idnits 2.17.1 draft-ietf-v6ops-rfc6555bis-07.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 abstract seems to contain references ([RFC6555]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. -- The document has examples using IPv4 documentation addresses according to RFC6890, but does not use any IPv6 documentation addresses. Maybe there should be IPv6 examples, too? -- The draft header indicates that this document obsoletes RFC6555, but the abstract doesn't seem to directly say this. It does mention RFC6555 though, so this could be OK. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (October 25, 2017) is 2373 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) == Missing Reference: 'Section 3' is mentioned on line 125, but not defined == Missing Reference: 'Section 4' is mentioned on line 169, but not defined == Missing Reference: 'Section 5' is mentioned on line 208, but not defined == Missing Reference: 'Section 6' is mentioned on line 175, but not defined ** Obsolete normative reference: RFC 6555 (Obsoleted by RFC 8305) Summary: 2 errors (**), 0 flaws (~~), 5 warnings (==), 3 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network D. Schinazi 3 Internet-Draft T. Pauly 4 Obsoletes: 6555 (if approved) Apple Inc. 5 Intended status: Standards Track October 25, 2017 6 Expires: April 28, 2018 8 Happy Eyeballs Version 2: Better Connectivity Using Concurrency 9 draft-ietf-v6ops-rfc6555bis-07 11 Abstract 13 Many communication protocols operated over the modern Internet use 14 host names. These often resolve to multiple IP addresses, each of 15 which may have different performance and connectivity 16 characteristics. Since specific addresses or address families (IPv4 17 or IPv6) may be blocked, broken, or sub-optimal on a network, clients 18 that attempt multiple connections in parallel have a higher chance of 19 establishing a connection sooner. This document specifies 20 requirements for algorithms that reduce this user-visible delay and 21 provides an example algorithm, referred to as "Happy Eyeballs". This 22 document obsoletes the original algorithm description in [RFC6555]. 24 Status of This Memo 26 This Internet-Draft is submitted in full conformance with the 27 provisions of BCP 78 and BCP 79. 29 Internet-Drafts are working documents of the Internet Engineering 30 Task Force (IETF). Note that other groups may also distribute 31 working documents as Internet-Drafts. The list of current Internet- 32 Drafts is at https://datatracker.ietf.org/drafts/current/. 34 Internet-Drafts are draft documents valid for a maximum of six months 35 and may be updated, replaced, or obsoleted by other documents at any 36 time. It is inappropriate to use Internet-Drafts as reference 37 material or to cite them other than as "work in progress." 39 This Internet-Draft will expire on April 28, 2018. 41 Copyright Notice 43 Copyright (c) 2017 IETF Trust and the persons identified as the 44 document authors. All rights reserved. 46 This document is subject to BCP 78 and the IETF Trust's Legal 47 Provisions Relating to IETF Documents 48 (https://trustee.ietf.org/license-info) in effect on the date of 49 publication of this document. Please review these documents 50 carefully, as they describe your rights and restrictions with respect 51 to this document. Code Components extracted from this document must 52 include Simplified BSD License text as described in Section 4.e of 53 the Trust Legal Provisions and are provided without warranty as 54 described in the Simplified BSD License. 56 Table of Contents 58 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 59 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 60 2. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . 3 61 3. Hostname Resolution Query Handling . . . . . . . . . . . . . 4 62 3.1. Handling Multiple DNS Server Addresses . . . . . . . . . 5 63 4. Sorting Addresses . . . . . . . . . . . . . . . . . . . . . . 5 64 5. Connection Attempts . . . . . . . . . . . . . . . . . . . . . 6 65 6. DNS Answer Changes during Happy Eyeballs Connection Setup . . 7 66 7. Supporting IPv6-only Networks with NAT64 and DNS64 . . . . . 8 67 7.1. IPv4 Address Literals . . . . . . . . . . . . . . . . . . 8 68 7.2. Host Names with Broken AAAA Records . . . . . . . . . . . 9 69 7.3. Virtual Private Networks . . . . . . . . . . . . . . . . 10 70 8. Summary of Configurable Values . . . . . . . . . . . . . . . 11 71 9. Limitations . . . . . . . . . . . . . . . . . . . . . . . . . 11 72 9.1. Path Maximum Transmission Unit Discovery . . . . . . . . 12 73 9.2. Application Layer . . . . . . . . . . . . . . . . . . . . 12 74 9.3. Hiding Operational Issues . . . . . . . . . . . . . . . . 12 75 10. Security Considerations . . . . . . . . . . . . . . . . . . . 12 76 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 12 77 12. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 12 78 13. References . . . . . . . . . . . . . . . . . . . . . . . . . 13 79 13.1. Normative References . . . . . . . . . . . . . . . . . . 13 80 13.2. Informative References . . . . . . . . . . . . . . . . . 14 81 Appendix A. Differences from RFC6555 . . . . . . . . . . . . . . 14 82 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 15 84 1. Introduction 86 Many communication protocols operated over the modern Internet use 87 host names. These often resolve to multiple IP addresses, each of 88 which may have different performance and connectivity 89 characteristics. Since specific addresses or address families (IPv4 90 or IPv6) may be blocked, broken, or sub-optimal on a network, clients 91 that attempt multiple connections in parallel have a higher chance of 92 establishing a connection sooner. This document specifies 93 requirements for algorithms that reduce this user-visible delay and 94 provides an example algorithm. 96 This document defines the algorithm for "Happy Eyeballs", a technique 97 of reducing user-visible delays on dual-stack hosts. This definition 98 obsoletes the original description in [RFC6555]. Now that this 99 approach has been deployed at scale and measured for several years, 100 the algorithm specification can be refined to improve its reliability 101 and generalization. 103 The Happy Eyeballs algorithm of racing resolved addresses has several 104 stages of ordering and racing to avoid delays to the user whenever 105 possible, while preferring the use of IPv6. This document discusses 106 how to handle DNS queries when starting a connection on a dual-stack 107 client, how to create an ordered list of destination addresses to 108 which to attempt connections, and how to race the connection 109 attempts. 111 1.1. Requirements Language 113 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 114 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 115 "OPTIONAL" in this document are to be interpreted as described in 116 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, 117 as shown here. 119 2. Overview 121 This document defines a method of connection establishment, named 122 "Happy Eyeballs Connection Setup". This approach has several 123 distinct phases: 125 1. Initiation of asynchronous DNS queries [Section 3] 127 2. Sorting of resolved destination addresses [Section 4] 129 3. Initiation of asynchronous connection attempts [Section 5] 131 4. Establishment of one connection, which cancels all other attempts 132 Note that this document assumes that the host destination address 133 preference policy favors IPv6 over IPv4. IPv6 has many desirable 134 properties designed to be improvements over IPv4 [RFC8200]. If the 135 host is configured to have a different preference, the 136 recommendations in this document can be easily adapted. 138 3. Hostname Resolution Query Handling 140 When a client has both IPv4 and IPv6 connectivity, and is trying to 141 establish a connection with a named host, it needs to send out both 142 AAAA and A DNS queries. Both queries SHOULD be made as soon after 143 one another as possible, with the AAAA query made first, immediately 144 followed by the A query. 146 Implementations SHOULD NOT wait for both families of answers to 147 return before attempting connection establishment. If one query 148 fails to return, or takes significantly longer to return, waiting for 149 the second address family can significantly delay the connection 150 establishment of the first one. Therefore, the client SHOULD treat 151 DNS resolution as asynchronous. Note that if the platform does not 152 offer an asynchronous DNS API, this behavior can be simulated by 153 making two separate synchronous queries on different threads, one per 154 address family. 156 The algorithm proceeds as follows: if a positive AAAA response (a 157 response with at least one valid AAAA record) is received first, the 158 first IPv6 connection attempt is immediately started. If a positive 159 A response is received first due to reordering, the client SHOULD 160 wait for a short time for the AAAA response to ensure preference is 161 given to IPv6 (it is common for the AAAA response to follow the A 162 response by a few milliseconds). This delay will be referred to as 163 the "Resolution Delay". The recommended value for the Resolution 164 Delay is 50 milliseconds. If a positive AAAA response is received 165 within the Resolution Delay period, the client immediately starts the 166 IPv6 connection attempt. If a negative AAAA response (no error, no 167 data) is received within the Resolution Delay period or the AAAA 168 response has not been received by the end of the Resolution Delay 169 period, the client SHOULD proceed to Sorting Addresses [Section 4] 170 and staggered connection attempts [Section 5] using any IPv4 171 addresses returned so far. If the AAAA response arrives while these 172 connection attempts are in progress, but before any connection has 173 been established, then the newly received IPv6 addresses are 174 incorporated into the list of available candidate addresses 175 [Section 6] and the process of connection attempts will continue with 176 the IPv6 addresses added, until one connection is established. 178 3.1. Handling Multiple DNS Server Addresses 180 If multiple DNS server addresses are configured for the current 181 network, the client may have the option of sending its DNS queries 182 over IPv4 or IPv6. In keeping with the Happy Eyeballs approach, 183 queries SHOULD be sent over IPv6 first (note that this is not 184 referring to the sending of AAAA or A queries, but rather the address 185 of the DNS server itself and IP version used to transport DNS 186 messages). If DNS queries sent to the IPv6 address do not receive 187 responses, that address may be marked as penalized, and queries can 188 be sent to other DNS server addresses. 190 As native IPv6 deployments become more prevalent, and IPv4 addresses 191 are exhausted, it is expected that IPv6 connectivity will have 192 preferential treatment within networks. If a DNS server is 193 configured to be accessible over IPv6, IPv6 should be assumed to be 194 the preferred address family. 196 Client systems SHOULD NOT have an explicit limit to the number of DNS 197 servers that can be configured, either manually or by the network. 198 If such a limit is required by hardware limitations, the client 199 SHOULD use at least one address from each address family from the 200 available list. 202 4. Sorting Addresses 204 Before attempting to connect to any of the resolved destination 205 addresses, the client should define the order in which to start the 206 attempts. Once the order has been defined, the client can use a 207 simple algorithm for racing each option after a short delay 208 [Section 5]. It is important that the ordered list involves all 209 addresses from both families that have been received by this point, 210 as this allows the client to get the racing effect of Happy Eyeballs 211 for the entire list, not just the first IPv4 and first IPv6 212 addresses. 214 First, the client MUST sort the addresses received up to this point 215 using Destination Address Selection ([RFC6724], Section 6). 217 If the client is stateful and has history of expected round-trip 218 times (RTT) for the routes to access each address, it SHOULD add a 219 Destination Address Selection rule between rules 8 and 9 that prefers 220 addresses with lower RTTs. If the client keeps track of which 221 addresses it has used in the past, it SHOULD add another destination 222 address selection rule between the RTT rule and rule 9, which prefers 223 used addresses over unused ones. This helps servers that use the 224 client's IP address during authentication, as is the case for TCP 225 Fast Open [RFC7413] and some HTTP cookies. This historical data MUST 226 NOT be used across different network interfaces, and SHOULD be 227 flushed whenever a device changes the network to which it is 228 attached. 230 Next, the client SHOULD modify the ordered list to interleave address 231 families. Whichever address family is first in the list should be 232 followed by an address of the other address family; that is, if the 233 first address in the sorted list is IPv6, then the first IPv4 address 234 should be moved up in the list to be second in the list. An 235 implementation MAY want to favor one address family more by allowing 236 multiple addresses of that family to be attempted before trying the 237 other family. The number of contiguous addresses of the first 238 address family will be referred to as the "First Address Family 239 Count", and can be a configurable value. This is performed to avoid 240 waiting through a long list of addresses from a given address family 241 if connectivity over that address family is impaired. 243 Note that the address selection described in this section only 244 applies to destination addresses; Source Address Selection 245 ([RFC6724], Section 5) is performed once per destination address and 246 is out of scope of this document. 248 5. Connection Attempts 250 Once the list of addresses received up to this point has been 251 constructed, the client will attempt to make connections. In order 252 to avoid unreasonable network load, connection attempts SHOULD NOT be 253 made simultaneously. Instead, one connection attempt to a single 254 address is started first, followed by the others in the list, one at 255 a time. Starting a new connection attempt does not affect previous 256 attempts, as multiple connection attempts may occur in parallel. 257 Once one of the connection attempts succeeds (generally when the TCP 258 handshake completes), all other connections attempts that have not 259 yet succeeded SHOULD be cancelled. Any address that was not yet 260 attempted as a connection SHOULD be ignored. At that time, the 261 asynchronous DNS query MAY be cancelled as new addresses will not be 262 used for this connection. However, the DNS client resolver SHOULD 263 still process DNS replies from the network for a short period of time 264 (recommended to be 1 second), as they will populate the DNS cache and 265 can be used for subsequent connections. 267 A simple implementation can have a fixed delay for how long to wait 268 before starting the next connection attempt. This delay is referred 269 to as the "Connection Attempt Delay". One recommended value for a 270 default delay is 250 milliseconds. A more nuanced implementation's 271 delay should correspond to the time when the previous attempt is 272 sending its second TCP SYN, based on TCP's retransmission timer 273 [RFC6298]. If the client has historical RTT data gathered from other 274 connections to the same host or prefix, it can use this information 275 to influence its delay. Note that this algorithm should only try to 276 approximate the time of the first SYN retransmission, and not any 277 further retransmissions which may be influenced by exponential timer 278 back off. 280 The Connection Attempt Delay MUST have a lower bound, especially if 281 it is computed using historical data. More specifically, a 282 subsequent connection MUST NOT be started within 10 milliseconds of 283 the previous attempt. The recommended minimum value is 100 284 milliseconds, which is referred to as the "Minimum Connection Attempt 285 Delay". This minimum value is required to avoid congestion collapse 286 in the presence of high packet loss rates. The Connection Attempt 287 Delay SHOULD have an upper bound, referred to as the "Maximum 288 Connection Attempt Delay". The current recommended value is 2 289 seconds. 291 6. DNS Answer Changes during Happy Eyeballs Connection Setup 293 If, during the course of connection establishment, the DNS answers 294 change either by adding resolved addresses (for example, due to DNS 295 push notifications [DNS-PUSH]), or removing previously resolved 296 addresses (for example, due to expiry of the TTL on that DNS record), 297 the client should react based on its current progress. 299 If an address is removed from the list that already had a connection 300 attempt started, the connection attempt SHOULD NOT be cancelled, but 301 rather be allowed to continue. If the removed address had not yet 302 had a connection attempt started, it SHOULD be removed from the list 303 of addresses to try. 305 If an address is added to the list, it should be sorted into the list 306 of addresses not yet attempted according to the rules above 307 (Section 4). 309 7. Supporting IPv6-only Networks with NAT64 and DNS64 311 While many IPv6 transition protocols have been standardized and 312 deployed, most are transparent to client devices. The combined use 313 of NAT64 [RFC6146] and DNS64 [RFC6147] is a popular solution that is 314 being deployed and requires changes in client devices. One possible 315 way to handle these networks is for the client device networking 316 stack to implement 464XLAT [RFC6877]. 464XLAT has the advantage of 317 not requiring changes to user space software, however it requires 318 per-packet translation if the application is using IPv4 literals and 319 does not encourage client application software to support native 320 IPv6. On platforms that do not support 464XLAT, the Happy Eyeballs 321 engine SHOULD follow the recommendations in this section to properly 322 support IPv6-only networks with NAT64 and DNS64. 324 The features described in this section SHOULD only be enabled when 325 the host detects one of these networks. A simple heuristic to 326 achieve that is to check if the network offers routable IPv6 327 addressing, does not offer routable IPv4 addressing, and offers a DNS 328 resolver address. 330 7.1. IPv4 Address Literals 332 If client applications or users wish to connect to IPv4 address 333 literals, the Happy Eyeballs engine will need to perform NAT64 334 address synthesis for them. The solution is similar to "Bump-in-the- 335 Host" [RFC6535] but is implemented inside the Happy Eyeballs library. 337 When an IPv4 address is passed in to the library instead of a host 338 name, the device queries the network for the NAT64 prefix using 339 "Discovery of the IPv6 Prefix Used for IPv6 Address Synthesis" 340 [RFC7050] then synthesizes an appropriate IPv6 address (or several) 341 using the encoding described in "IPv6 Addressing of IPv4/IPv6 342 Translators" [RFC6052]. The synthesized addresses are then inserted 343 into the list of addresses as if they were results from DNS queries; 344 connection attempts follow the algorithm described above (Section 5). 346 7.2. Host Names with Broken AAAA Records 348 At the time of writing, there exist a small but non negligible number 349 of host names that resolve to valid A records and broken AAAA 350 records, which we define as AAAA records that contain seemingly valid 351 IPv6 addresses but those addresses never reply when contacted on the 352 usual ports. These can be for example caused by: 354 o Mistyping of the IPv6 address in the DNS zone configuration 356 o Routing black holes 358 o Service outages 360 While an algorithm complying with the other sections of this document 361 would correctly handle such host names on a dual-stack network, they 362 will not necessarily function correctly on IPv6-only networks with 363 NAT64 and DNS64. Since DNS64 recursive resolvers rely on the 364 authoritative name servers sending negative ("no error no answer") 365 responses for AAAA records in order to synthesize, they will not 366 synthesize records for these particular host names, and will instead 367 pass through the broken AAAA record. 369 In order to support these scenarios, the client device needs to query 370 the DNS for the A record then perform local synthesis. Since these 371 types of host names are rare, and in order to minimize load on DNS 372 servers, this A query should only be performed when the client has 373 given up on the AAAA records it initially received. This can be 374 achieved by using a longer timeout, referred to as the "Last Resort 375 Local Synthesis Delay" and recommended to be 2 seconds. The timer is 376 started when the last connection attempt is fired. If no connection 377 attempt has succeeded when this timer fires, the device queries the 378 DNS for the IPv4 address and on reception of a valid A record, treats 379 it as if it were provided by the application (Section 7.1). 381 7.3. Virtual Private Networks 383 Some Virtual Private Networks (VPN) may be configured to handle DNS 384 queries from the device. The configuration could encompass all 385 queries, or a subset such as "*.internal.example.com". These VPNs 386 can also be configured to only route part of the IPv4 address space, 387 such as 192.0.2.0/24. However, if an internal hostname resolves to 388 an external IPv4 address, these can cause issues if the underlying 389 network is IPv6-only. As an example, let's assume that 390 "www.internal.example.com" has exactly one A record, 198.51.100.42, 391 and no AAAA records. The client will send the DNS query to the 392 company's recursive resolver and that resolver will reply with these 393 records. The device now only has an IPv4 address to connect to, and 394 no route to that address. Since the company's resolver does not know 395 the NAT64 prefix of the underlying network, it cannot synthesize the 396 address. Similarly, the underlying network's DNS64 recursive 397 resolver does not know the company's internal addresses, so it cannot 398 resolve the hostname. Because of this, the client device needs to 399 resolve the A record using the company's resolver then locally 400 synthesize an IPv6 address, as if the resolved IPv4 address were 401 provided by the application (Section 7.1). 403 8. Summary of Configurable Values 405 The values that may be configured as defaults on a client for use in 406 Happy Eyeballs are as follows: 408 o Resolution Delay (Section 3): The time to wait for a AAAA response 409 after receiving an A response. Recommended to be 50 milliseconds. 411 o First Address Family Count (Section 4): The number of addresses 412 belonging to the first address family (such as IPv6) that should 413 be attempted before attempting another address family. 414 Recommended to be 1, or 2 to more aggressively favor one address 415 family. 417 o Connection Attempt Delay (Section 5): The time to wait between 418 connection attempts in the absence of RTT data. Recommended to be 419 250 milliseconds. 421 o Minimum Connection Attempt Delay (Section 5): The minimum time to 422 wait between connection attempts. Recommended to be 100 423 milliseconds. MUST NOT be less than 10 milliseconds. 425 o Maximum Connection Attempt Delay (Section 5): The maximum time to 426 wait between connection attempts. Recommended to be 2 seconds. 428 o Last Resort Local Synthesis Delay (Section 7.2): The time to wait 429 after starting the last IPv6 attempt and before sending the A 430 query. Recommended to be 2 seconds. 432 The delay values described in this section were determined 433 empirically by measuring the timing of connections on a very wide set 434 of production devices. They were picked to reduce wait times noticed 435 by users while minimizing load on the network. As time passes, it is 436 expected that the properties of networks will evolve. For that 437 reason, it is expected that these values will change over time. 438 Implementors should feel welcome to use different values without 439 changing this specification. Since IPv6 issues are expected to be 440 less common, the delays SHOULD be increased with time as client 441 software is updated. 443 9. Limitations 445 Happy Eyeballs will handle initial connection failures at the TCP/IP 446 layer, however other failures or performance issues may still affect 447 the chosen connection. 449 9.1. Path Maximum Transmission Unit Discovery 451 Since Happy Eyeballs is only active during the initial handshake and 452 TCP does not pass the initial handshake, issues related to MTU can be 453 masked and go unnoticed during Happy Eyeballs. Solving this issue is 454 out of scope of this document. One solution is to use Packetization 455 Layer Path MTU Discovery [RFC4821]. 457 9.2. Application Layer 459 If the DNS returns multiple addresses for different application 460 servers, the application itself may not be operational and functional 461 on all of them. Common examples include Transport Layer Security 462 (TLS) and the Hypertext Transport Protocol (HTTP). 464 9.3. Hiding Operational Issues 466 It has been observed in practice that Happy Eyeballs can hide issues 467 in networks. For example, if a misconfiguration causes IPv6 to 468 consistently fail on a given network while IPv4 is still functional, 469 Happy Eyeballs may impair the operator's ability to notice the issue. 470 It is recommended that network operators deploy external means of 471 monitoring to ensure functionality of all address families. 473 10. Security Considerations 475 Note that applications should not rely upon a stable hostname-to- 476 address mapping to ensure any security properties, since DNS results 477 may change between queries. Happy Eyeballs may make it more likely 478 that subsequent connections to a single hostname use different IP 479 addresses. 481 11. IANA Considerations 483 This memo includes no request to IANA. 485 12. Acknowledgments 487 The authors thank Dan Wing, Andrew Yourtchenko, and everyone else who 488 worked on the original Happy Eyeballs design [RFC6555], Josh 489 Graessley, Stuart Cheshire, and the rest of team at Apple that helped 490 implement and instrument this algorithm, and Jason Fesler and Paul 491 Saab who helped measure and refine this algorithm. The authors would 492 also like to thank Fred Baker, Nick Chettle, Lorenzo Colitti, Igor 493 Gashinsky, Geoff Huston, Jen Linkova, Paul Hoffman, Philip Homburg, 494 Warren Kumari, Erik Nygren, Jordi Palet Martinez, Rui Paulo, Stephen 495 Strowes, Jinmei Tatuya, Dave Thaler, Joe Touch and James Woodyatt for 496 their input and contributions. 498 13. References 500 13.1. Normative References 502 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 503 Requirement Levels", BCP 14, RFC 2119, 504 DOI 10.17487/RFC2119, March 1997, 505 . 507 [RFC4821] Mathis, M. and J. Heffner, "Packetization Layer Path MTU 508 Discovery", RFC 4821, DOI 10.17487/RFC4821, March 2007, 509 . 511 [RFC6052] Bao, C., Huitema, C., Bagnulo, M., Boucadair, M., and X. 512 Li, "IPv6 Addressing of IPv4/IPv6 Translators", RFC 6052, 513 DOI 10.17487/RFC6052, October 2010, 514 . 516 [RFC6146] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful 517 NAT64: Network Address and Protocol Translation from IPv6 518 Clients to IPv4 Servers", RFC 6146, DOI 10.17487/RFC6146, 519 April 2011, . 521 [RFC6147] Bagnulo, M., Sullivan, A., Matthews, P., and I. van 522 Beijnum, "DNS64: DNS Extensions for Network Address 523 Translation from IPv6 Clients to IPv4 Servers", RFC 6147, 524 DOI 10.17487/RFC6147, April 2011, 525 . 527 [RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, 528 "Computing TCP's Retransmission Timer", RFC 6298, 529 DOI 10.17487/RFC6298, June 2011, 530 . 532 [RFC6535] Huang, B., Deng, H., and T. Savolainen, "Dual-Stack Hosts 533 Using "Bump-in-the-Host" (BIH)", RFC 6535, 534 DOI 10.17487/RFC6535, February 2012, 535 . 537 [RFC6555] Wing, D. and A. Yourtchenko, "Happy Eyeballs: Success with 538 Dual-Stack Hosts", RFC 6555, DOI 10.17487/RFC6555, April 539 2012, . 541 [RFC6724] Thaler, D., Ed., Draves, R., Matsumoto, A., and T. Chown, 542 "Default Address Selection for Internet Protocol Version 6 543 (IPv6)", RFC 6724, DOI 10.17487/RFC6724, September 2012, 544 . 546 [RFC7050] Savolainen, T., Korhonen, J., and D. Wing, "Discovery of 547 the IPv6 Prefix Used for IPv6 Address Synthesis", 548 RFC 7050, DOI 10.17487/RFC7050, November 2013, 549 . 551 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 552 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 553 May 2017, . 555 13.2. Informative References 557 [DNS-PUSH] 558 Pusateri, T. and S. Cheshire, "DNS Push Notifications", 559 Work in Progress, draft-ietf-dnssd-push, March 2017. 561 [RFC6877] Mawatari, M., Kawashima, M., and C. Byrne, "464XLAT: 562 Combination of Stateful and Stateless Translation", 563 RFC 6877, DOI 10.17487/RFC6877, April 2013, 564 . 566 [RFC7413] Cheng, Y., Chu, J., Radhakrishnan, S., and A. Jain, "TCP 567 Fast Open", RFC 7413, DOI 10.17487/RFC7413, December 2014, 568 . 570 [RFC8200] Deering, S. and R. Hinden, "Internet Protocol, Version 6 571 (IPv6) Specification", STD 86, RFC 8200, 572 DOI 10.17487/RFC8200, July 2017, 573 . 575 Appendix A. Differences from RFC6555 577 "Happy Eyeballs: Success with Dual-Stack Hosts" [RFC6555] mostly 578 concentrates on how to stagger connections to a hostname that has an 579 AAAA and an A record. This document additionally discusses: 581 o how to perform DNS queries to obtain these addresses 583 o how to handle multiple addresses from each address family 585 o how to handle DNS updates while connections are being raced 587 o how to leverage historical information 589 o how to support IPv6-only networks with NAT64 and DNS64 591 Note that a simple implementation of the algorithm described in this 592 document is still compliant with the previous specification 594 [RFC6555]. Implementations should take the new considerations into 595 account when applicable to optimize their behavior. 597 Authors' Addresses 599 David Schinazi 600 Apple Inc. 601 1 Infinite Loop 602 Cupertino, California 95014 603 US 605 Email: dschinazi@apple.com 607 Tommy Pauly 608 Apple Inc. 609 1 Infinite Loop 610 Cupertino, California 95014 611 US 613 Email: tpauly@apple.com