idnits 2.17.1 draft-ietf-conex-tcp-modifications-06.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 (November 13, 2014) is 3453 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Unused Reference: 'DCTCP' is defined on line 552, but no explicit reference was found in the text == Unused Reference: 'I-D.briscoe-tsvwg-re-ecn-tcp' is defined on line 557, but no explicit reference was found in the text Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Congestion Exposure (ConEx) M. Kuehlewind, Ed. 3 Internet-Draft ETH Zurich 4 Intended status: Experimental R. Scheffenegger 5 Expires: May 17, 2015 NetApp, Inc. 6 November 13, 2014 8 TCP modifications for Congestion Exposure 9 draft-ietf-conex-tcp-modifications-06 11 Abstract 13 Congestion Exposure (ConEx) is a mechanism by which senders inform 14 the network about the congestion encountered by previous packets on 15 the same flow. This document describes the necessary modifications 16 to use ConEx with the Transmission Control Protocol (TCP). 18 Status of This Memo 20 This Internet-Draft is submitted in full conformance with the 21 provisions of BCP 78 and BCP 79. 23 Internet-Drafts are working documents of the Internet Engineering 24 Task Force (IETF). Note that other groups may also distribute 25 working documents as Internet-Drafts. The list of current Internet- 26 Drafts is at http://datatracker.ietf.org/drafts/current/. 28 Internet-Drafts are draft documents valid for a maximum of six months 29 and may be updated, replaced, or obsoleted by other documents at any 30 time. It is inappropriate to use Internet-Drafts as reference 31 material or to cite them other than as "work in progress." 33 This Internet-Draft will expire on May 17, 2015. 35 Copyright Notice 37 Copyright (c) 2014 IETF Trust and the persons identified as the 38 document authors. All rights reserved. 40 This document is subject to BCP 78 and the IETF Trust's Legal 41 Provisions Relating to IETF Documents 42 (http://trustee.ietf.org/license-info) in effect on the date of 43 publication of this document. Please review these documents 44 carefully, as they describe your rights and restrictions with respect 45 to this document. Code Components extracted from this document must 46 include Simplified BSD License text as described in Section 4.e of 47 the Trust Legal Provisions and are provided without warranty as 48 described in the Simplified BSD License. 50 Table of Contents 52 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 53 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 54 2. Sender-side Modifications . . . . . . . . . . . . . . . . . . 3 55 3. Accounting congestion . . . . . . . . . . . . . . . . . . . . 4 56 3.1. Loss Detection . . . . . . . . . . . . . . . . . . . . . 5 57 3.1.1. Without SACK Support . . . . . . . . . . . . . . . . 6 58 3.2. ECN . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 59 3.2.1. Accurate ECN feedback . . . . . . . . . . . . . . . . 8 60 3.2.2. Classic ECN support . . . . . . . . . . . . . . . . . 8 61 4. Setting the ConEx Bits . . . . . . . . . . . . . . . . . . . 9 62 4.1. Setting the E and the L Bit . . . . . . . . . . . . . . . 9 63 4.2. Credit Bits . . . . . . . . . . . . . . . . . . . . . . . 9 64 5. Loss of ConEx information . . . . . . . . . . . . . . . . . . 11 65 6. Timeliness of the ConEx Signals . . . . . . . . . . . . . . . 11 66 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 11 67 8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 11 68 9. Security Considerations . . . . . . . . . . . . . . . . . . . 11 69 10. References . . . . . . . . . . . . . . . . . . . . . . . . . 12 70 10.1. Normative References . . . . . . . . . . . . . . . . . . 12 71 10.2. Informative References . . . . . . . . . . . . . . . . . 12 72 Appendix A. Revision history . . . . . . . . . . . . . . . . . . 13 73 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 14 75 1. Introduction 77 Congestion Exposure (ConEx) is a mechanism by which senders inform 78 the network about the congestion encountered by previous packets on 79 the same flow. ConEx concepts and use cases are further explained in 80 [RFC6789]. The abstract ConEx mechanism is explained in 81 [draft-ietf-conex-abstract-mech]. This document describes the 82 necessary modifications to use ConEx with the Transmission Control 83 Protocol (TCP). 85 The needed markings to provide ConEx signaling are defined in the 86 ConEx Destination Option (CDO) for IPv6 [draft-ietf-conex-destopt]. 87 Specifically, the use of four bits are defined: the X (ConEx- 88 capable), the L (loss experienced), the E (ECN experienced) and C 89 (credit) bit. 91 ConEx signaling is based on loss or Explicit Congestion Notification 92 (ECN) marks [RFC3168] as a congestion indication. This congestion 93 information is retrieved by the sender based on existing feedback 94 mechanisms from the receiver to the sender in TCP. No changes are 95 needed at the receiver to implement ConEx signaling. Therefore no 96 additional negotiation is needed to implement and use ConEx at the 97 sender. This document specifies actions needed by sender to provide 98 meaningful ConEx information to the network. 100 Section 2 provides an overview of the needed modifications for TCP 101 senders to implement ConEx. First congestion information have to be 102 extracted from loss or ECN feedback in TCP as described in section 103 3". Section 4 details how to set the CDO marking based on the 104 accounted congestion information. Section 6 finally discusses 105 timeliness of the ConEx feedback signal as congestion is a temporary 106 state. 108 This document describes congestion accounting for both TCP with and 109 without the Selective Acknowledgment (SACK) extension [RFC2018] in 110 section 3.1. However, ConEx benefits from more accurate information 111 about the number of packets dropped in the network. It is therefore 112 recommend to use the SACK extension when using TCP with ConEx. The 113 detailed mechanism to respectively set the L bit in response to loss- 114 based congestion feedback signal is given in section 4.1. 116 While loss-based congestion feedback should be minimized, ECN could 117 actually provide more fine-grained feedback information. ConEx-based 118 traffic measurement or management mechanisms would benefit from this. 119 Unfortunately, the current ECN feedback mechanism does not reflect 120 multiple congestion markings which occur within the same Round-Trip 121 Time (RTT). A more accurate feedback extension to ECN is proposed in 122 a separate document [draft-kuehlewind-tcpm-accurate-ecn], as this is 123 also useful for other mechanisms. 125 The congestion accounting for both, with the classic ECN feedback as 126 well as a more accurate ECN feedback are explained in detail in 127 section 3.2 while the setting of the E bit in response to ECN-based 128 congestion feedback is again detailed in section 4.1. 130 1.1. Requirements Language 132 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 133 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 134 document are to be interpreted as described in [RFC2119]. 136 2. Sender-side Modifications 138 This section gives an overview of actions that need to be taken by a 139 TCP sender that would like to use ConEx signaling. 141 A ConEx sender MUST negotiate for both SACK and ECN or the more 142 accurate ECN feedback in the TCP handshake if these TCP extension are 143 available at the sender. Therefore a ConEx sender SHOULD also 144 implement SACK and ECN. Depending on the capability of the receiver, 145 the following operation modes exist: 147 o SACK-accECN-ConEx (SACK and accurate ECN feedback) 149 o accECN-ConEx (no SACK but accurate ECN feedback) 151 o ECN-ConEx (no SACK and no accurate ECN feedback but 'classic' ECN) 153 o SACK-ECN-ConEx (SACK and 'classic' instead of accurate ECN) 155 o SACK-ConEx (SACK but no ECN at all) 157 o Basic-ConEx (neither SACK nor ECN) 159 A ConEx sender MUST expose all congestion information to the network 160 according to the congestion information received by ECN or based on 161 loss information provided by the TCP feedback loop. A TCP sender 162 SHOULD account congestion byte-wise (and not packet-wise). A sender 163 MUST mark subsequent packets (after the congestion notification) with 164 the respective ConEx bit in the IP header. Furthermore, a ConEx 165 sender must send enough credit to cover all experienced congestion 166 for the connection so far, as well as the risk of congestion for the 167 current transmission (see Section 4.2). 169 With SACK only the number of lost payload bytes is known, but not the 170 number of packets carrying these bytes. With classic ECN only an 171 indication is given that a marking occurred but not the exact number 172 of payload bytes nor packets. As network congestion is usually byte- 173 congestion [draft-briscoe-tsvwg-byte-pkt-mark], the exact number of 174 bytes should be taken into account, if available, to make the ConEx 175 signal as exact as possible. 177 Detailed mechanisms for congestion accounting in each operation mode 178 are described in the next section. Further handling of the IPv6 bits 179 itself if congestion was accounted is described in the subsequent 180 section afterwards. 182 3. Accounting congestion 184 A ConEx sender, thats accounts congestion byte-wise based on the 185 congestion information received by loss detection or ECN provided by 186 TCP, will maintain two different counters. These counters hold the 187 number of outstanding bytes that should be ConEx marked either with 188 the E bit or the L bit in subsequent packets. 190 The outstanding bytes for congestion indications based on loss are 191 maintained in the loss exposure gauge (LEG) and the accounting is 192 explained in Section 3.1. 194 The outstanding bytes accounted based on ECN feedback information are 195 maintained in the congestion exposure gauge (CEG). The accounting of 196 these bytes from the ECN feedback is explained in more detail next in 197 Section 3.2. 199 Furthermore, those counters will be reduced every time a ConEx 200 capable packet with the E or L bit set is sent. This is explained 201 for both counters in Section 4.1. 203 Usually all bytes of an IP packet must be accounted. Therefore the 204 sender SHOULD take the headers into account, too. If equal sized 205 packets, or at least equally distributed packet sizes can be assumed, 206 the sender MAY only account the TCP payload bytes. In this case 207 there should be about the same number of ConEx marked packets as the 208 original packets that were causing the congestion. Thus both contain 209 about the same number of header bytes. This case is assumed for 210 simplification in the following sections. 212 Otherwise if this is not the case and a sender sends different sized 213 packets (with unequally distributed packet sizes), the sender needs 214 to memorize or estimate the number of ECN-marked or lost packets. A 215 sender might be able to reconstruct the number of packets and thus 216 the header bytes if the packet sizes of all packets that were sent 217 during the last RTT are known. Otherwise if no additional 218 information is available the worst case number of packets and thus 219 header bytes should be estimated in a conservative way based on a 220 minimum packet size (of all packets sent in the last RTT). If the 221 number of ConEx marked packets is smaller (or larger) than the 222 estimated number of ECN-marked or lost packets, the additional header 223 bytes should the added to (or can be subtracted from) the respective 224 counter. 226 3.1. Loss Detection 228 A ConEx sender MUST maintain a loss exposure gauge (LEG), indicating 229 the number of outstanding bytes that must be sent with the ConEx L 230 bit. When a data segment is retransmitted, LEG will be increased by 231 the size of the TCP payload bytes contained by the retransmission, 232 assuming equal sized segments such that the retransmitted packet will 233 have the same number of header bytes as the original ones. 235 Any retransmission may be spurious. To accommodate that, a ConEx 236 sender SHOULD make use of heuristics to detect such spurious 237 retransmissions (e.g. F-RTO [RFC5682], DSACK [RFC3708], and Eifel 239 [RFC3522], [RFC4015]). When such a heuristic has determined, that a 240 certain number of packets were retransmitted erroneously, the ConEx 241 sender should subtract the payload size of these TCP packets from 242 LEG. 244 3.1.1. Without SACK Support 246 If multiple losses occur within one RTT and SACK is not used, it may 247 take several RTTs until all lost data is retransmitted. With the 248 scheme described above, the ConEx information will be delayed 249 strongly but timeliness is important for ConEx. 251 For ConEx it is not important to know which data got lost but only 252 how much. During the first RTT after the initial loss detection, the 253 amount of received data and thus also the amount of lost data can be 254 estimated based on the number of received ACKs. Thus without SACK, 255 the needed information for the ConEx feedback can be available with 256 an additionally delay of one RTT by using the following estimation 257 algorithm: 259 If SACK information is not available, a ConEx sender should maintain 260 an additional Loss Estimation Counter (LEC). With the first 261 retransmission of a congestion event LEC is set to: 263 LEC = f - 3*SMSS 265 where f the is current flight size in bytes. At this point of time 266 in the transmission, in the worst case, all packets in flight minus 267 three that trigged the dupACks could have been lost. For each 268 retransmission that is sent, the LEG will still be increased but the 269 LEC will also be decreased by the payload size of the retransmission. 270 During the following RTT, LEC should be reduced by SMSS for each ACK 271 that is received. Thus after one RTT the LEC estimates the number of 272 outstanding bytes that should be ConEx L marked. To not further 273 delay this information, now LEG should be increased by LEC. From 274 then on every following retransmission should only reduce the LEC and 275 not increase the LEG until the LEC is zero, as those bytes were 276 already accounted. 278 3.2. ECN 280 ECN [RFC3168] is an IP/TCP mechanism that allows network nodes to 281 mark packets with the Congestion Experienced (CE) mark instead of 282 (early) dropping them when congestion occurs. As soon as a CE mark 283 is seen at the receiver, with classic ECN it will feed this 284 information back to the sender by setting the Echo Congestion 285 Experienced (ECE) bit in the TCP header of all subsequent ACKs until 286 a packet with Congestion Window Reduced (CWR) bit in the TCP header 287 is received to acknowledge the reception of the congestion 288 notification. The sender sets the CWR bit in the TCP header once 289 when the first ECE of a congestion notification is received. 291 A receiver can support 'classic' ECN, a more accurate ECN feedback 292 scheme, or neither. In the case ECN is not supported at all, of 293 course, no ECN marks will occur, thus the E bit will never be set. 294 Otherwise, a ConEx sender must maintain a counter, the congestion 295 exposure gauge (CEG), for the number of outstanding bytes that have 296 to be ConEx marked with the E bit. 298 The CEG is increased when ECN information is received from an ECN- 299 capable receiver supporting the 'classic' ECN scheme or the accurate 300 ECN feedback scheme. When the ConEx sender receives an ACK 301 indicating one or more segments were received with a CE mark, CEG is 302 increased by the appropriate number of bytes as described further 303 below. 305 Unfortunately in case of duplicate acknowledgements the number of 306 newly acknowledged bytes will be zero even though (CE marked) data 307 has been received. Therefore, we increase the CEG by DeliveredData, 308 as defined below: 310 DeliveredData = acked_bytes + SACK_diff + (is_dup)*1SMSS - 311 (is_after_dup)*num_dup*1SMSS 313 DeliveredData covers the number of bytes which has been newly 314 delivered to the receiver. Therefore on each arrival of an ACK, 315 DeliveredData will be increased by the newly acknowledged bytes 316 (acked_bytes) as indicated by the current ACK, relative to all past 317 ACKs. 319 Moreover with SACK, DeliveredData is increased by the number of bytes 320 provided by (new) SACK information (SACK_diff). Note, if less 321 unacknowledged bytes are announced in the new SACK information than 322 in the previous ACK, SACK_diff can be negative. In this case, data 323 is newly acknowledged (in acked_byte), that has previously already 324 been accounted to DeliveredData based on SACK information. 326 Without SACK, DeliveredData is estimated to be 1 SMSS on duplicate 327 acknowledgements. For the subsequent partial or full ACK, 328 DeliveredData is estimated to be the newly acknowledged bytes, minus 329 one SMSS for each preceding duplicate ACK. Therefore is_dup is one 330 if the current ACK is a duplicated ACK without SACK, and zero 331 otherwise. is_after_dup is only one for the next full or partial ACK 332 after a number of duplicated ACKs without SACK and num_dup counts the 333 number of duplicated ACKs in a row. 335 The two cases, with and without more accurate ECN depending on the 336 receiver capability, are discussed in the following sections. 338 3.2.1. Accurate ECN feedback 340 With a more accurate ECN feedback scheme either the number of marked 341 packets/received CE marks or directly the number of marked bytes is 342 known. In the later case the CEG can directly be increased by the 343 number of marked bytes. Otherwise if D is assumed to be the number 344 of marks, the gauge CEG will be conservatively increased by one SMSS 345 for each marking or at max the number of newly acknowledged bytes: 347 CEG += min(SMSS*D, DeliveredData) 349 3.2.2. Classic ECN support 351 If the ConEx sender fully conforms to the semantics of the ECN 352 signaling as defined by [RFC5562], it will receive one full RTT of 353 ACKs with the ECE flag set whenever at least one CE mark was received 354 by the receiver. As the sender cannot estimate how much packets have 355 actually been CE marked during this RTT, the most conservative 356 assumption should be taken, namely assuming that all packets were 357 marked. This can be achieved by increasing the CEG by DeliveredData 358 for each ACK with the ECE flag: 360 CEG += DeliveredData 362 Optionally a ConEx sender could implement an Advanced Compatibility 363 Mode: 365 To extract more than one ECE indication per RTT, a ConEx sender could 366 set the CWR flag opportunistically to force the receiver to signal 367 only one ECE per CE mark. Unfortunately, the use of delayed ACKs 368 [RFC5681], as it is usually done today, will prevent a feedback of 369 every CE mark. If an CWR confirmation will be received before the 370 ECE can be sent out with the next ACK, ECN feedback information 371 information could get lost. Thus a sender should set CWR only on 372 those data segments, that will actually trigger a (delayed) ACK. The 373 sender would need an additional control loop to estimated which data 374 segment will trigger an ACK. But such a more sophisticated 375 heuristics could extract congestion notifications more timely. Still 376 the CEG need to be increased by DeliveredData, as one or more CE 377 marked packets could be acknowledged by one delayed ACK. 379 4. Setting the ConEx Bits 381 By setting the X bit a packet is marked as ConEx-capable. All 382 packets carrying payload MUST be marked with the X bit set including 383 retransmissions. No congestion feedback information are available 384 about control packets such as pure ACKs which are not carrying any 385 payload. Thus these packets should not be taken into account when 386 determining ConEx information. These packet MUST carry a ConEx 387 Destination Option with the X bit unset. 389 4.1. Setting the E and the L Bit 391 As long as the CEG or LEG counter is positive, ConEx-capable packets 392 SHOULD be marked with E or L respectively, and the CEG or LEG counter 393 is decreased by the TCP payload bytes carried in this packet. If the 394 CEG or LEG counter is negative, the respective counter SHOULD be 395 reset to zero within one RTT after it was decreased the last time or 396 one RTT after recovery if no further congestion occurred. 398 If SACK information is not available spurious retransmission are more 399 likely. In this case it might be valuable to slightly delay the 400 ConEx loss feedback until a spurious retransmission might be 401 detected. But the ConEx signal MUST NOT be delayed more than one RTT 402 if as long as data packets are sent out. 404 4.2. Credit Bits 406 The ConEx abstract mechanism requires that sufficient credit must be 407 signaled in advance to cover the expected congestion during the 408 feedback delay of one RTT. A ConEx sender should maintain a counter 409 of the sent credits c in bytes. If congestion occurs, credits will 410 be consumed and the c counter should be reduced by the number of 411 bytes that where lost or estimated to be ECN-marked. If the risk of 412 congestion was estimated wrongly and thus too few credits were sent, 413 the c counter becomes zero but can not get negative. 415 The number of credits sent should always equal the number of bytes in 416 flight, as all packets could potentially get lost or congestion 417 marked. Thus a ConEx sender should monitor the number of bytes in 418 flight f. If f ever becomes larger than c, the ConEx sender SHOULD 419 send new credits. Remember that c will be decreased if congestion 420 occurs. 422 In TCP Slow Start, the congestion window might grow much larger than 423 during the rest of the transmission. Thus a sender could consider to 424 sent fewer than f credits but risking potential penalization by an 425 audit. In any case the credits should at least cover the increase in 426 sending rate. As the sending rate increases exponentially in Slow 427 Start, thus double every RTT, a ConEx sender should at least cover 428 half the number of packets in flight by credits. Note, that the 429 number of losses or markings within one RTT does not only depend 430 actions taken by the sender. In general, the behavior of the cross 431 traffic, and if Active Queue Management (AQM) is used, the respective 432 parameterization influence how many packets get dropped or marked. 433 But if the used AQM is not overly aggressive with ECN marking, 434 sending halve the flight size as credits should be sufficient for 435 both, congestion signaled by loss or ECN. Marking every fourth 436 packet will allow the respective number of credits in Slow Start as 437 it can be seen in Figure Figure 1. 439 RTT1 |------XC------>| 440 |------X------->| 441 |------X------->| credit=1 in_flight=3 442 | | 443 RTT2 |------X------->| 444 |------XC------>| 445 |------X------->| 446 |------X------->| 447 |------X------->| 448 |------XC------>| credit=3 in_flight=6 449 | | 450 RTT3 |------X------->| 451 |------X------->| 452 |------X------->| 453 |------XC------>| 454 |------X------->| 455 |------X------->| 456 |------X------->| 457 |------XC------>| 458 |------X------->| 459 |------X------->| 460 |------X------->| 461 |------XC------>| credit=6 in_flight=12 462 | . | 463 | : | 465 Figure 1: Credits in Slow Start (with an initial window of 3) 467 It is possible that the audit looses state due to e.g. rerouting or 468 memory limitations. Therefore, the sender needs to detect this case 469 and resend credits. Thus a ConEx sender should reset the credit 470 count c to zero if losses occur in two subsequent RTTs (assuming that 471 the sending rate was correctly reduced based on the received 472 congestion signal). 474 5. Loss of ConEx information 476 Of course also packets that carry a ConEx marking can get lost. A 477 ConEx sender must remember which packet was marked with either the L, 478 the E or the C bit. If one of these packets is detected to be lost, 479 the should increase the respective gauge, LEG or CEG, by the number 480 of lost payload bytes. 482 6. Timeliness of the ConEx Signals 484 ConEx signals can only be evaluated by a network nodewith a time 485 delay of about one RTT after the congestion occured. To avoiad 486 further delays, a ConEx sender SHOULD sent the ConEx signaling with 487 the next available packet. In cases where it is preferable to 488 slightly delay the ConEx signal, the sender MUST NOT delay the ConEx 489 signal more than one RTT. 491 Multiple ConEx bits may become available for signaling at the same 492 time, for example when an ACK is received by the sender, that 493 indicates at the same time that at least one segment has been lost, 494 and that one or more ECN marks were received. This may happen during 495 excessive congestion, where the queues overflow even though ECN was 496 used and currently all packets are marked, while others have to be 497 dropped nevertheless. Another possibility when this may happen are 498 lost ACKs, so that a subsequent ACK carries summary information not 499 previously available to the sender. As ConEx-capable packet can 500 carry different ConEx marks at the same time, these information do 501 not need to be distributed over several packets and thus can be sent 502 without further delay. 504 7. Acknowledgements 506 The authors would like to thank Bob Briscoe who contributed with this 507 initial ideas and valuable feedback. Moreover, thanks to Jana 508 Iyengar who provided valuable feedback. 510 8. IANA Considerations 512 This document does not have any requests to IANA. 514 9. Security Considerations 516 With some of the advanced ECN compatibility modes it is possible to 517 miss congestion notifications. Thus a sender will not decrease its 518 sending rate. If the congestion is persistent, the likelihood to 519 receive a congestion notification increases. In the worst case the 520 sender will still react correctly to loss. This will prevent a 521 congestion collapse. 523 10. References 525 10.1. Normative References 527 [RFC2018] Mathis, M., Mahdavi, J., Floyd, S., and A. Romanow, "TCP 528 Selective Acknowledgment Options", RFC 2018, October 1996. 530 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 531 Requirement Levels", BCP 14, RFC 2119, March 1997. 533 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 534 of Explicit Congestion Notification (ECN) to IP", RFC 535 3168, September 2001. 537 [RFC5681] Allman, M., Paxson, V., and E. Blanton, "TCP Congestion 538 Control", RFC 5681, September 2009. 540 [draft-ietf-conex-abstract-mech] 541 Mathis, M. and B. Briscoe, "Congestion Exposure (ConEx) 542 Concepts and Abstract Mechanism", draft-ietf-conex- 543 abstract-mech-06 (work in progress), October 2012. 545 [draft-ietf-conex-destopt] 546 Krishnan, S., Kuehlewind, M., and C. Ucendo, "IPv6 547 Destination Option for ConEx", draft-ietf-conex-destopt-04 548 (work in progress), March 2013. 550 10.2. Informative References 552 [DCTCP] Alizadeh, M., Greenberg, A., Maltz, D., Padhye, J., Patel, 553 P., Prabhakar, B., Sengupta, S., and M. Sridharan, "DCTCP: 554 Efficient Packet Transport for the Commoditized Data 555 Center", Jan 2010. 557 [I-D.briscoe-tsvwg-re-ecn-tcp] 558 Briscoe, B., Jacquet, A., Moncaster, T., and A. Smith, 559 "Re-ECN: Adding Accountability for Causing Congestion to 560 TCP/IP", draft-briscoe-tsvwg-re-ecn-tcp-09 (work in 561 progress), October 2010. 563 [RFC3522] Ludwig, R. and M. Meyer, "The Eifel Detection Algorithm 564 for TCP", RFC 3522, April 2003. 566 [RFC3708] Blanton, E. and M. Allman, "Using TCP Duplicate Selective 567 Acknowledgement (DSACKs) and Stream Control Transmission 568 Protocol (SCTP) Duplicate Transmission Sequence Numbers 569 (TSNs) to Detect Spurious Retransmissions", RFC 3708, 570 February 2004. 572 [RFC4015] Ludwig, R. and A. Gurtov, "The Eifel Response Algorithm 573 for TCP", RFC 4015, February 2005. 575 [RFC5562] Kuzmanovic, A., Mondal, A., Floyd, S., and K. 576 Ramakrishnan, "Adding Explicit Congestion Notification 577 (ECN) Capability to TCP's SYN/ACK Packets", RFC 5562, June 578 2009. 580 [RFC5682] Sarolahti, P., Kojo, M., Yamamoto, K., and M. Hata, 581 "Forward RTO-Recovery (F-RTO): An Algorithm for Detecting 582 Spurious Retransmission Timeouts with TCP", RFC 5682, 583 September 2009. 585 [RFC6789] Briscoe, B., Woundy, R., and A. Cooper, "Congestion 586 Exposure (ConEx) Concepts and Use Cases", RFC 6789, 587 December 2012. 589 [draft-briscoe-tsvwg-byte-pkt-mark] 590 Briscoe, B. and J. Manner, "Byte and Packet Congestion 591 Notification", draft-briscoe-tsvwg-byte-pkt-mark-010 (work 592 in progress), May 2013. 594 [draft-kuehlewind-tcpm-accurate-ecn] 595 Kuehlewind, M. and R. Scheffenegger, "More Accurate ECN 596 Feedback in TCP", draft-kuehlewind-tcpm-accurate-ecn-02 597 (work in progress), Jun 2013. 599 Appendix A. Revision history 601 RFC Editior: This section is to be removed before RFC publication. 603 00 ... initial draft, early submission to meet deadline. 605 01 ... refined draft, updated LEG "drain" from per-packet to RTT- 606 based. 608 02 ... added Section 5 and expanded discussion about ECN interaction. 610 03 ... expanded the discussion around credit bits. 612 04 ... review comments of Jana addressed. (Change in full compliance 613 mode.) 615 05 ... changes on Loss Detection without SACK, support of classic ECN 616 and credit handling. 618 Authors' Addresses 620 Mirja Kuehlewind (editor) 621 ETH Zurich 622 Switzerland 624 Email: mirja.kuehlewind@tik.ee.ethz.ch 626 Richard Scheffenegger 627 NetApp, Inc. 628 Am Euro Platz 2 629 Vienna 1120 630 Austria 632 Phone: +43 1 3676811 3146 633 Email: rs@netapp.com