idnits 2.17.1 draft-ietf-ice-trickle-05.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 (January 31, 2017) is 2642 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) == Outdated reference: A later version (-18) exists of draft-ietf-mmusic-trickle-ice-sip-06 -- Obsolete informational reference (is this intentional?): RFC 4566 (Obsoleted by RFC 8866) -- Obsolete informational reference (is this intentional?): RFC 5389 (Obsoleted by RFC 8489) -- Obsolete informational reference (is this intentional?): RFC 5766 (Obsoleted by RFC 8656) Summary: 0 errors (**), 0 flaws (~~), 2 warnings (==), 4 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group E. Ivov 3 Internet-Draft Atlassian 4 Intended status: Standards Track E. Rescorla 5 Expires: August 4, 2017 RTFM, Inc. 6 J. Uberti 7 Google 8 P. Saint-Andre 9 Filament 10 January 31, 2017 12 Trickle ICE: Incremental Provisioning of Candidates for the Interactive 13 Connectivity Establishment (ICE) Protocol 14 draft-ietf-ice-trickle-05 16 Abstract 18 This document describes "Trickle ICE", an extension to the 19 Interactive Connectivity Establishment (ICE) protocol that enables 20 ICE agents to send and receive candidates incrementally rather than 21 exchanging complete lists. With such incremental provisioning, ICE 22 agents can begin connectivity checks while they are still gathering 23 candidates and considerably shorten the time necessary for ICE 24 processing to complete. 26 Status of This Memo 28 This Internet-Draft is submitted in full conformance with the 29 provisions of BCP 78 and BCP 79. 31 Internet-Drafts are working documents of the Internet Engineering 32 Task Force (IETF). Note that other groups may also distribute 33 working documents as Internet-Drafts. The list of current Internet- 34 Drafts is at http://datatracker.ietf.org/drafts/current/. 36 Internet-Drafts are draft documents valid for a maximum of six months 37 and may be updated, replaced, or obsoleted by other documents at any 38 time. It is inappropriate to use Internet-Drafts as reference 39 material or to cite them other than as "work in progress." 41 This Internet-Draft will expire on August 4, 2017. 43 Copyright Notice 45 Copyright (c) 2017 IETF Trust and the persons identified as the 46 document authors. All rights reserved. 48 This document is subject to BCP 78 and the IETF Trust's Legal 49 Provisions Relating to IETF Documents 50 (http://trustee.ietf.org/license-info) in effect on the date of 51 publication of this document. Please review these documents 52 carefully, as they describe your rights and restrictions with respect 53 to this document. Code Components extracted from this document must 54 include Simplified BSD License text as described in Section 4.e of 55 the Trust Legal Provisions and are provided without warranty as 56 described in the Simplified BSD License. 58 Table of Contents 60 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 61 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4 62 3. Determining Support for Trickle ICE . . . . . . . . . . . . . 5 63 4. Sending the Initial ICE Description . . . . . . . . . . . . . 6 64 5. Receiving the Initial ICE Description . . . . . . . . . . . . 7 65 5.1. Sending the Initial Response . . . . . . . . . . . . . . 7 66 5.2. Forming Check Lists and Beginning Connectivity 67 Checks . . . . . . . . . . . . . . . . . . . . . . . . . 7 68 6. Receiving the Initial Answer . . . . . . . . . . . . . . . . 8 69 7. Performing Connectivity Checks . . . . . . . . . . . . . . . 8 70 7.1. Scheduling Checks . . . . . . . . . . . . . . . . . . . . 8 71 7.2. Check List and Timer State Updates . . . . . . . . . . . 9 72 8. Discovering and Sending Additional Local Candidates . . . . . 9 73 8.1. Pairing Newly Learned Candidates and Updating 74 Check Lists . . . . . . . . . . . . . . . . . . . . . . . 11 75 8.2. Announcing End of Candidates . . . . . . . . . . . . . . 12 76 9. Receiving Additional Remote Candidates . . . . . . . . . . . 14 77 10. Receiving an End-Of-Candidates Notification . . . . . . . . . 14 78 11. Trickle ICE and Peer Reflexive Candidates . . . . . . . . . . 14 79 12. Concluding ICE Processing . . . . . . . . . . . . . . . . . . 15 80 13. Subsequent Exchanges . . . . . . . . . . . . . . . . . . . . 15 81 14. Unilateral Use of Trickle ICE (Half Trickle) . . . . . . . . 15 82 15. Requirements for Signaling Protocols . . . . . . . . . . . . 16 83 16. Example Flow . . . . . . . . . . . . . . . . . . . . . . . . 16 84 17. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 17 85 18. Security Considerations . . . . . . . . . . . . . . . . . . . 17 86 19. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 17 87 20. References . . . . . . . . . . . . . . . . . . . . . . . . . 17 88 20.1. Normative References . . . . . . . . . . . . . . . . . . 18 89 20.2. Informative References . . . . . . . . . . . . . . . . . 18 90 Appendix A. Interaction with Regular ICE . . . . . . . . . . . . 19 91 Appendix B. Interaction with ICE Lite . . . . . . . . . . . . . 20 92 Appendix C. Preserving Candidate Order while Trickling . . . . . 21 93 Appendix D. Changes from Earlier Versions . . . . . . . . . . . 22 94 D.1. Changes from draft-ietf-ice-trickle-04 . . . . . . . . . 22 95 D.2. Changes from draft-ietf-ice-trickle-03 . . . . . . . . . 23 96 D.3. Changes from draft-ietf-ice-trickle-03 . . . . . . . . . 23 97 D.4. Changes from draft-ietf-ice-trickle-02 . . . . . . . . . 23 98 D.5. Changes from draft-ietf-ice-trickle-01 . . . . . . . . . 23 99 D.6. Changes from draft-ietf-ice-trickle-00 . . . . . . . . . 23 100 D.7. Changes from draft-mmusic-trickle-ice-02 . . . . . . . . 23 101 D.8. Changes from draft-ivov-01 and draft-mmusic-00 . . . . . 24 102 D.9. Changes from draft-ivov-00 . . . . . . . . . . . . . . . 24 103 D.10. Changes from draft-rescorla-01 . . . . . . . . . . . . . 25 104 D.11. Changes from draft-rescorla-00 . . . . . . . . . . . . . 26 105 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 26 107 1. Introduction 109 The Interactive Connectivity Establishment (ICE) protocol 110 [rfc5245bis] describes mechanisms for gathering candidates, 111 prioritizing them, choosing default ones, exchanging them with a 112 remote party, pairing them, and ordering them into check lists. Once 113 all of these actions have been completed (and only then), the parties 114 can begin a phase of connectivity checks and eventually select the 115 pair of candidates that will be used in a media session or for a 116 given media stream. 118 Although the sequence described above has the advantage of being 119 relatively straightforward to implement and debug once deployed, it 120 can also be rather lengthy. Candidate gathering often involves 121 things like querying STUN [RFC5389] servers and allocating relayed 122 candidates at TURN [RFC5766] servers. All of these actions can be 123 delayed for a noticeable amount of time; although they can be run in 124 parallel, they still need to respect the pacing requirements from 125 [rfc5245bis], which is likely to delay them even further. Some or 126 all of these actions also need be completed by the remote agent. 127 Both agents would next perform connectivity checks and only then 128 would they be ready to begin streaming media. 130 These factors can lead to relatively lengthy session establishment 131 times and thus to a degraded user experience. 133 This document defines an alternative or supplementary mode of 134 operation for ICE implementations, known as "Trickle ICE", in which 135 candidates can be exchanged incrementally. This enables ICE agents 136 to exchange candidates as soon as an ICE negotiation session has been 137 initiated. Connectivity checks for a media stream can also start as 138 soon as the first candidates for that stream become available. 140 Trickle ICE can reduce session establishment times in cases where 141 connectivity is confirmed for the first exchanged candidates (e.g., 142 where candidates for one of the agents are directly reachable from 143 the second agent, such as candidates at a media relay). Even when 144 this is not the case, performing candidate gathering for both agents 145 and connectivity checks in parallel can considerably shorten ICE 146 processing times. 148 It is worth noting that there is quite a bit of operational 149 experience with the Trickle ICE technique, going back as far as 2005 150 (when the XMPP Jingle extension defined a "dribble mode" as specified 151 in [XEP-0176]); this document incorporates feedback from those who 152 have implemented and deployed the technique. 154 In addition to the basics of Trickle ICE, this document also 155 describes how to discover support for Trickle ICE, how regular ICE 156 processing needs to be modified when building and updating check 157 lists, and how Trickle ICE implementations interoperate with agents 158 that only implement regular ICE processing as defined in 159 [rfc5245bis]. 161 This specification does not define the usage of Trickle ICE with any 162 specific signaling protocol (however, see 163 [I-D.ietf-mmusic-trickle-ice-sip] for usage with SIP [RFC3261] and 164 [XEP-0176] for usage with XMPP [RFC6120]). Similarly, it does not 165 define Trickle ICE in terms of the Session Description Protocol (SDP) 166 [RFC4566] or the offer/answer model [RFC3264] because the technique 167 can be and already is used in application protocols that are not tied 168 to SDP or to offer/answer semantics. However, because SDP and the 169 offer/answer model are familiar to most readers of this 170 specification, some examples in this document use those particulars 171 in order to explain the underlying concepts. 173 2. Terminology 175 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 176 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 177 document are to be interpreted as described in [RFC2119]. 179 This specification makes use of all terminology defined for 180 Interactive Connectivity Establishment in [rfc5245bis]. In addition, 181 it defines the following terms: 183 Candidate Gatherer: A module used by an ICE agent to obtain local 184 candidates. Candidate gatherers use different mechanisms for 185 discovering local candidates, such as STUN and TURN. 187 Generation: The complete set of candidates sent within an ICE 188 negotiation session. 190 ICE Description: Any session-related (as opposed to candidate- 191 related) attributes required to configure an ICE agent. These 192 include but are not limited to "ice-ufrag", "ice-pwd", and "ice- 193 options". 195 ICE Negotiation Session: A virtual session involving all of the 196 interactions between ICE agents up until an ICE restart (if any). 198 Initiator: The ICE agent that starts an ICE negotiation session. 200 Responder: The ICE agent with which an initiator starts an ICE 201 negotiation session. 203 Trickled Candidates: Candidates that a Trickle ICE agent sends after 204 sending an initial ICE description or responding to an initial ICE 205 description, but within the same ICE negotiation session. 206 Trickled candidates can be sent in parallel with candidate 207 gathering and connectivity checks. 209 Trickling: The act of sending trickled candidates. 211 Half Trickle: A Trickle ICE mode of operation where the initiator 212 gathers its first generation of candidates strictly before 213 creating and sending the initial ICE description. Once sent, that 214 ICE description can be processed by regular ICE agents and does 215 not require support for this specification. It also allows 216 Trickle ICE capable responders to still gather candidates and 217 perform connectivity checks in a non-blocking way, thus roughly 218 providing "half" the advantages of Trickle ICE. The mechanism is 219 mostly meant for use in cases where the remote agent's support for 220 Trickle ICE cannot be confirmed prior to sending an initial ICE 221 description. 223 Full Trickle: The typical mode of operation for Trickle ICE agents, 224 in which an initial ICE description can include any number of 225 candidates (even zero candidates) and does not need to include the 226 entire first generation of candidates as in half trickle. 228 3. Determining Support for Trickle ICE 230 To fully support Trickle ICE, applications SHOULD incorporate one of 231 the following mechanisms to enable implementations to determine 232 whether Trickle ICE is supported: 234 1. Provide a capabilities discovery method so that agents can verify 235 support of Trickle ICE prior to initiating a session (XMPP's 236 Service Discovery [XEP-0030] is one such mechanism). 238 2. Make support for Trickle ICE mandatory so that user agents can 239 assume support. 241 If an application protocol does not provide a method of determining 242 ahead of time whether Trickle ICE is supported, agents can make use 243 of the half trickle procedure described in Section 14. 245 Prior to sending an initial ICE description, agents using signaling 246 protocols that support capabilities discovery can attempt to verify 247 whether or not the remote party supports Trickle ICE. If an agent 248 determines that the remote party does not support Trickle ICE, it 249 MUST fall back to using regular ICE or abandon the entire session. 251 Even if a signaling protocol does not include a capabilities 252 discovery method, a user agent can provide an indication within the 253 ICE description that it supports Trickle ICE (e.g., in SDP this would 254 be a token of "trickle" in the ice-options attribute). 256 Dedicated discovery semantics and half trickle are needed only prior 257 to session initiation. After a session is established and Trickle 258 ICE support is confirmed for both parties, either agent can use full 259 trickle for subsequent exchanges. 261 4. Sending the Initial ICE Description 263 An agent can start gathering candidates as soon as it has an 264 indication that communication is imminent (e.g., a user interface cue 265 or an explicit request to initiate a session). Unlike in regular 266 ICE, in Trickle ICE implementations do not need to gather candidates 267 in a blocking manner. Therefore, unless half trickle is being used, 268 agents SHOULD generate and transmit their initial ICE description as 269 early as possible, so that the remote party can start gathering and 270 trickling candidates. 272 Trickle ICE agents MAY include any mix of candidates in an ICE 273 description. This includes the possibility of sending an ICE 274 description that contains all the candidates that the agent plans to 275 use (as in half trickle mode), sending an ICE description that 276 contains only a publicly-reachable IP address (e.g., a candidate at a 277 media relay that is known to not be behind a firewall), or sending an 278 ICE description with no candidates at all (in which case the 279 initiator can obtain the responder's initial candidate list sooner 280 and the responder can begin candidate gathering more quickly). 282 Methods for calculating priorities and foundations, as well as 283 determining redundancy of candidates, work just as with regular ICE 284 (with the exception of pruning of duplicate peer reflexive candidates 285 as described under Section 5.2). 287 5. Receiving the Initial ICE Description 289 When a responder receives an initial ICE description, it will first 290 check if the ICE description or initiator indicates support for 291 Trickle ICE as explained in Section 3. If this is not the case, the 292 agent MUST process the ICE description according to regular ICE 293 procedures [rfc5245bis] (or, if no ICE support is detected at all, 294 according to relevant processing rules for the underlying signaling 295 protocol, such as offer/answer processing rules [RFC3264]). 297 If support for Trickle ICE is confirmed, an agent will automatically 298 assume support for regular ICE as well even if the support 299 verification procedure in [rfc5245bis] indicates otherwise. 300 Specifically, the rules from [rfc5245bis] would imply that ICE itself 301 is not supported if the initial ICE description includes no 302 candidates; however, such a conclusion is not warranted if the 303 responder can confirm that the initiator supports Trickle ICE; in 304 this case, fallback to [RFC3264] is not necessary. 306 If the initial ICE description does indicate support for Trickle ICE, 307 the agent will determine its role and start gathering and 308 prioritizing candidates; while doing so, it will also respond by 309 sending its own ICE description, so that both agents can start 310 forming check lists and begin connectivity checks. 312 5.1. Sending the Initial Response 314 An agent can respond to an initial ICE description at any point while 315 gathering candidates. Here again the ICE description MAY contain any 316 set of candidates, including all candidates or no candidates. (The 317 benefit of including no candidates is to send the ICE description as 318 quickly as possible, so that both parties can consider the overall 319 session to be under active negotiation as soon as possible.) 321 As noted in Section 3, in application protocols that use SDP the 322 responder's ICE description can indicate support for Trickle ICE by 323 including a token of "trickle" in the ice-options attribute. 325 5.2. Forming Check Lists and Beginning Connectivity Checks 327 After the initiator and responder exchange ICE descriptions, and as 328 soon as they have obtained local and remote candidates, agents begin 329 forming candidate pairs, computing candidate pair priorities, 330 ordering candidate pairs, pruning duplicate pairs, and creating check 331 lists according to regular ICE procedures [rfc5245bis]. 333 According to those procedures, in order for candidate pairing to be 334 possible and for duplicate candidates to be pruned, the candidates 335 would need to be provided in the relevant ICE descriptions. Under 336 Trickle ICE, check lists can be empty until candidate pairs are sent 337 or received. Therefore Trickle ICE agents handle check lists and 338 candidate pairing in a slightly different way than regular ICE 339 agents: the agents still create the check lists, but they populate 340 the check lists only after they actually have the candidate pairs. 342 A Trickle ICE agent initially considers all check lists to be frozen. 343 It then inspects the first check list and attempts to unfreeze all 344 candidate pairs it has received so far that belong to the first 345 component on the first media stream (i.e., the first media stream 346 that was reported to the ICE implementation from the using 347 application). If that first component of the first media stream does 348 not contain candidates for one or more of the currently known pair 349 foundations, and if candidate pairs already exist for that foundation 350 in one of the following components or media streams, then the agent 351 unfreezes the first of those candidate pairs. 353 With regard to pruning of duplicate candidate pairs, a Trickle ICE 354 agent SHOULD follow a policy of "highest priority wins, except for 355 peer reflexive candidates". 357 6. Receiving the Initial Answer 359 When processing an ICE description from a responder, the initiator 360 follows regular ICE procedures to determine its role, after which it 361 forms check lists (as described in Section 5.2) and begins 362 connectivity checks. 364 7. Performing Connectivity Checks 366 For the most part, Trickle ICE agents perform connectivity checks 367 following regular ICE procedures. However, the fact that gathering 368 and communicating candidates is asynchronous in Trickle ICE imposes a 369 number of changes as described in the following sections. 371 7.1. Scheduling Checks 373 The ICE specification [rfc5245bis], Section 5.8, requires that agents 374 terminate the timer for a triggered check in relation to an active 375 check list once the agent has exhausted all frozen pairs in the check 376 list. This will not work with Trickle ICE, because more pairs will 377 be added to the check list incrementally. 379 Therefore, a Trickle ICE agent SHOULD NOT terminate the timer until 380 the state of the check list is Completed or Failed as specified 381 herein (see Section 8.2). 383 7.2. Check List and Timer State Updates 385 The ICE specification [rfc5245bis], Section 7.1.3.3, requires that 386 agents update check lists and timer states upon completing a 387 connectivity check transaction. During such an update, regular ICE 388 agents would set the state of a check list to Failed if both of the 389 following two conditions are satisfied: 391 o all of the pairs in the check list are either in the Failed state 392 or Succeeded state; and 394 o there is not a pair in the valid list for each component of the 395 media stream. 397 With Trickle ICE, the above situation would often occur when 398 candidate gathering and trickling are still in progress, even though 399 it is quite possible that future checks will succeed. For this 400 reason, Trickle ICE agents add the following conditions to the above 401 list: 403 o all candidate gatherers have completed and the agent is not 404 expecting to discover any new local candidates; 406 o the remote agent has sent an end-of-candidates indication for that 407 check list as described in Section 8.2. 409 Regular ICE requires that agents then update all other check lists, 410 placing one pair from each of them into the Waiting state, 411 effectively unfreezing all remaining check lists. However, under 412 Trickle ICE other check lists might still be empty at that point. 413 Therefore a Trickle ICE agent MUST monitor whether a check list is 414 active or frozen independently of the state of the candidate pairs 415 that the check list contains, and MUST consider a check list to be 416 active when unfreezing the first candidate pair in the check list. 417 When there is no candidate pair in a check list (i.e., when the check 418 list is empty), a Trickle ICE agent MAY consider it to be either 419 active or frozen. An empty frozen check list SHOULD be changed to 420 active if another check list is completely finished (i.e., every pair 421 is either Successful or Failed), or if another checklist has a valid 422 candidate pair for all components. 424 8. Discovering and Sending Additional Local Candidates 426 After ICE descriptions have been sent, agents will most likely 427 continue discovering new local candidates as STUN, TURN, and other 428 non-host candidate gathering mechanisms begin to yield results. 429 Whenever an agent discovers such a new candidate it will compute its 430 priority, type, foundation and component ID according to regular ICE 431 procedures. 433 The new candidate is then checked for redundancy against the existing 434 list of local candidates. If its transport address and base match 435 those of an existing candidate, it will be considered redundant and 436 will be ignored. This would often happen for server reflexive 437 candidates that match the host addresses they were obtained from 438 (e.g., when the latter are public IPv4 addresses). Contrary to 439 regular ICE, Trickle ICE agents will consider the new candidate 440 redundant regardless of its priority. 442 Next the agent sends (i.e., trickles) the newly discovered 443 candidate(s) to the remote agent. The actual delivery of the new 444 candidates is handled by a signaling protocol such as SIP or XMPP. 445 Trickle ICE imposes no restrictions on the way this is done (e.g., 446 some applications may choose not to send trickle updates for server 447 reflexive candidates and instead rely on the discovery of peer 448 reflexive ones). 450 When trickle updates are sent, each candidate MUST be delivered to 451 the receiving Trickle ICE implementation not more than once. If 452 there are any candidate retransmissions, they need to be hidden from 453 the ICE implementation. 455 Also, candidate trickling needs to be correlated to a specific ICE 456 negotiation session, so that if there is an ICE restart, any delayed 457 updates for a previous session can be recognized as such and ignored 458 by the receiving party. For example, applications that choose to 459 signal candidates via SDP may include a ufrag value in the 460 corresponding a=candidate line such as: 462 a=candidate:1 1 UDP 2130706431 2001:db8::1 5000 typ host ufrag 8hhY 464 Or as another example, WebRTC implementations may include a ufrag in 465 the JavaScript objects that represent candidates. 467 Note: The signaling protocol needs to provide a mechanism for both 468 parties to indicate and agree on the ICE negotiation session in force 469 (as identified by the ufrag) so that they have a consistent view of 470 which candidates are to be paired. This is especially important in 471 the case of ICE restarts (see Section 13). 473 Once the candidate has been sent to the remote party, the agent 474 checks if any remote candidates are currently known for this same 475 stream. If not, the new candidate will simply be added to the list 476 of local candidates. 478 Otherwise, if the agent has already learned of one or more remote 479 candidates for this stream and component, it will begin pairing the 480 new local candidates with them and adding the pairs to the existing 481 check lists according to their priority. 483 Note: A Trickle ICE agent MUST NOT pair a local candidate until it 484 has been trickled to the remote agent. 486 8.1. Pairing Newly Learned Candidates and Updating Check Lists 488 Forming candidate pairs works as described in the ICE specification 489 [rfc5245bis]. However, actually adding the new pair to a check list 490 happens according to the rules described below. 492 If the check list where the pair is to be added already contains the 493 maximum number of candidate pairs (100 by default as per 494 [rfc5245bis]), the new pair is discarded. 496 If the new pair's local candidate is server reflexive, the server 497 reflexive candidate MUST be replaced by its base before adding the 498 pair to the list. 500 Once this is done, the agent examines the check list looking for 501 another pair that would be redundant with the new one. If such a 502 pair exists and the type of its remote candidate is not peer 503 reflexive, the pair with the higher priority is kept and the one with 504 the lower priority is discarded. If, on the other hand, the type of 505 the remote candidate in the pre-existing pair is peer reflexive, the 506 agent MUST replace it with the new candidate it received (regardless 507 of their respective priorities); this is done by setting the priority 508 of the new candidate to the priority of the pre-existing candidate 509 and then re-sorting the check list. 511 Note: So that both agents will have the same view of candidate 512 priorities, it is important to replacing existing pairs with 513 seemingly equivalent higher-priority ones and to always update 514 peer-reflexive candidates if equivalent alternatives are received 515 through signaling. 517 For all other pairs, including those with a server reflexive local 518 candidate that were not found to be redundant: 520 o if this check list is frozen, then the new pair will be assigned a 521 state of Frozen. 523 o else if the check list is active and it is either empty or 524 contains only candidates in the Succeeded and Failed states, then 525 the new pair's state is set to Waiting. 527 o else if the check list is non-empty and active, then the state of 528 the new pair will be set as follows: 530 Frozen: if there is at least one pair in any check list whose 531 foundation matches the one in the new pair, whose state is 532 neither Succeeded nor Failed (eventually the new pair will get 533 unfrozen after the ongoing check for the existing pair 534 concludes), and whose component ID precedes the one of the new 535 pair; 537 Waiting: if the list contains no pairs with the same foundation 538 as the new one, or in case such pairs exist but they are all in 539 either the Succeeded or Failed state or the new pair's 540 component ID precedes theirs. 542 8.2. Announcing End of Candidates 544 Once all candidate gathering is completed or expires for a specific 545 media stream, the agent will generate an "end-of-candidates" 546 indication for that stream and send it to the remote agent via the 547 signaling channel. The exact form of the indication depends on the 548 application protocol. The indication can be sent in the following 549 ways: 551 o As part of an initiation request (which would typically be the 552 case with an initial ICE description for half trickle) 554 o Along with the last candidate an agent can send for a stream 556 o As a standalone notification (e.g., after STUN Binding requests or 557 TURN Allocate requests to a server time out and the agent has no 558 other active gatherers) 560 Sending an end-of-candidates indication in a timely manner is 561 important in order to avoid ambiguities and speed up the conclusion 562 of ICE processing. In particular: 564 o A controlled Trickle ICE agent SHOULD send an end-of-candidates 565 indication after it has completed gathering for a media stream, 566 unless ICE processing terminates before the agent has had a chance 567 to complete gathering. 569 o A controlling agent MAY conclude ICE processing prior to sending 570 end-of-candidates indications for all streams. However, it is 571 RECOMMENDED for a controlling agent to send end-of-candidates 572 indications whenever possible for the sake of consistency and to 573 keep middleboxes and controlled agents up-to-date on the state of 574 ICE processing. 576 When sending an end-of-candidates indication during trickling (rather 577 than as a part of an initial ICE description or response), it is the 578 responsibility of the using protocol to define methods for relating 579 the indication to one or more specific media streams. 581 Receiving an end-of-candidates indication enables an agent to update 582 check list states and, in case valid pairs do not exist for every 583 component in every media stream, determine that ICE processing has 584 failed. It also enables agents to speed up the conclusion of ICE 585 processing when a candidate pair has been validated but it involves 586 the use of lower-preference transports such as TURN. In such 587 situations, an implementation MAY choose to wait and see if higher- 588 priority candidates are received; in this case the end-of-candidates 589 indication provides a notification that such candidates are not 590 forthcoming. 592 An agent MAY also choose to generate an end-of-candidates indication 593 before candidate gathering has actually completed, if the agent 594 determines that gathering has continued for more than an acceptable 595 period of time. However, an agent MUST NOT send any more candidates 596 after it has sent an end-of-candidates indication. 598 When performing half trickle, an agent SHOULD send an end-of- 599 candidates indication together with its initial ICE description 600 unless it is planning to potentially send additional candidates 601 (e.g., in case the remote party turns out to support Trickle ICE). 603 After an agent sends the end-of-candidates indication, it will update 604 the state of the corresponding check list as explained in 605 Section 7.2. Past that point, an agent MUST NOT send any new 606 candidates within this ICE negotiation session. After an agent has 607 received an end-of-candidates indication, it MUST also ignore any 608 newly received candidates for that media stream or media session. 609 Therefore, adding new candidates to the negotiation is possible only 610 through an ICE restart (see Section 13). 612 This specification does not override regular ICE semantics for 613 concluding ICE processing. Therefore, even if end-of-candidates 614 indications are sent, agents will still have to go through pair 615 nomination. Also, if pairs have been nominated for components and 616 media streams, ICE processing MAY still conclude even if end-of- 617 candidates indications have not been received for all streams. 619 9. Receiving Additional Remote Candidates 621 At any time during ICE processing, a Trickle ICE agent might receive 622 new candidates from the remote agent. When this happens and no local 623 candidates are currently known for this same stream, the new remote 624 candidates are added to the list of remote candidates. 626 Otherwise, the new candidates are used for forming candidate pairs 627 with the pool of local candidates and they are added to the local 628 check lists as described in Section 8.1. 630 Once the remote agent has completed candidate gathering, it will send 631 an end-of-candidates indication. Upon receiving such an indication, 632 the local agent MUST update check list states as per Section 7.2. 633 This might lead to some check lists being marked as Failed. 635 10. Receiving an End-Of-Candidates Notification 637 When an agent receives an end-of-candidates indication for a specific 638 media stream, it will update the state of the relevant check list as 639 per Section 7.2. If the check list is still in the Active state 640 after the update, the agent will persist the fact that an end-of- 641 candidates indication has been received and take it into account in 642 future updates to the check list. 644 11. Trickle ICE and Peer Reflexive Candidates 646 Even though Trickle ICE does not explicitly modify the procedures for 647 handling peer-reflexive candidates, use of Trickle ICE can have an 648 impact on how they are processed. With Trickle ICE, it is possible 649 that server reflexive candidates can be discovered as peer reflexive 650 in cases where incoming connectivity checks are received from these 651 candidates before the trickle updates that carry them. 653 While this would certainly increase the number of cases where ICE 654 processing nominates and selects candidates discovered as peer- 655 reflexive, it does not require any change in processing. 657 It is also likely that some applications would prefer not to trickle 658 server reflexive candidates to entities that are known to be publicly 659 accessible and where sending a direct STUN binding request is likely 660 to reach the destination faster than the trickle update that travels 661 through the signaling path. 663 12. Concluding ICE Processing 665 This specification does not directly modify the procedures for ending 666 ICE processing described in Section 8 of [rfc5245bis], and Trickle 667 ICE implementations follow the same rules. 669 13. Subsequent Exchanges 671 Either agent MAY generate a subsequent ICE description at any time 672 allowed by [RFC3264]. When this happens agents will use [rfc5245bis] 673 semantics to determine whether or not the new ICE description 674 requires an ICE restart. If an ICE restart occurs, the user agents 675 can assume that Trickle ICE is still supported if support was 676 determined previously, and thus can engage in Trickle ICE behavior as 677 they would in an initial exchange of ICE descriptions where support 678 was determined through a capabilities discovery method. 680 14. Unilateral Use of Trickle ICE (Half Trickle) 682 In half trickle mode, the initiator sends a regular ICE description 683 with a complete generation of candidates. This ensures that the ICE 684 description can be processed by a regular ICE responder and is mostly 685 meant for use in cases where support for Trickle ICE cannot be 686 confirmed prior to sending an initial ICE description. The initial 687 ICE description indicates support for Trickle ICE, which means the 688 responder can respond with an incomplete generation of candidates and 689 continue trickling the rest. A half trickle ICE description would 690 typically contain an end-of-candidates indication, although this is 691 not mandatory because if trickle support is confirmed then the 692 initiator can choose to trickle additional candidates before it sends 693 an end-of-candidates indication. 695 The half trickle mechanism can be used in cases where there is no way 696 for an agent to verify in advance whether a remote party supports 697 Trickle ICE. Because the initial ICE description contains a full 698 generation of candidates, it can thus be handled by a regular ICE 699 agent, while still allowing a Trickle ICE agent to use the 700 optimization defined in this specification. This prevents 701 negotiation from failing in the former case while still giving 702 roughly half the Trickle ICE benefits in the latter (hence the name 703 of the mechanism). 705 Use of half trickle is only necessary during an initial exchange of 706 ICE descriptions. After both parties have received a session 707 description from their peer, they can each reliably determine Trickle 708 ICE support and use it for all subsequent exchanges. 710 In some instances, using half trickle might bring more than just half 711 the improvement in terms of user experience. This can happen when an 712 agent starts gathering candidates upon user interface cues that the 713 user will soon be initiating an interaction, such as activity on a 714 keypad or the phone going off hook. This would mean that some or all 715 of the candidate gathering could be completed before the agent 716 actually needs to send the ICE description. Because the responder 717 will be able to trickle candidates, both agents will be able to start 718 connectivity checks and complete ICE processing earlier than with 719 regular ICE and potentially even as early as with full trickle. 721 However, such anticipation is not always possible. For example, a 722 multipurpose user agent or a WebRTC web page where communication is a 723 non-central feature (e.g., calling a support line in case of a 724 problem with the main features) would not necessarily have a way of 725 distinguishing between call intentions and other user activity. In 726 such cases, using full trickle is most likely to result in an ideal 727 user experience. Even so, using half trickle would be an improvement 728 over regular ICE because it would result in a better experience for 729 responders. 731 15. Requirements for Signaling Protocols 733 In order to fully enable the use of Trickle ICE, this specification 734 defines the following requirements for signaling protocols. 736 o A signaling protocol SHOULD provide a way for parties to advertise 737 and discover support for Trickle ICE before an ICE negotiation 738 session begins (see Section 3). 740 o A signaling protocol MUST provide methods for incrementally 741 sending (i.e., "trickling") additional candidates after sending 742 the initial ICE description (see Section 8). 744 o A signaling protocol MUST provide a mechanism for both parties to 745 indicate and agree on the ICE negotiation session in force (see 746 Section 8). 748 o A signaling protocol MUST provide a way for parties to communicate 749 the end-of-candidates indication (see Section 8.2). 751 16. Example Flow 753 As an example, a typical successful Trickle ICE exchange with a 754 signaling protocol that follows the offer/answer model would look 755 this way: 757 Alice Bob 758 | Offer | 759 |---------------------------------------------->| 760 | Additional Candidates | 761 |---------------------------------------------->| 762 | | 763 | Answer | 764 |<----------------------------------------------| 765 | Additional Candidates | 766 |<----------------------------------------------| 767 | | 768 | Additional Candidates and Connectivity Checks | 769 |<--------------------------------------------->| 770 | | 771 |<=============== MEDIA FLOWS =================>| 773 Figure 1: Example 775 17. IANA Considerations 777 This specification requests no actions from IANA. 779 18. Security Considerations 781 This specification inherits most of its semantics from [rfc5245bis] 782 and as a result all security considerations described there apply to 783 Trickle ICE. 785 If the privacy implications of revealing host addresses on an 786 endpoint device are a concern, agents can generate an ICE description 787 that contains no candidates and then only trickle candidates that do 788 not reveal host addresses (e.g., relayed candidates). 790 19. Acknowledgements 792 The authors would like to thank Bernard Aboba, Flemming Andreasen, 793 Rajmohan Banavi, Taylor Brandstetter, Christer Holmberg, Jonathan 794 Lennox, Enrico Marocco, Pal Martinsen, Martin Thomson, Dale R. 795 Worley, and Brandon Williams for their reviews and suggestions on 796 improving this document. 798 20. References 799 20.1. Normative References 801 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 802 Requirement Levels", BCP 14, RFC 2119, March 1997. 804 [rfc5245bis] 805 Keranen, A., Keranen, A., and J. Rosenberg, "Interactive 806 Connectivity Establishment (ICE): A Protocol for Network 807 Address Translator (NAT) Traversal", draft-ietf-ice- 808 rfc5245bis-08 (work in progress), December 2016. 810 20.2. Informative References 812 [I-D.ietf-mmusic-trickle-ice-sip] 813 Ivov, E., Thomas, T., Marocco, E., and C. Holmberg, "A 814 Session Initiation Protocol (SIP) usage for Trickle ICE", 815 draft-ietf-mmusic-trickle-ice-sip-06 (work in progress), 816 October 2016. 818 [RFC1918] Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot, G., 819 and E. Lear, "Address Allocation for Private Internets", 820 BCP 5, RFC 1918, DOI 10.17487/RFC1918, February 1996, 821 . 823 [RFC3261] Rosenberg, J., Schulzrinne, H., Camarillo, G., Johnston, 824 A., Peterson, J., Sparks, R., Handley, M., and E. 825 Schooler, "SIP: Session Initiation Protocol", RFC 3261, 826 June 2002. 828 [RFC3264] Rosenberg, J. and H. Schulzrinne, "An Offer/Answer Model 829 with Session Description Protocol (SDP)", RFC 3264, June 830 2002. 832 [RFC4566] Handley, M., Jacobson, V., and C. Perkins, "SDP: Session 833 Description Protocol", RFC 4566, July 2006. 835 [RFC4787] Audet, F., Ed. and C. Jennings, "Network Address 836 Translation (NAT) Behavioral Requirements for Unicast 837 UDP", BCP 127, RFC 4787, DOI 10.17487/RFC4787, January 838 2007, . 840 [RFC5389] Rosenberg, J., Mahy, R., Matthews, P., and D. Wing, 841 "Session Traversal Utilities for NAT (STUN)", RFC 5389, 842 DOI 10.17487/RFC5389, October 2008, 843 . 845 [RFC5766] Mahy, R., Matthews, P., and J. Rosenberg, "Traversal Using 846 Relays around NAT (TURN): Relay Extensions to Session 847 Traversal Utilities for NAT (STUN)", RFC 5766, April 2010. 849 [RFC6120] Saint-Andre, P., "Extensible Messaging and Presence 850 Protocol (XMPP): Core", RFC 6120, March 2011. 852 [XEP-0030] 853 Hildebrand, J., Millard, P., Eatmon, R., and P. Saint- 854 Andre, "XEP-0030: Service Discovery", XEP XEP-0030, June 855 2008. 857 [XEP-0176] 858 Beda, J., Ludwig, S., Saint-Andre, P., Hildebrand, J., 859 Egan, S., and R. McQueen, "XEP-0176: Jingle ICE-UDP 860 Transport Method", XEP XEP-0176, June 2009. 862 Appendix A. Interaction with Regular ICE 864 The ICE protocol was designed to be flexible enough to work in and 865 adapt to as many network environments as possible. Despite that 866 flexibility, ICE as specified in [rfc5245bis] does not by itself 867 support trickle ICE. This section describes how trickling of 868 candidates interacts with ICE. 870 [rfc5245bis] describes the conditions required to update check lists 871 and timer states while an ICE agent is in the Running state. These 872 conditions are verified upon transaction completion and one of them 873 stipulates that: 875 If there is not a pair in the valid list for each component of the 876 media stream, the state of the check list is set to Failed. 878 This could be a problem and cause ICE processing to fail prematurely 879 in a number of scenarios. Consider the following case: 881 1. Alice and Bob are both located in different networks with Network 882 Address Translation (NAT). Alice and Bob themselves have 883 different address but both networks use the same [RFC1918] block. 885 2. Alice sends Bob the candidate 2001:db8:a0b:12f0::10 which also 886 happens to correspond to an existing host on Bob's network. 888 3. Bob creates a check list consisting solely of 889 2001:db8:a0b:12f0::10 and starts checks. 891 4. These checks reach the host at 2001:db8:a0b:12f0::10 in Bob's 892 network, which responds with an ICMP "port unreachable" error and 893 per [rfc5245bis] Bob marks the transaction as Failed. 895 At this point the check list only contains Failed candidates and the 896 valid list is empty. This causes the media stream and potentially 897 all ICE processing to fail. 899 A similar race condition would occur if the initial ICE description 900 from Alice only contains candidates that can be determined as 901 unreachable from any of the candidates that Bob has gathered (e.g., 902 this would be the case if Bob's candidates only contain IPv4 903 addresses and the first candidate that he receives from Alice is an 904 IPv6 one). 906 Another potential problem could arise when a non-trickle ICE 907 implementation initiates an interaction with a Trickle ICE 908 implementation. Consider the following case: 910 1. Alice's client has a non-Trickle ICE implementation. 912 2. Bob's client has support for Trickle ICE. 914 3. Alice and Bob are behind NATs with address-dependent filtering 915 [RFC4787]. 917 4. Bob has two STUN servers but one of them is currently 918 unreachable. 920 After Bob's agent receives Alice's initial ICE description it would 921 immediately start connectivity checks. It would also start gathering 922 candidates, which would take a long time because of the unreachable 923 STUN server. By the time Bob's answer is ready and sent to Alice, 924 Bob's connectivity checks may well have failed: until Alice gets 925 Bob's answer, she won't be able to start connectivity checks and 926 punch holes in her NAT. The NAT would hence be filtering Bob's 927 checks as originating from an unknown endpoint. 929 Appendix B. Interaction with ICE Lite 931 The behavior of ICE lite agents that are capable of Trickle ICE does 932 not require any particular rules other than those already defined in 933 this specification and [rfc5245bis]. This section is hence provided 934 only for informational purposes. 936 An ICE lite agent would generate an ICE description as per 937 [rfc5245bis] and would indicate support for Trickle ICE. Given that 938 they will contain a complete generation of candidates, these ICE 939 descriptions would also be accompanied by an end-of-candidates 940 indication. 942 When performing full trickle, a full ICE implementation could send an 943 initial ICE description or response with no candidates. After 944 receiving a response that identifies the remote agent as an ICE lite 945 implementation, the initiator can choose to not send any additional 946 candidates. The same is also true in the case when the ICE lite 947 agent initiates the interaction and the full ICE agent is the 948 responder. In these cases the connectivity checks would be enough 949 for the ICE lite implementation to discover all potentially useful 950 candidates as peer reflexive. The following example illustrates one 951 such ICE session using SDP syntax: 953 ICE Lite Bob 954 Agent 955 | Offer (a=ice-lite a=ice-options:trickle) | 956 |---------------------------------------------->| 957 | |no cand 958 | Answer (a=ice-options:trickle) |trickling 959 |<----------------------------------------------| 960 | Connectivity Checks | 961 |<--------------------------------------------->| 962 peer rflx| | 963 cand disco| | 964 | | 965 |<=============== MEDIA FLOWS =================>| 967 Figure 2: Example 969 In addition to reducing signaling traffic this approach also removes 970 the need to discover STUN bindings or make TURN allocations, which 971 may considerably lighten ICE processing. 973 Appendix C. Preserving Candidate Order while Trickling 975 One important aspect of regular ICE is that connectivity checks for a 976 specific foundation and component are attempted simultaneously by 977 both agents, so that any firewalls or NATs fronting the agents would 978 whitelist both endpoints and allow all except for the first 979 ("suicide") packets to go through. This is also important to 980 unfreezing candidates at the right time. While not crucial, 981 preserving this behavior in Trickle ICE is likely to improve ICE 982 performance. 984 To achieve this, when trickling candidates agents MUST respect the 985 order in which the components and streams as they have been 986 negotiated appear (implicitly or explicitly) in the relevant ICE 987 descriptions. Therefore a candidate for a specific component MUST 988 NOT be sent prior to candidates for other components within the same 989 foundation. 991 For example, the following SDP description contains two components 992 (RTP and RTCP) and two foundations (host and server reflexive): 994 v=0 995 o=jdoe 2890844526 2890842807 IN IP6 2001:db8:a0b:12f0::1 996 s= 997 c=IN IP4 2001:db8:a0b:12f0::1 998 t=0 0 999 a=ice-pwd:asd88fgpdd777uzjYhagZg 1000 a=ice-ufrag:8hhY 1001 m=audio 5000 RTP/AVP 0 1002 a=rtpmap:0 PCMU/8000 1003 a=candidate:1 1 UDP 2130706431 2001:db8:a0b:12f0::1 5000 typ host 1004 a=candidate:1 2 UDP 2130706431 2001:db8:a0b:12f0::1 5001 typ host 1005 a=candidate:2 1 UDP 1694498815 2001:db8:a0b:12f0::3 5000 typ srflx 1006 raddr 2001:db8:a0b:12f0::1 rport 8998 1007 a=candidate:2 2 UDP 1694498815 2001:db8:a0b:12f0::3 5001 typ srflx 1008 raddr 2001:db8:a0b:12f0::1 rport 8998 1010 For this description the RTCP host candidate MUST NOT be sent prior 1011 to the RTP host candidate. Similarly the RTP server reflexive 1012 candidate MUST be sent together with or prior to the RTCP server 1013 reflexive candidate. 1015 Similar considerations apply at the level of media streams in 1016 addition to foundations; this is covered by the requirement to always 1017 start unfreezing candidates starting from the first media stream as 1018 described under Section 5.2. 1020 Appendix D. Changes from Earlier Versions 1022 Note to the RFC-Editor: please remove this section prior to 1023 publication as an RFC. 1025 D.1. Changes from draft-ietf-ice-trickle-04 1027 o Removed dependency on SDP and offer/answer model. 1029 o Removed mentions of aggressive nomination, since it is deprecated 1030 in 5245bis. 1032 o Added section on requirements for signaling protocols. 1034 o Clarified terminology. 1036 o Addressed various WG feedback. 1038 D.2. Changes from draft-ietf-ice-trickle-03 1040 o Copy edit. 1042 D.3. Changes from draft-ietf-ice-trickle-03 1044 o Provided more detailed description of unfreezing behavior, 1045 specifically how to replace pre-existing peer-reflexive candidates 1046 with higher-priority ones received via trickling. 1048 D.4. Changes from draft-ietf-ice-trickle-02 1050 o Adjusted unfreezing behavior when there are disparate foundations. 1052 D.5. Changes from draft-ietf-ice-trickle-01 1054 o Changed examples to use IPv6. 1056 D.6. Changes from draft-ietf-ice-trickle-00 1058 o Removed dependency on SDP (which is to be provided in a separate 1059 specification). 1061 o Clarified text about the fact that a check list can be empty if no 1062 candidates have been sent or received yet. 1064 o Clarified wording about check list states so as not to define new 1065 states for "Active" and "Frozen" because those states are not 1066 defined for check lists (only for candidate pairs) in ICE core. 1068 o Removed open issues list because it was out of date. 1070 o Completed a thorough copy edit. 1072 D.7. Changes from draft-mmusic-trickle-ice-02 1074 o Addressed feedback from Rajmohan Banavi and Brandon Williams. 1076 o Clarified text about determining support and about how to proceed 1077 if it can be determined that the answering agent does not support 1078 Trickle ICE. 1080 o Clarified text about check list and timer updates. 1082 o Clarified when it is appropriate to use half trickle or to send no 1083 candidates in an offer or answer. 1085 o Updated the list of open issues. 1087 D.8. Changes from draft-ivov-01 and draft-mmusic-00 1089 o Added a requirement to trickle candidates by order of components 1090 to avoid deadlocks in the unfreezing algorithm. 1092 o Added an informative note on peer-reflexive candidates explaining 1093 that nothing changes for them semantically but they do become a 1094 more likely occurrence for Trickle ICE. 1096 o Limit the number of pairs to 100 to comply with 5245. 1098 o Added clarifications on the non-importance of how newly discovered 1099 candidates are trickled/sent to the remote party or if this is 1100 done at all. 1102 o Added transport expectations for trickled candidates as per Dale 1103 Worley's recommendation. 1105 D.9. Changes from draft-ivov-00 1107 o Specified that end-of-candidates is a media level attribute which 1108 can of course appear as session level, which is equivalent to 1109 having it appear in all m-lines. Also made end-of-candidates 1110 optional for cases such as aggressive nomination for controlled 1111 agents. 1113 o Added an example for ICE lite and Trickle ICE to illustrate how, 1114 when talking to an ICE lite agent doesn't need to send or even 1115 discover any candidates. 1117 o Added an example for ICE lite and Trickle ICE to illustrate how, 1118 when talking to an ICE lite agent doesn't need to send or even 1119 discover any candidates. 1121 o Added wording that explicitly states ICE lite agents have to be 1122 prepared to receive no candidates over signaling and that they 1123 should not freak out if this happens. (Closed the corresponding 1124 open issue). 1126 o It is now mandatory to use MID when trickling candidates and using 1127 m-line indexes is no longer allowed. 1129 o Replaced use of 0.0.0.0 to IP6 :: in order to avoid potential 1130 issues with RFC2543 SDP libraries that interpret 0.0.0.0 as an on- 1131 hold operation. Also changed the port number here from 1 to 9 1132 since it already has a more appropriate meaning. (Port change 1133 suggested by Jonathan Lennox). 1135 o Closed the Open Issue about use about what to do with cands 1136 received after end-of-cands. Solution: ignore, do an ICE restart 1137 if you want to add something. 1139 o Added more terminology, including trickling, trickled candidates, 1140 half trickle, full trickle, 1142 o Added a reference to the SIP usage for Trickle ICE as requested at 1143 the Boston interim. 1145 D.10. Changes from draft-rescorla-01 1147 o Brought back explicit use of Offer/Answer. There are no more 1148 attempts to try to do this in an O/A independent way. Also 1149 removed the use of ICE Descriptions. 1151 o Added SDP specification for trickled candidates, the trickle 1152 option and 0.0.0.0 addresses in m-lines, and end-of-candidates. 1154 o Support and Discovery. Changed that section to be less abstract. 1155 As discussed in IETF85, the draft now says implementations and 1156 usages need to either determine support in advance and directly 1157 use trickle, or do half trickle. Removed suggestion about use of 1158 discovery in SIP or about letting implementing protocols do what 1159 they want. 1161 o Defined Half Trickle. Added a section that says how it works. 1162 Mentioned that it only needs to happen in the first o/a (not 1163 necessary in updates), and added Jonathan's comment about how it 1164 could, in some cases, offer more than half the improvement if you 1165 can pre-gather part or all of your candidates before the user 1166 actually presses the call button. 1168 o Added a short section about subsequent offer/answer exchanges. 1170 o Added a short section about interactions with ICE Lite 1171 implementations. 1173 o Added two new entries to the open issues section. 1175 D.11. Changes from draft-rescorla-00 1177 o Relaxed requirements about verifying support following a 1178 discussion on MMUSIC. 1180 o Introduced ICE descriptions in order to remove ambiguous use of 1181 3264 language and inappropriate references to offers and answers. 1183 o Removed inappropriate assumption of adoption by RTCWEB pointed out 1184 by Martin Thomson. 1186 Authors' Addresses 1188 Emil Ivov 1189 Atlassian 1190 303 Colorado Street, #1600 1191 Austin 78701 1192 USA 1194 Phone: +1-512-640-3000 1195 Email: eivov@atlassian.com 1197 Eric Rescorla 1198 RTFM, Inc. 1199 2064 Edgewood Drive 1200 Palo Alto, CA 94303 1201 USA 1203 Phone: +1 650 678 2350 1204 Email: ekr@rtfm.com 1206 Justin Uberti 1207 Google 1208 747 6th St S 1209 Kirkland, WA 98033 1210 USA 1212 Phone: +1 857 288 8888 1213 Email: justin@uberti.name 1214 Peter Saint-Andre 1215 Filament 1216 P.O. Box 787 1217 Parker, CO 80134 1218 USA 1220 Phone: +1 720 256 6756 1221 Email: peter@filament.com 1222 URI: https://filament.com/