 The TCP sender response is currently specified in section 6.1.2 of 152 the ECN specification [RFC3168], updated by [RFC8311]: 154 The indication of congestion should be treated just as a 155 congestion loss in non-ECN-Capable TCP. That is, the TCP source 156 halves the congestion window "cwnd" and reduces the slow start 157 threshold "ssthresh", unless otherwise specified by an 158 Experimental RFC in the IETF document stream. 160 Following publication of RFC 8311, this document specifies a sender- 161 side change to TCP: 163 Receipt of a packet with the ECN-Echo flag SHOULD trigger the TCP 164 source to set the slow start threshold (ssthresh) to 0.8 times the 165 FlightSize, with a lower bound of 2 * SMSS applied to the result. 166 As in [RFC5681], the TCP sender also reduces the cwnd value to no 167 more than the new ssthresh value. RFC 3168 section 6.1.2 provides 168 guidance on setting a cwnd less than 2 * SMSS. 170 3.1. Choice of ABE Multiplier 172 ABE decouples the reaction of a TCP sender to inferred packet loss 173 and indication of ECN-signalled congestion in the congestion 174 avoidance phase. To achieve this, ABE uses a different scaling 175 factor in Equation 4 in Section 3.1 of [RFC5681]. The description 176 respectively uses beta_{loss} and beta_{ecn} to refer to the 177 multiplicative decrease factors applied in response to inferred 178 packet loss, and in response to a receiver indicating ECN-signalled 179 congestion. For non-ECN-enabled TCP connections, only beta_{loss} 180 applies. 182 In other words, in response to inferred packet loss: 184 ssthresh = max (FlightSize * beta_{loss}, 2 * SMSS) 186 and in response to an indication of an ECN-signalled congestion: 188 ssthresh = max (FlightSize * beta_{ecn}, 2 * SMSS) 189 and 191 cwnd = ssthresh 193 (If ssthresh == 2 * SMSS, RFC 3168 section 6.1.2 provides guidance 194 on setting a cwnd lower than 2 * SMSS.) 196 where FlightSize is the amount of outstanding data in the network, 197 upper-bounded by the smaller of the sender's cwnd and the receiver's 198 advertised window (rwnd) [RFC5681]. The higher the values of 199 beta_{loss} and beta_{ecn}, the less aggressive the response of any 200 individual backoff event. 202 The appropriate choice for beta_{loss} and beta_{ecn} values is a 203 balancing act between path utilisation and draining the bottleneck 204 queue. More aggressive backoff (smaller beta_*) risks underutilising 205 the path, while less aggressive backoff (larger beta_*) can result in 206 slower draining of the bottleneck queue. 208 The Internet has already been running with at least two different 209 beta_{loss} values for several years: the standard value is 0.5 210 [RFC5681], and the Linux implementation of CUBIC [RFC8312] has used a 211 multiplier of 0.7 since kernel version 2.6.25 released in 2008. ABE 212 does not change the value of beta_{loss} used by current TCP 213 implementations. 215 The recommendation in this document specifies a value of 216 beta_{ecn}=0.8. This recommended beta_{ecn} value is only applicable 217 for the standard TCP congestion control [RFC5681]. The selection of 218 beta_{ecn} enables tuning the response of a TCP connection to shallow 219 AQM marking thresholds. beta_{loss} characterizes the response of a 220 congestion control algorithm to packet loss, i.e., exhaustion of 221 buffers (of unknown depth). Different values for beta_{loss} have 222 been suggested for TCP congestion control algorithms. Consequently, 223 beta_{ecn} is likely to be an algorithm-specific parameter rather 224 than a constant multiple of the algorithm's existing beta_{loss}. 226 A range of tests (section IV, [ABE2017]) with NewReno and CUBIC over 227 CoDel and PIE in lightly-multiplexed scenarios have explored this 228 choice of parameter. The results of these tests indicate that CUBIC 229 connections benefit from beta_{ecn} of 0.85 (cf. beta_{loss} = 0.7), 230 and NewReno connections see improvements with beta_{ecn} in the range 231 0.7 to 0.85 (cf. beta_{loss} = 0.5). 233 4. Discussion 235 Much of the technical background to ABE can be found in a research 236 paper [ABE2017]. This paper used a mix of experiments, theory and 237 simulations with NewReno [RFC5681] and CUBIC [RFC8312] to evaluate 238 the technique. The technique was shown to present "...significant 239 performance gains in lightly-multiplexed [few concurrent flows] 240 scenarios, without losing the delay-reduction benefits of deploying 241 CoDel or PIE". The performance improvement is achieved when reacting 242 to ECN-Echo in congestion avoidance (when ssthresh > cwnd) by 243 multiplying cwnd and ssthresh with a value in the range [0.7,0.85]. 244 Applying ABE when cwnd <= ssthresh is not currently recommended, but 245 may benefit from additional attention, experimentation and 246 specification. 248 4.1. Why Use ECN to Vary the Degree of Backoff? 250 AQM mechanisms such as CoDel [RFC8289] and PIE [RFC8033] set a delay 251 target in routers and use congestion notifications to constrain the 252 queuing delays experienced by packets, rather than in response to 253 impending or actual bottleneck buffer exhaustion. With current 254 default delay targets, CoDel and PIE both effectively emulate a 255 bottleneck with a short queue (section II, [ABE2017]) while also 256 allowing short traffic bursts into the queue. This provides 257 acceptable performance for TCP connections over a path with a low 258 BDP, or in highly multiplexed scenarios (many concurrent transport 259 flows). However, in a lightly-multiplexed case over a path with a 260 large BDP, conventional TCP backoff leads to gaps in packet 261 transmission and under-utilisation of the path. 263 Instead of discarding packets, an AQM mechanism is allowed to mark 264 ECN-Capable packets with an ECN CE-mark. The reception of a CE-mark 265 feedback not only indicates congestion on the network path, it also 266 indicates that an AQM mechanism exists at the bottleneck along the 267 path, and hence the CE-mark likely came from a bottleneck with a 268 controlled short queue. Reacting differently to an ECN-signalled 269 congestion than to an inferred packet loss can then yield the benefit 270 of a reduced back-off when queues are short. Using ECN can also be 271 advantageous for several other reasons [RFC8087]. 273 The idea of reacting differently to inferred packet loss and 274 detection of an ECN-signalled congestion pre-dates this 275 specification. For example, previous research proposed using ECN CE- 276 marked feedback to modify TCP congestion control behaviour via a 277 larger multiplicative decrease factor in conjunction with a smaller 278 additive increase factor [ICC2002]. The goal of this former work was 279 to operate across AQM bottlenecks using Random Early Detection (RED) 280 that were not necessarily configured to emulate a short queue (The 281 current usage of RED as an Internet AQM method is limited [RFC7567]). 283 4.2. An RTT-based response to indicated congestion 285 This specification applies to the use of ECN feedback as defined in 286 [RFC3168], which specifies a response to indicated congestion that is 287 no more frequent that once per path round trip time. Since ABE 288 responds to indicated congestion once per RTT, it therefore does not 289 respond to any further loss within the same RTT, because an ABE 290 sender has already reduced the congestion window. If congestion 291 persists after such reduction, ABE continues to reduce the congestion 292 window in each consecutive RTT. This consecutive reduction can 293 protect the network against long-standing unfairness in the case of 294 AQM algorithms that do not keep a small average queue length. The 295 mechanism does not rely on Accurate ECN 296 ([I-D.ietf-tcpm-accurate-ecn]). 298 In contrast, transport protocol mechanisms can also be designed to 299 utilise more frequent and detailed ECN feedback (e.g., Accurate ECN 300 [I-D.ietf-tcpm-accurate-ecn]), which then permit a congestion control 301 response that adjusts the sending rate more frequently. Datacenter 302 TCP (DCTCP) [RFC8257] is an example of this approach. 304 5. ABE Deployment Requirements 306 This update is a sender-side only change. Like other changes to 307 congestion control algorithms, it does not require any change to the 308 TCP receiver or to network devices. It does not require any ABE- 309 specific changes in routers or the use of Accurate ECN feedback 310 [I-D.ietf-tcpm-accurate-ecn] by a receiver. 312 If the method is only deployed by some senders, and not by others, 313 the senders that use this method can gain some advantage, possibly at 314 the expense of other flows that do not use this updated method. 315 Because this advantage applies only to ECN-marked packets and not to 316 packet loss indications, an ECN-Capable bottleneck will still fall 317 back to dropping packets if an TCP sender using ABE is too 318 aggressive, and the result is no different than if the TCP sender was 319 using traditional loss-based congestion control. 321 When used with bottlenecks that do not support ECN-marking the 322 specification does not modify the transport protocol. 324 6. ABE Experiment Goals 326 [RFC3168] states that the congestion control response following an 327 indication of ECN-signalled congestion is the same as the response to 328 a dropped packet. [RFC8311] updates this specification to allow 329 systems to provide a different behaviour when they experience ECN- 330 signalled congestion rather than packet loss. The present 331 specification defines such an experiment and has thus been assigned 332 an Experimental status before being proposed as a Standards-Track 333 update. 335 The purpose of the Internet experiment is to collect experience with 336 deployment of ABE, and confirm acceptable safety in deployed networks 337 that use this update to TCP congestion control. To evaluate ABE, 338 this experiment therefore requires support in AQM routers for ECN- 339 marking of packets carrying the ECN-Capable Transport, ECT(0), 340 codepoint [RFC3168]. 342 The result of this Internet experiment ought to include an 343 investigation of the implications of experiencing an ECN-CE mark 344 followed by loss within the same RTT. At the end of the experiment, 345 this will be reported to the TCPM WG or the IESG. 347 7. Acknowledgements 349 Authors N. Khademi, M. Welzl and G. Fairhurst were part-funded by 350 the European Community under its Seventh Framework Programme through 351 the Reducing Internet Transport Latency (RITE) project (ICT-317700). 352 The views expressed are solely those of the authors. 354 Author G. Armitage performed most of his work on this document while 355 employed by Swinburne University of Technology, Melbourne, Australia. 357 The authors would like to thank Stuart Cheshire for many suggestions 358 when revising the draft, and the following people for their 359 contributions to [ABE2017]: Chamil Kulatunga, David Ros, Stein 360 Gjessing, Sebastian Zander. Thanks also to (in alphabetical order) 361 Roland Bless, Bob Briscoe, David Black, Markku Kojo, John Leslie, 362 Lawrence Stewart, Dave Taht and the TCPM Working Group for providing 363 valuable feedback on this document. 365 The authors would finally like to thank everyone who provided 366 feedback on the congestion control behaviour specified in this update 367 received from the IRTF Internet Congestion Control Research Group 368 (ICCRG). 370 8. IANA Considerations 372 XX RFC ED - PLEASE REMOVE THIS SECTION XXX 374 This document includes no request to IANA. 376 9. Implementation Status 378 ABE is implemented as a patch for Linux and FreeBSD. This is meant 379 for research and available for download from 380 http://heim.ifi.uio.no/michawe/research/abe/. This code was used to 381 produce the test results that are reported in [ABE2017]. The FreeBSD 382 code has been committed to the mainline kernel on March 19, 2018 383 [ABE-FreeBSD]. 385 10. Security Considerations 387 The described method is a sender-side only transport change, and does 388 not change the protocol messages exchanged. The security 389 considerations for ECN [RFC3168] therefore still apply. 391 This is a change to TCP congestion control with ECN that will 392 typically lead to a change in the capacity achieved when flows share 393 a network bottleneck. This could result in some flows receiving more 394 than their fair share of capacity. Similar unfairness in the way 395 that capacity is shared is also exhibited by other congestion control 396 mechanisms that have been in use in the Internet for many years 397 (e.g., CUBIC [RFC8312]). Unfairness may also be a result of other 398 factors, including the round trip time experienced by a flow. ABE 399 applies only when ECN-marked packets are received, not when packets 400 are lost, hence use of ABE cannot lead to congestion collapse. 402 11. Revision Information 404 XX RFC ED - PLEASE REMOVE THIS SECTION XXX 406 -10. Incorported changes following the Gen-ART review by Russ 407 Housley. Correction to URL. 409 -09. Changed to "Following publication of RFC 8311, this document 410 specifies a sender-side change to TCP:" 412 -08. Addressed comments from AD review on the document structure, 413 and relationship to existing RFCs. 415 -07. Addressed comments following WGLC. 417 o Updated Reference citations. 419 o Removed paragraph containing a wrong statement related to timeout 420 in section 4.1. 422 o Discuss what happens when cwnd <= ssthresh. 424 o Added text on Concern about lower bound of 2*SMSS. 426 -06. Addressed Michael Scharf's comments. 428 -05. Refined the description of the experiment based on feedback at 429 IETF-100. Incorporated comments from David Black. 431 -04. Incorporates review comments from Lawrence Stewart and the 432 remaining comments from Roland Bless. References are updated. 434 -03. Several review comments from Roland Bless are addressed. 435 Consistent terminology and equations. Clarification on the scope of 436 recommended beta_{ecn} value. 438 -02. Corrected the equations in Section 3.1. Updated the 439 affiliations. Lower bound for cwnd is defined. A recommendation for 440 window-based transport protocols is changed to cover all transport 441 protocols that implement a congestion control reduction to an ECN 442 congestion signal. Added text about ABE's FreeBSD mainline kernel 443 status including a reference to the FreeBSD code review page. 444 References are updated. 446 -01. Text improved, mainly incorporating comments from Stuart 447 Cheshire. The reference to a technical report has been updated to a 448 published version of the tests [ABE2017]. Used "AQM Mechanism" 449 throughout in place of other alternatives, and more consistent use of 450 technical language and clarification on the intended purpose of the 451 experiments required by EXP status. There was no change to the 452 technical content. 454 -00. draft-ietf-tcpm-alternativebackoff-ecn-00 replaces draft- 455 khademi-tcpm-alternativebackoff-ecn-01. Text describing the nature 456 of the experiment was added. 458 Individual draft -01. This I-D now refers to draft-black-tsvwg-ecn- 459 experimentation-02, which replaces draft-khademi-tsvwg-ecn- 460 response-00 to make a broader update to RFC 3168 for the sake of 461 allowing experiments. As a result, some of the motivating and 462 discussing text that was moved from draft-khademi-alternativebackoff- 463 ecn-03 to draft-khademi-tsvwg-ecn-response-00 has now been re- 464 inserted here. 466 Individual draft -00. draft-khademi-tsvwg-ecn-response-00 and draft- 467 khademi-tcpm-alternativebackoff-ecn-00 replace draft-khademi- 468 alternativebackoff-ecn-03, following discussion in the TSVWG and TCPM 469 working groups. 471 12. References 473 12.1. Normative References 475 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 476 Requirement Levels", BCP 14, RFC 2119, 477 DOI 10.17487/RFC2119, March 1997, 478 . 480 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 481 of Explicit Congestion Notification (ECN) to IP", 482 RFC 3168, DOI 10.17487/RFC3168, September 2001, 483 . 485 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 486 Control", RFC 5681, DOI 10.17487/RFC5681, September 2009, 487 . 489 [RFC7567] Baker, F., Ed. and G. Fairhurst, Ed., "IETF 490 Recommendations Regarding Active Queue Management", 491 BCP 197, RFC 7567, DOI 10.17487/RFC7567, July 2015, 492 . 494 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 495 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 496 May 2017, . 498 [RFC8257] Bensley, S., Thaler, D., Balasubramanian, P., Eggert, L., 499 and G. Judd, "Data Center TCP (DCTCP): TCP Congestion 500 Control for Data Centers", RFC 8257, DOI 10.17487/RFC8257, 501 October 2017, . 503 [RFC8311] Black, D., "Relaxing Restrictions on Explicit Congestion 504 Notification (ECN) Experimentation", RFC 8311, 505 DOI 10.17487/RFC8311, January 2018, 506 . 508 12.2. Informative References 510 [ABE-FreeBSD] 511 "ABE patch review in FreeBSD", 512 . 515 [ABE2017] Khademi, N., Armitage, G., Welzl, M., Fairhurst, G., 516 Zander, S., and D. Ros, "Alternative Backoff: Achieving 517 Low Latency and High Throughput with ECN and AQM", IFIP 518 NETWORKING 2017, Stockholm, Sweden, June 2017. 520 [BUFFERBLOAT] 521 Gettys, J. and K. Nichols, "Bufferbloat: Dark Buffers in 522 the Internet", November 2011. 524 [CODEL2012] 525 Nichols, K. and V. Jacobson, "Controlling Queue Delay", 526 July 2012, . 528 [I-D.ietf-tcpm-accurate-ecn] 529 Briscoe, B., Kuehlewind, M., and R. Scheffenegger, "More 530 Accurate ECN Feedback in TCP", draft-ietf-tcpm-accurate- 531 ecn-06 (work in progress), March 2018. 533 [ICC2002] Kwon, M. and S. Fahmy, "TCP Increase/Decrease Behavior 534 with Explicit Congestion Notification (ECN)", IEEE 535 ICC 2002, New York, New York, USA, May 2002, 536 . 538 [RFC7713] Mathis, M. and B. Briscoe, "Congestion Exposure (ConEx) 539 Concepts, Abstract Mechanism, and Requirements", RFC 7713, 540 DOI 10.17487/RFC7713, December 2015, 541 . 543 [RFC8033] Pan, R., Natarajan, P., Baker, F., and G. White, 544 "Proportional Integral Controller Enhanced (PIE): A 545 Lightweight Control Scheme to Address the Bufferbloat 546 Problem", RFC 8033, DOI 10.17487/RFC8033, February 2017, 547 . 549 [RFC8087] Fairhurst, G. and M. Welzl, "The Benefits of Using 550 Explicit Congestion Notification (ECN)", RFC 8087, 551 DOI 10.17487/RFC8087, March 2017, 552 . 554 [RFC8289] Nichols, K., Jacobson, V., McGregor, A., Ed., and J. 555 Iyengar, Ed., "Controlled Delay Active Queue Management", 556 RFC 8289, DOI 10.17487/RFC8289, January 2018, 557 . 559 [RFC8312] Rhee, I., Xu, L., Ha, S., Zimmermann, A., Eggert, L., and 560 R. Scheffenegger, "CUBIC for Fast Long-Distance Networks", 561 RFC 8312, DOI 10.17487/RFC8312, February 2018, 562 . 564 Authors' Addresses 566 Naeem Khademi 567 University of Oslo 568 PO Box 1080 Blindern 569 Oslo N-0316 570 Norway 572 Email: naeemk@ifi.uio.no 574 Michael Welzl 575 University of Oslo 576 PO Box 1080 Blindern 577 Oslo N-0316 578 Norway 580 Email: michawe@ifi.uio.no 582 Grenville Armitage 583 Netflix Inc. 585 Email: garmitage@netflix.com 587 Godred Fairhurst 588 University of Aberdeen 589 School of Engineering, Fraser Noble Building 590 Aberdeen AB24 3UE 591 UK 593 Email: gorry@erg.abdn.ac.uk