idnits 2.17.1 draft-ietf-quic-recovery-01.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 Security Considerations section. ** There are 3 instances of too long lines in the document, the longest one being 10 characters in excess of 72. 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. (The document does seem to have the reference to RFC 2119 which the ID-Checklist requires). -- The document date (January 14, 2017) is 2658 days in the past. Is this intentional? -- Found something which looks like a code comment -- if you have code sections in the document, please surround them with '' and '' lines. Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Unused Reference: 'QUIC-TLS' is defined on line 539, but no explicit reference was found in the text -- Possible downref: Non-RFC (?) normative reference: ref. 'QUIC-TLS' -- Possible downref: Non-RFC (?) normative reference: ref. 'QUIC-TRANSPORT' Summary: 2 errors (**), 0 flaws (~~), 3 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 QUIC J. Iyengar, Ed. 3 Internet-Draft I. Swett, Ed. 4 Intended status: Standards Track Google 5 Expires: July 18, 2017 January 14, 2017 7 QUIC Loss Detection and Congestion Control 8 draft-ietf-quic-recovery-01 10 Abstract 12 QUIC is a new multiplexed and secure transport atop UDP. QUIC builds 13 on decades of transport and security experience, and implements 14 mechanisms that make it attractive as a modern general-purpose 15 transport. QUIC implements the spirit of known TCP loss detection 16 mechanisms, described in RFCs, various Internet-drafts, and also 17 those prevalent in the Linux TCP implementation. This document 18 describes QUIC loss detection and congestion control, and attributes 19 the TCP equivalent in RFCs, Internet-drafts, academic papers, and TCP 20 implementations. 22 Note to Readers 24 Discussion of this draft takes place on the QUIC working group 25 mailing list (quic@ietf.org), which is archived at 26 https://mailarchive.ietf.org/arch/search/?email_list=quic . 28 Working Group information can be found at https://github.com/quicwg ; 29 source code and issues list for this draft can be found at 30 https://github.com/quicwg/base-drafts/labels/recovery . 32 Status of This Memo 34 This Internet-Draft is submitted in full conformance with the 35 provisions of BCP 78 and BCP 79. 37 Internet-Drafts are working documents of the Internet Engineering 38 Task Force (IETF). Note that other groups may also distribute 39 working documents as Internet-Drafts. The list of current Internet- 40 Drafts is at http://datatracker.ietf.org/drafts/current/. 42 Internet-Drafts are draft documents valid for a maximum of six months 43 and may be updated, replaced, or obsoleted by other documents at any 44 time. It is inappropriate to use Internet-Drafts as reference 45 material or to cite them other than as "work in progress." 47 This Internet-Draft will expire on July 18, 2017. 49 Copyright Notice 51 Copyright (c) 2017 IETF Trust and the persons identified as the 52 document authors. All rights reserved. 54 This document is subject to BCP 78 and the IETF Trust's Legal 55 Provisions Relating to IETF Documents 56 (http://trustee.ietf.org/license-info) in effect on the date of 57 publication of this document. Please review these documents 58 carefully, as they describe your rights and restrictions with respect 59 to this document. Code Components extracted from this document must 60 include Simplified BSD License text as described in Section 4.e of 61 the Trust Legal Provisions and are provided without warranty as 62 described in the Simplified BSD License. 64 Table of Contents 66 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 67 1.1. Notational Conventions . . . . . . . . . . . . . . . . . 3 68 2. Design of the QUIC Transmission Machinery . . . . . . . . . . 3 69 2.1. Relevant Differences Between QUIC and TCP . . . . . . . . 4 70 2.1.1. Monotonically Increasing Packet Numbers . . . . . . . 4 71 2.1.2. No Reneging . . . . . . . . . . . . . . . . . . . . . 5 72 2.1.3. More ACK Ranges . . . . . . . . . . . . . . . . . . . 5 73 2.1.4. Explicit Correction For Delayed Acks . . . . . . . . 5 74 3. Loss Detection . . . . . . . . . . . . . . . . . . . . . . . 5 75 3.1. Constants of interest . . . . . . . . . . . . . . . . . . 5 76 3.2. Variables of interest . . . . . . . . . . . . . . . . . . 6 77 3.3. Initialization . . . . . . . . . . . . . . . . . . . . . 6 78 3.4. Setting the Loss Detection Alarm . . . . . . . . . . . . 7 79 3.5. On Sending a Packet . . . . . . . . . . . . . . . . . . . 8 80 3.6. On Ack Receipt . . . . . . . . . . . . . . . . . . . . . 8 81 3.7. On Packet Acknowledgment . . . . . . . . . . . . . . . . 9 82 3.8. On Alarm Firing . . . . . . . . . . . . . . . . . . . . . 10 83 3.9. Detecting Lost Packets . . . . . . . . . . . . . . . . . 10 84 4. Congestion Control . . . . . . . . . . . . . . . . . . . . . 10 85 5. TCP mechanisms in QUIC . . . . . . . . . . . . . . . . . . . 10 86 5.1. RFC 6298 (RTO computation) . . . . . . . . . . . . . . . 11 87 5.2. FACK Loss Recovery (paper) . . . . . . . . . . . . . . . 11 88 5.3. RFC 3782, RFC 6582 (NewReno Fast Recovery) . . . . . . . 11 89 5.4. TLP (draft) . . . . . . . . . . . . . . . . . . . . . . . 11 90 5.5. RFC 5827 (Early Retransmit) with Delay Timer . . . . . . 11 91 5.6. RFC 5827 (F-RTO) . . . . . . . . . . . . . . . . . . . . 12 92 5.7. RFC 6937 (Proportional Rate Reduction) . . . . . . . . . 12 93 5.8. TCP Cubic (draft) with optional RFC 5681 (Reno) . . . . . 12 94 5.9. Hybrid Slow Start (paper) . . . . . . . . . . . . . . . . 12 95 5.10. RACK (draft) . . . . . . . . . . . . . . . . . . . . . . 12 96 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 12 97 7. Normative References . . . . . . . . . . . . . . . . . . . . 12 98 Appendix A. Acknowledgments . . . . . . . . . . . . . . . . . . 13 99 Appendix B. Change Log . . . . . . . . . . . . . . . . . . . . . 13 100 B.1. Since draft-ietf-quic-recovery-00: . . . . . . . . . . . 13 101 B.2. Since draft-iyengar-quic-loss-recovery-01: . . . . . . . 13 102 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 13 104 1. Introduction 106 QUIC is a new multiplexed and secure transport atop UDP. QUIC builds 107 on decades of transport and security experience, and implements 108 mechanisms that make it attractive as a modern general-purpose 109 transport. The QUIC protocol is described in [QUIC-TRANSPORT]. 111 QUIC implements the spirit of known TCP loss recovery mechanisms, 112 described in RFCs, various Internet-drafts, and also those prevalent 113 in the Linux TCP implementation. This document describes QUIC 114 congestion control and loss recovery, and where applicable, 115 attributes the TCP equivalent in RFCs, Internet-drafts, academic 116 papers, and/or TCP implementations. 118 This document first describes pre-requisite parts of the QUIC 119 transmission machinery, then discusses QUIC's default congestion 120 control and loss detection mechanisms, and finally lists the various 121 TCP mechanisms that QUIC loss detection implements (in spirit.) 123 1.1. Notational Conventions 125 The words "MUST", "MUST NOT", "SHOULD", and "MAY" are used in this 126 document. It's not shouting; when they are capitalized, they have 127 the special meaning defined in [RFC2119]. 129 2. Design of the QUIC Transmission Machinery 131 All transmissions in QUIC are sent with a packet-level header, which 132 includes a packet sequence number (referred to below as a packet 133 number). These packet numbers never repeat in the lifetime of a 134 connection, and are monotonically increasing, which makes duplicate 135 detection trivial. This fundamental design decision obviates the 136 need for disambiguating between transmissions and retransmissions and 137 eliminates significant complexity from QUIC's interpretation of TCP 138 loss detection mechanisms. 140 Every packet may contain several frames. We outline the frames that 141 are important to the loss detection and congestion control machinery 142 below. 144 o Retransmittable frames are frames requiring reliable delivery. 145 The most common are STREAM frames, which typically contain 146 application data. 148 o Crypto handshake data is also sent as STREAM data, and uses the 149 reliability machinery of QUIC underneath. 151 o ACK frames contain acknowledgment information. QUIC uses a SACK- 152 based scheme, where acks express up to 256 ranges. The ACK frame 153 also includes a receive timestamp for each packet newly acked. 155 2.1. Relevant Differences Between QUIC and TCP 157 There are some notable differences between QUIC and TCP which are 158 important for reasoning about the differences between the loss 159 recovery mechanisms employed by the two protocols. We briefly 160 describe these differences below. 162 2.1.1. Monotonically Increasing Packet Numbers 164 TCP conflates transmission sequence number at the sender with 165 delivery sequence number at the receiver, which results in 166 retransmissions of the same data carrying the same sequence number, 167 and consequently to problems caused by "retransmission ambiguity". 168 QUIC separates the two: QUIC uses a packet sequence number (referred 169 to as the "packet number") for transmissions, and any data that is to 170 be delivered to the receiving application(s) is sent in one or more 171 streams, with stream offsets encoded within STREAM frames inside of 172 packets that determine delivery order. 174 QUIC's packet number is strictly increasing, and directly encodes 175 transmission order. A higher QUIC packet number signifies that the 176 packet was sent later, and a lower QUIC packet number signifies that 177 the packet was sent earlier. When a packet containing frames is 178 deemed lost, QUIC rebundles necessary frames in a new packet with a 179 new packet number, removing ambiguity about which packet is 180 acknowledged when an ACK is received. Consequently, more accurate 181 RTT measurements can be made, spurious retransmissions are trivially 182 detected, and mechanisms such as Fast Retransmit can be applied 183 universally, based only on packet number. 185 This design point significantly simplifies loss detection mechanisms 186 for QUIC. Most TCP mechanisms implicitly attempt to infer 187 transmission ordering based on TCP sequence numbers - a non-trivial 188 task, especially when TCP timestamps are not available. 190 2.1.2. No Reneging 192 QUIC ACKs contain information that is equivalent to TCP SACK, but 193 QUIC does not allow any acked packet to be reneged, greatly 194 simplifying implementations on both sides and reducing memory 195 pressure on the sender. 197 2.1.3. More ACK Ranges 199 QUIC supports up to 256 ACK ranges, opposed to TCP's 3 SACK ranges. 200 In high loss environments, this speeds recovery. 202 2.1.4. Explicit Correction For Delayed Acks 204 QUIC ACKs explicitly encode the delay incurred at the receiver 205 between when a packet is received and when the corresponding ACK is 206 sent. This allows the receiver of the ACK to adjust for receiver 207 delays, specifically the delayed ack timer, when estimating the path 208 RTT. This mechanism also allows a receiver to measure and report the 209 delay from when a packet was received by the OS kernel, which is 210 useful in receivers which may incur delays such as context-switch 211 latency before a userspace QUIC receiver processes a received packet. 213 3. Loss Detection 215 We now describe QUIC's loss detection as functions that should be 216 called on packet transmission, when a packet is acked, and timer 217 expiration events. 219 3.1. Constants of interest 221 Constants used in loss recovery and congestion control are based on a 222 combination of RFCs, papers, and common practice. Some may need to 223 be changed or negotiated in order to better suit a variety of 224 environments. 226 o kMaxTLPs: 2 Maximum number of tail loss probes before an RTO 227 fires. 229 o kReorderingThreshold: 3 Maximum reordering in packet number space 230 before FACK style loss detection considers a packet lost. 232 o kTimeReorderingThreshold: 1/8 Maximum reordering in time sapce 233 before time based loss detection considers a packet lost. In 234 fraction of an RTT. 236 o kMinTLPTimeout: 10ms Minimum time in the future a tail loss probe 237 alarm may be set for. 239 o kMinRTOTimeout: 200ms Minimum time in the future an RTO alarm may 240 be set for. 242 o kDelayedAckTimeout: 25ms The length of the peer's delayed ack 243 timer. 245 3.2. Variables of interest 247 We first describe the variables required to implement the loss 248 detection mechanisms described in this section. 250 o loss_detection_alarm: Multi-modal alarm used for loss detection. 252 o alarm_mode: QUIC maintains a single loss detection alarm, which 253 switches between various modes. This mode is used to determine 254 the duration of the alarm. 256 o handshake_count: The number of times the handshake packets have 257 been retransmitted without receiving an ack. 259 o tlp_count: The number of times a tail loss probe has been sent 260 without receiving an ack. 262 o rto_count: The number of times an rto has been sent without 263 receiving an ack. 265 o smoothed_rtt: The smoothed RTT of the connection, computed as 266 described in [RFC6298] 268 o rttvar: The RTT variance. 270 o reordering_threshold: The largest delta between the largest acked 271 retransmittable packet and a packet containing retransmittable 272 frames before it's declared lost. 274 o use_time_loss: When true, loss detection operates solely based on 275 reordering threshold in time, rather than in packet number gaps. 277 o sent_packets: An association of packet numbers to information 278 about them. 280 3.3. Initialization 282 At the beginning of the connection, initialize the loss detection 283 variables as follows: 285 loss_detection_alarm.reset(); 286 handshake_count = 0; 287 tlp_count = 0; 288 rto_count = 0; 289 reordering_threshold = kReorderingThreshold; 290 use_time_loss = false; 291 smoothed_rtt = 0; 292 rttvar = 0; 294 3.4. Setting the Loss Detection Alarm 296 QUIC loss detection uses a single alarm for all timer-based loss 297 detection. The duration of the alarm is based on the alarm's mode, 298 which is set in the packet and timer events further below. The 299 function SetLossDetectionAlarm defined below shows how the single 300 timer is set based on the alarm mode. 302 Pseudocode for SetLossDetectionAlarm follows: 304 SetLossDetectionAlarm(): 305 if (retransmittable packets are not outstanding): 306 loss_detection_alarm.cancel(); 307 return; 309 if (handshake packets are outstanding): 310 // Handshake retransmission alarm. 311 alarm_duration = max(1.5 * smoothed_rtt, kMinTLPTimeout) << handshake_count; 312 handshake_count++; 313 else if (largest sent packet is acked): 314 // Early retransmit alarm. 315 alarm_duration = 0.25 x smoothed_rtt; 316 else if (tlp_count < kMaxTLPs): 317 // Tail Loss Probe alarm. 318 if (retransmittable_packets_outstanding = 1): 319 alarm_duration = max(1.5 x smoothed_rtt + kDelayedAckTimeout, 320 2 x smoothed_rtt); 321 else: 322 alarm_duration = max (kMinTLPTimeout, 2 x smoothed_rtt); 323 tlp_count++; 324 else: 325 // RTO alarm. 326 if (rto_count = 0): 327 alarm_duration = max(kMinRTOTimeout, smoothed_rtt + 4 x rttvar); 328 else: 329 alarm_duration = loss_detection_alarm.get_delay() << 1; 330 rto_count++; 332 loss_detection_alarm.set(now + alarm_duration); 334 3.5. On Sending a Packet 336 After any packet is sent, be it a new transmission or a rebundled 337 transmission, the following OnPacketSent function is called. The 338 parameters to OnPacketSent are as follows: 340 o packet_number: The packet number of the sent packet. 342 o is_retransmittble: A boolean that indicates whether the packet 343 contains at least one frame requiring reliable deliver. The 344 retransmittability of various QUIC frames is described in 345 [QUIC-TRANSPORT]. If false, it is still acceptable for an ack to 346 be received for this packet. However, a caller MUST NOT set 347 is_retransmittable to true if an ack is not expected. 349 Pseudocode for OnPacketSent follows: 351 OnPacketSent(packet_number, is_retransmittable): 352 # TODO: Clarify the data in sent_packets. 353 sent_packets[packet_number] = {now} 354 if is_retransmittable: 355 SetLossDetectionAlarm() 357 3.6. On Ack Receipt 359 When an ack is received, it may acknowledge 0 or more packets. 361 Pseudocode for OnAckReceived and UpdateRtt follow: 363 OnAckReceived(ack): 364 // If the largest acked is newly acked, update the RTT. 365 if (sent_packets[ack.largest_acked]): 366 rtt_sample = now - sent_packets[ack.largest_acked] 367 if (rtt_sample > ack.ack_delay): 368 rtt_sample -= ack.delay; 369 UpdateRtt(rtt_sample) 370 // Find all newly acked packets. 371 for acked_packet in DetermineNewlyAckedPackets(): 372 OnPacketAcked(acked_packet) 374 DetectLostPackets(ack.largest_acked_packet); 375 SetLossDetectionAlarm(); 377 UpdateRtt(rtt_sample): 378 if (smoothed_rtt == 0): 379 smoothed_rtt = rtt_sample 380 rttvar = rtt_sample / 2 381 else: 382 rttvar = 3/4 * rttvar + 1/4 * (smoothed_rtt - rtt_sample) 383 smoothed_rtt = 7/8 * smoothed_rtt + 1/8 * rtt_sample 385 3.7. On Packet Acknowledgment 387 When a packet is acked for the first time, the following 388 OnPacketAcked function is called. Note that a single ACK frame may 389 newly acknowledge several packets. OnPacketAcked must be called once 390 for each of these newly acked packets. 392 OnPacketAcked takes one parameter, acked_packet, which is the packet 393 number of the newly acked packet, and returns a list of packet 394 numbers that are detected as lost. 396 Pseudocode for OnPacketAcked follows: 398 OnPacketAcked(acked_packet): 399 handshake_count = 0; 400 tlp_count = 0; 401 rto_count = 0; 402 # TODO: Don't remove packets immediately, since they can be used for 403 # detecting spurous retransmits. 404 sent_packets.remove(acked_packet); 406 3.8. On Alarm Firing 408 QUIC uses one loss recovery alarm, which when set, can be in one of 409 several modes. When the alarm fires, the mode determines the action 410 to be performed. OnAlarm returns a list of packet numbers that are 411 detected as lost. 413 Pseudocode for OnAlarm follows: 415 OnAlarm(acked_packet): 416 lost_packets = DetectLostPackets(acked_packet); 417 MaybeRetransmitLostPackets(); 418 SetLossDetectionAlarm(); 420 3.9. Detecting Lost Packets 422 Packets in QUIC are only considered lost once a larger packet number 423 is acknowledged. DetectLostPackets is called every time there is a 424 new largest packet or if the loss detection alarm fires the previous 425 largest acked packet is supplied. 427 DetectLostPackets takes one parameter, acked_packet, which is the 428 packet number of the largest acked packet, and returns a list of 429 packet numbers detected as lost. 431 Pseudocode for DetectLostPackets follows: 433 DetectLostPackets(acked_packet): 434 lost_packets = {}; 435 foreach (unacked_packet less than acked_packet): 436 if (unacked_packet.time_sent < 437 acked_packet.time_sent - kTimeReorderThreshold * smoothed_rtt): 438 lost_packets.insert(unacked_packet.packet_number); 439 else if (unacked_packet.packet_number < 440 acked_packet.packet_number - reordering_threshold) 441 lost_packets.insert(unacked_packet.packet_number); 442 return lost_packets; 444 4. Congestion Control 446 (describe NewReno-style congestion control for QUIC.) 448 5. TCP mechanisms in QUIC 450 QUIC implements the spirit of a variety of RFCs, Internet drafts, and 451 other well-known TCP loss recovery mechanisms, though the 452 implementation details differ from the TCP implementations. 454 5.1. RFC 6298 (RTO computation) 456 QUIC calculates SRTT and RTTVAR according to the standard formulas. 457 An RTT sample is only taken if the delayed ack correction is smaller 458 than the measured RTT (otherwise a negative RTT would result), and 459 the ack's contains a new, larger largest observed packet number. 460 min_rtt is only based on the observed RTT, but SRTT uses the delayed 461 ack correction delta. 463 As described above, QUIC implements RTO with the standard timeout and 464 CWND reduction. However, QUIC retransmits the earliest outstanding 465 packets rather than the latest, because QUIC doesn't have 466 retransmission ambiguity. QUIC uses the commonly accepted min RTO of 467 200ms instead of the 1s the RFC specifies. 469 5.2. FACK Loss Recovery (paper) 471 QUIC implements the algorithm for early loss recovery described in 472 the FACK paper (and implemented in the Linux kernel.) QUIC uses the 473 packet number to measure the FACK reordering threshold. Currently 474 QUIC does not implement an adaptive threshold as many TCP 475 implementations (i.e., the Linux kernel) do. 477 5.3. RFC 3782, RFC 6582 (NewReno Fast Recovery) 479 QUIC only reduces its CWND once per congestion window, in keeping 480 with the NewReno RFC. It tracks the largest outstanding packet at 481 the time the loss is declared and any losses which occur before that 482 packet number are considered part of the same loss event. It's worth 483 noting that some TCP implementations may do this on a sequence number 484 basis, and hence consider multiple losses of the same packet a single 485 loss event. 487 5.4. TLP (draft) 489 QUIC always sends two tail loss probes before RTO is triggered. QUIC 490 invokes tail loss probe even when a loss is outstanding, which is 491 different than some TCP implementations. 493 5.5. RFC 5827 (Early Retransmit) with Delay Timer 495 QUIC implements early retransmit with a timer in order to minimize 496 spurious retransmits. The timer is set to 1/4 SRTT after the final 497 outstanding packet is acked. 499 5.6. RFC 5827 (F-RTO) 501 QUIC implements F-RTO by not reducing the CWND and SSThresh until a 502 subsequent ack is received and it's sure the RTO was not spurious. 503 Conceptually this is similar, but it makes for a much cleaner 504 implementation with fewer edge cases. 506 5.7. RFC 6937 (Proportional Rate Reduction) 508 PRR-SSRB is implemented by QUIC in the epoch when recovering from a 509 loss. 511 5.8. TCP Cubic (draft) with optional RFC 5681 (Reno) 513 TCP Cubic is the default congestion control algorithm in QUIC. Reno 514 is also an easily available option which may be requested via 515 connection options and is fully implemented. 517 5.9. Hybrid Slow Start (paper) 519 QUIC implements hybrid slow start, but disables ack train detection, 520 because it has shown to falsely trigger when coupled with packet 521 pacing, which is also on by default in QUIC. Currently the minimum 522 delay increase is 4ms, the maximum is 16ms, and within that range 523 QUIC exits slow start if the min_rtt within a round increases by more 524 than one eighth of the connection mi 526 5.10. RACK (draft) 528 QUIC's loss detection is by it's time-ordered nature, very similar to 529 RACK. Though QUIC defaults to loss detection based on reordering 530 threshold in packets, it could just as easily be based on fractions 531 of an rtt, as RACK does. 533 6. IANA Considerations 535 This document has no IANA actions. Yet. 537 7. Normative References 539 [QUIC-TLS] 540 Thomson, M., Ed. and S. Turner, Ed, Ed., "Using Transport 541 Layer Security (TLS) to Secure QUIC". 543 [QUIC-TRANSPORT] 544 Iyengar, J., Ed. and M. Thomson, Ed., "QUIC: A UDP-Based 545 Multiplexed and Secure Transport". 547 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 548 Requirement Levels", BCP 14, RFC 2119, 549 DOI 10.17487/RFC2119, March 1997, 550 . 552 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 553 "Computing TCP's Retransmission Timer", RFC 6298, 554 DOI 10.17487/RFC6298, June 2011, 555 . 557 Appendix A. Acknowledgments 559 Appendix B. Change Log 561 *RFC Editor's Note:* Please remove this section prior to 562 publication of a final version of this document. 564 B.1. Since draft-ietf-quic-recovery-00: 566 o Improved description of constants and ACK behavior 568 B.2. Since draft-iyengar-quic-loss-recovery-01: 570 o Adopted as base for draft-ietf-quic-recovery. 572 o Updated authors/editors list. 574 o Added table of contents. 576 Authors' Addresses 578 Jana Iyengar (editor) 579 Google 581 Email: jri@google.com 583 Ian Swett (editor) 584 Google 586 Email: ianswett@google.com