idnits 2.17.1 draft-ietf-v6ops-rfc6555bis-02.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 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 mention this, which it should. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (July 3, 2017) is 2482 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 120, but not defined == Missing Reference: 'Section 4' is mentioned on line 159, but not defined == Missing Reference: 'Section 5' is mentioned on line 197, but not defined == Missing Reference: 'Section 6' is mentioned on line 164, but not defined ** Obsolete normative reference: RFC 6555 (Obsoleted by RFC 8305) Summary: 1 error (**), 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 July 3, 2017 6 Expires: January 4, 2018 8 Happy Eyeballs Version 2: Better Connectivity Using Concurrency 9 draft-ietf-v6ops-rfc6555bis-02 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. 23 Status of This Memo 25 This Internet-Draft is submitted in full conformance with the 26 provisions of BCP 78 and BCP 79. 28 Internet-Drafts are working documents of the Internet Engineering 29 Task Force (IETF). Note that other groups may also distribute 30 working documents as Internet-Drafts. The list of current Internet- 31 Drafts is at http://datatracker.ietf.org/drafts/current/. 33 Internet-Drafts are draft documents valid for a maximum of six months 34 and may be updated, replaced, or obsoleted by other documents at any 35 time. It is inappropriate to use Internet-Drafts as reference 36 material or to cite them other than as "work in progress." 38 This Internet-Draft will expire on January 4, 2018. 40 Copyright Notice 42 Copyright (c) 2017 IETF Trust and the persons identified as the 43 document authors. All rights reserved. 45 This document is subject to BCP 78 and the IETF Trust's Legal 46 Provisions Relating to IETF Documents 47 (http://trustee.ietf.org/license-info) in effect on the date of 48 publication of this document. Please review these documents 49 carefully, as they describe your rights and restrictions with respect 50 to this document. Code Components extracted from this document must 51 include Simplified BSD License text as described in Section 4.e of 52 the Trust Legal Provisions and are provided without warranty as 53 described in the Simplified BSD License. 55 Table of Contents 57 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 58 1.1. Requirements Language . . . . . . . . . . . . . . . . . . 3 59 2. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . 3 60 3. Hostname Resolution Query Handling . . . . . . . . . . . . . 4 61 3.1. Handling Multiple DNS Server Addresses . . . . . . . . . 4 62 4. Sorting Addresses . . . . . . . . . . . . . . . . . . . . . . 5 63 5. Connection Attempts . . . . . . . . . . . . . . . . . . . . . 6 64 6. DNS Answer Changes during Happy Eyeballs Connection Setup . . 7 65 7. Supporting IPv6-only Networks with NAT64 and DNS64 . . . . . 7 66 7.1. IPv4 Address Literals . . . . . . . . . . . . . . . . . . 8 67 7.2. Host Names with Broken AAAA Records . . . . . . . . . . . 8 68 7.3. Virtual Private Networks . . . . . . . . . . . . . . . . 9 69 8. Summary of Configurable Values . . . . . . . . . . . . . . . 10 70 9. Limitations . . . . . . . . . . . . . . . . . . . . . . . . . 10 71 9.1. Path Maximum Transmission Unit Discovery . . . . . . . . 10 72 9.2. Application Layer . . . . . . . . . . . . . . . . . . . . 11 73 9.3. Hiding Operational Issues . . . . . . . . . . . . . . . . 11 74 10. Security Considerations . . . . . . . . . . . . . . . . . . . 11 75 11. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 11 76 12. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 11 77 13. References . . . . . . . . . . . . . . . . . . . . . . . . . 12 78 13.1. Normative References . . . . . . . . . . . . . . . . . . 12 79 13.2. Informative References . . . . . . . . . . . . . . . . . 13 80 Appendix A. Differences from RFC6555 . . . . . . . . . . . . . . 13 81 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 13 83 1. Introduction 85 Many communication protocols operated over the modern Internet use 86 host names. These often resolve to multiple IP addresses, each of 87 which may have different performance and connectivity 88 characteristics. Since specific addresses or address families (IPv4 89 or IPv6) may be blocked, broken, or sub-optimal on a network, clients 90 that attempt multiple connections in parallel have a higher chance of 91 establishing a connection sooner. This document specifies 92 requirements for algorithms that reduce this user-visible delay and 93 provides an example algorithm. 95 This documents expands on "Happy Eyeballs" [RFC6555], a technique of 96 reducing user-visible delays on dual-stack hosts. Now that this 97 approach has been deployed at scale and measured for several years, 98 the algorithm specification can be refined to improve its reliability 99 and generalization. This document recommends an algorithm of racing 100 resolved addresses that has several stages of ordering and racing to 101 avoid delays to the user whenever possible, while preferring the use 102 of IPv6. Specifically, it discusses how to handle DNS queries when 103 starting a connection on a dual-stack client, how to create an 104 ordered list of addresses to which to attempt connections, and how to 105 race the connection attempts. 107 1.1. Requirements Language 109 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 110 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 111 "OPTIONAL" in this document are to be interpreted as described in 112 "Key words for use in RFCs to Indicate Requirement Levels" [RFC2119]. 114 2. Overview 116 This document defines a method of connection establishment, named 117 "Happy Eyeballs Connection Setup". This approach has several 118 distinct phases: 120 1. Initiation of asynchronous DNS queries [Section 3] 122 2. Sorting of resolved addresses [Section 4] 124 3. Initiation of asynchronous connection attempts [Section 5] 126 4. Establishment of one connection, which cancels all other attempts 128 Note that this document assumes that the host address preference 129 policy favors IPv6 over IPv4. If the host is configured differently, 130 the recommendations in this document can be easily adapted. 132 3. Hostname Resolution Query Handling 134 When a client has both IPv4 and IPv6 connectivity, and is trying to 135 establish a connection with a named host, it needs to send out both 136 AAAA and A DNS queries. Both queries SHOULD be made as soon after 137 one another as possible, with the AAAA query made first, immediately 138 followed by the A query. 140 Implementations MUST NOT wait for both families of answers to return 141 before attempting connection establishment. If one query fails to 142 return, or takes significantly longer to return, waiting for the 143 second address family can significantly delay the connection 144 establishment of the first one. Therefore, the client MUST treat DNS 145 resolution as asynchronous. Note that if the platform does not offer 146 an asynchronous DNS API, this behavior can be simulated by making two 147 separate synchronous queries on different threads, one per address 148 family. If the AAAA query returns first, the first IPv6 connection 149 attempt MUST be immediately started. If the A query returns first 150 due to reordering, the client SHOULD wait for a short time for the 151 AAAA response to ensure preference is given to IPv6 (it is common for 152 the AAAA response to follow the A response by a few milliseconds). 153 This delay will be referred to as the "Resolution Delay". The 154 RECOMMENDED value for the Resolution Delay is 50 milliseconds. If 155 the AAAA response is received within the Resolution Delay period, the 156 client MUST immediately start the IPv6 connection attempt. If, at 157 the end of the Resolution Delay period, the AAAA response has not 158 been received but the A response has been received, the client SHOULD 159 proceed to Sorting Addresses [Section 4] and staggered connection 160 attempts [Section 5] using only the IPv4 addresses returned so far. 161 If the AAAA response arrives while these connection attempts are in 162 progress, but before any connection has been established, then the 163 newly received IPv6 addresses are incorporated into the list of 164 available candidate addresses [Section 6] and the process of 165 connection attempts will continue with the IPv6 addresses added, 166 until one connection is established. 168 3.1. Handling Multiple DNS Server Addresses 170 If multiple DNS server addresses are configured for the current 171 network, the client may have the option of sending its DNS queries 172 over IPv4 or IPv6. In keeping with the Happy Eyeballs approach, 173 queries SHOULD be sent over IPv6 first (note that this is not 174 referring to the sending of AAAA or A queries, but rather the address 175 of the DNS server itself and IP version used to transport DNS 176 messages). If DNS queries sent to the IPv6 address do not receive 177 responses, that address may be marked as penalized, and queries can 178 be sent to other DNS server addresses. 180 As native IPv6 deployments become more prevalent, and IPv4 addresses 181 are exhausted, it is expected that IPv6 connectivity will have 182 preferential treatment within networks. If a DNS server is 183 configured to be accessible over IPv6, IPv6 should be assumed to be 184 the preferred address family. 186 Client systems SHOULD NOT have an explicit limit to the number of DNS 187 servers that can be configured, either manually or by the network. 188 If such a limit is required by hardware limitations, it is 189 RECOMMENDED to use at least one address from each address family from 190 the available list. 192 4. Sorting Addresses 194 Before attempting to connect to any of the resolved addresses, the 195 client should define the order in which to start the attempts. Once 196 the order has been defined, the client can use a simple algorithm for 197 racing each option after a short delay [Section 5]. It is important 198 that the ordered list involves all addresses from both families, as 199 this allows the client to get the racing effect of Happy Eyeballs for 200 the entire list, not just the first IPv4 and first IPv6 addresses. 202 First, the client MUST sort the addresses using Destination Address 203 Selection ([RFC6724], Section 6). 205 If the client is stateful and has history of expected round-trip 206 times (RTT) for the routes to access each address, it SHOULD add a 207 Destination Address Selection rule between rules 8 and 9 that prefers 208 addresses with lower RTTs. If the client keeps track of which 209 addresses it has used in the past, it SHOULD add another destination 210 address selection rule between the RTT rule and rule 9, which prefers 211 used addresses over unused ones. This helps servers that use the 212 client's IP address during authentication, as is the case for TCP 213 Fast Open ([RFC7413]) and some HTTP cookies. This historical data 214 MUST NOT be used across networks, and SHOULD be flushed on network 215 changes. 217 Next, the client SHOULD modify the ordered list to interleave address 218 families. Whichever address family is first in the list should be 219 followed by an address of the other address family; that is, if the 220 first address in the sorted list is IPv6, then the first IPv4 address 221 should be moved up in the list to be second in the list. An 222 implementation MAY want to favor one address family more by allowing 223 multiple addresses of that family to be attempted before trying the 224 other family. The number of contiguous addresses of the first 225 address family will be referred to as the "First Address Family 226 Count", and can be a configurable value. This is performed to avoid 227 waiting through a long list of addresses from a given address family 228 if connectivity over that address family is impaired. 230 5. Connection Attempts 232 Once the list of addresses has been constructed, the client will 233 attempt to make connections. In order to avoid unreasonable network 234 load, connection attempts SHOULD NOT be made simultaneously. 235 Instead, one connection attempt to a single address is started first, 236 followed by the others in the list, one at a time. Starting a new 237 connection attempt does not affect previous attempts, as multiple 238 connection attempts may occur in parallel. Once one of the 239 connection attempts succeeds (generally when the TCP handshake 240 completes), all other connections attempts that have not yet 241 succeeded SHOULD be cancelled. Any address that was not yet 242 attempted as a connection SHOULD be ignored. 244 A simple implementation can have a fixed delay for how long to wait 245 before starting the next connection attempt. This delay is referred 246 to as the "Connection Attempt Delay". One recommended value for this 247 delay is 250 milliseconds. If the client has historical RTT data, it 248 can also use the expected RTT to choose a more nuanced delay value. 249 The recommended formula for calculating the delay after starting a 250 connection attempt is: MAX( 1.25 * RTT_MEAN + 4 * RTT_VARIANCE, 2 * 251 RTT_MEAN ), where the RTT values are based on the statistics for 252 previous address used. If the TCP implementation leverages 253 historical RTT data to compute SYN timeout, these algorithms should 254 match so that a new attempt will be started at the same time as the 255 previous is sending its second TCP SYN. While TCP implementations 256 often leverage an exponential backoff when they detect packet loss, 257 the "Connection Attempt Delay" SHOULD NOT such an aggressive backoff, 258 as it would harm user experience. 260 The Connection Attempt Delay MUST have a lower bound, especially if 261 it is computed using historical data. More specifically, a 262 subsequent connection MUST NOT be started within 10 milliseconds of 263 the previous attempt. The recommended minimum value is 100 264 milliseconds, which is referred to as the "Minimum Connection Attempt 265 Delay". This minimum value is required to avoid congestion collapse 266 in the presence of high packet loss rates. The Connection Attempt 267 Delay SHOULD have an upper bound, referred to as the "Maximum 268 Connection Attempt Delay". The current recommended value is 2 269 seconds. 271 6. DNS Answer Changes during Happy Eyeballs Connection Setup 273 If, during the course of connection establishment, the DNS answers 274 change either by adding resolved addresses (for example, due to DNS 275 push notifications [DNS-PUSH]), or removing previously resolved 276 addresses (for example, due to expiry of the TTL on that DNS record), 277 the client should react based on its current progress. 279 If an address is removed from the list that already had a connection 280 attempt started, the connection attempt SHOULD NOT be cancelled, but 281 rather be allowed to continue. If the removed address had not yet 282 had a connection attempt started, it SHOULD be removed from the list 283 of addresses to try. 285 If an address is added to the list, it should be sorted into the list 286 of addresses not yet attempted according to the rules above 287 (Section 4). 289 7. Supporting IPv6-only Networks with NAT64 and DNS64 291 While many IPv6 transition protocols have been standardized and 292 deployed, most are transparent to client devices. The combined use 293 of NAT64 [RFC6146] and DNS64 [RFC6147] is a popular solution that is 294 being deployed and requires changes in client devices. One possible 295 way to handle these networks is for the client device networking 296 stack to implement 464XLAT [RFC6877]. 464XLAT has the advantage of 297 not requiring changes to user space software, however it requires 298 per-packet translation if the application is using IPv4 literals and 299 does not encourage client application software to support native 300 IPv6. On platforms that do not support 464XLAT, the Happy Eyeballs 301 engine SHOULD follow the recommendations in this section to properly 302 support IPv6-only networks with NAT64 and DNS64. 304 The features described in this section SHOULD only be enabled when 305 the host detects one of these networks. A simple heuristic to 306 achieve that is to check if the networks offers routable IPv6 307 addressing, does not offer routable IPv4 addressing, and offers a DNS 308 resolver address. 310 7.1. IPv4 Address Literals 312 If client applications or users wish to connect to IPv4 address 313 literals, the Happy Eyeballs engine will need to perform NAT64 314 address synthesis for them. The solution is similar to "Bump-in-the- 315 Host" [RFC6535] but is implemented inside the Happy Eyeballs library. 317 When an IPv4 address is passed in to the library instead of a host 318 name, the device queries the network for the NAT64 prefix using 319 "Discovery of the IPv6 Prefix Used for IPv6 Address Synthesis" 320 [RFC7050] then synthesizes an appropriate IPv6 address (or several) 321 using the encoding described in "IPv6 Addressing of IPv4/IPv6 322 Translators" [RFC6052]. The synthesized addresses are then inserted 323 into the list of addresses as if they were results from DNS queries; 324 connection attempts follow the algorithm described above (Section 5). 326 7.2. Host Names with Broken AAAA Records 328 At the time of writing, there exist a small but non negligible number 329 of host names that resolve to valid A records and broken AAAA 330 records, which we define as AAAA records that contain seemingly valid 331 IPv6 addresses but those addresses never reply when contacted on the 332 usual ports. These can be for example caused by: 334 o Mistyping of the IPv6 address in the DNS zone configuration 336 o Routing black holes 338 o Service outages 340 While an algorithm complying with the other sections of this document 341 would correctly handle such host names on a dual-stack network, they 342 will not necessarily function correctly on IPv6-only networks with 343 NAT64 and DNS64. Since DNS64 recursive resolvers rely on the 344 authoritative name servers sending negative ("no error no answer") 345 responses for AAAA records in order to synthesize, they will not 346 synthesize records for these particular host names, and will instead 347 pass through the broken AAAA record. 349 In order to support these scenarios, the client device needs to query 350 the DNS for the A record then perform local synthesis. Since these 351 types of host names are rare, and in order to minimize load on DNS 352 servers, this A query should only be performed when the client has 353 given up on the AAAA records it initially received. This can be 354 achieved by using a longer timeout, referred to as the "Last Resort 355 Local Synthesis Delay" and RECOMMENDED at 2 seconds. The timer is 356 started when the last connection attempt is fired. If no connection 357 attempt has succeeded when this timer fires, the device queries the 358 DNS for the IPv4 address and on reception of a valid A record, treats 359 it as if it were provided by the application (Section 7.1). 361 7.3. Virtual Private Networks 363 Some Virtual Private Networks (VPN) may be configured to handle DNS 364 queries from the device. The configuration could encompass all 365 queries, or a subset such as "*.internal.example.com". These VPNs 366 can also be configured to only route part of the IPv4 address space, 367 such as 192.0.2.0/24. However, if an internal hostname resolves to 368 an external IPv4 address, these can cause issues if the underlying 369 network is IPv6-only. As an example, let's assume that 370 "www.internal.example.com" has exactly one A record, 198.51.100.42, 371 and no AAAA records. The client will send the DNS query to the 372 company's recursive resolver and that resolver will reply with these 373 records. The device now only has an IPv4 address to connect to, and 374 no route to that address. Since the company's resolver does not know 375 the NAT64 prefix of the underlying network, it cannot synthesize the 376 address. Similarly, the underlying network's DNS64 recursive 377 resolver does not know the company's internal addresses, so it cannot 378 resolve the hostname. Because of this, the client device needs to 379 resolve the A record using the company's resolver then locally 380 synthesize an IPv6 address, as if the resolved IPv4 address were 381 provided by the application (Section 7.1). 383 8. Summary of Configurable Values 385 The values that may be configured as defaults on a client for use in 386 Happy Eyeballs are as follows: 388 o Resolution Delay (Section 3): The time to wait for a AAAA response 389 after receiving an A response. RECOMMENDED at 50 milliseconds. 391 o First Address Family Count (Section 4): The number of addresses 392 belonging to the first address family (such as IPv6) that should 393 be attempted before attempting another address family. 394 RECOMMENDED as 1, or 2 to more aggressively favor one address 395 family. 397 o Connection Attempt Delay (Section 5): The time to wait between 398 connection attempts in the absence of RTT data. RECOMMENDED at 399 250 milliseconds. 401 o Minimum Connection Attempt Delay (Section 5): The minimum time to 402 wait between connection attempts. RECOMMENDED at 100 403 milliseconds. MUST NOT be less than 10 milliseconds. 405 o Maximum Connection Attempt Delay (Section 5): The maximum time to 406 wait between connection attempts. RECOMMENDED at 2 seconds. 408 o Last Resort Local Synthesis Delay (Section 7.2): The time to wait 409 after starting the last IPv6 attempt and before sending the A 410 query. RECOMMENDED at 2 seconds. 412 As time advances, it is expected that the properties of networks will 413 evolve. For that reason, it is expected that these values will 414 change over time. Implementors should feel welcome to use different 415 values without changing this specification. In particular, IPv6 416 issues are expected to be less common, therefore the Resolution Delay 417 SHOULD be increased with time as client software is updated. 419 9. Limitations 421 Happy Eyeballs will handle initial connection failures at the TCP/IP 422 layer, however other failures or performance issues may still affect 423 the chosen connection. 425 9.1. Path Maximum Transmission Unit Discovery 427 Since Happy Eyeballs is only active during the initial handshake and 428 TCP does not pass the initial handshake, issues related to MTU can be 429 masked and go unnoticed during Happy Eyeballs. Solving this issue is 430 out of scope of this document. 432 9.2. Application Layer 434 If the DNS returns multiple addresses for different application 435 servers, the application itself may not be operational and functional 436 on all of them. Common examples include Transport Layer Security 437 (TLS) and the Hypertext Transport Protocol (HTTP). 439 9.3. Hiding Operational Issues 441 It has been observed in practice that Happy Eyeballs can hide issues 442 in networks. For example, if a misconfiguration causes IPv6 to 443 consistently fail on a given network while IPv4 is still functional, 444 Happy Eyeballs may impair the operator's ability to notice the issue. 445 It is recommended that network operators deploy external means of 446 monitoring to ensure functionality of all address families. 448 10. Security Considerations 450 This memo has no direct security considerations. 452 11. IANA Considerations 454 This memo includes no request to IANA. 456 12. Acknowledgments 458 The authors thank Dan Wing, Andrew Yourtchenko, and everyone else who 459 worked on the original Happy Eyeballs design ([RFC6555]), Josh 460 Graessley, Stuart Cheshire, and the rest of team at Apple that helped 461 implement and instrument this algorithm, and Jason Fesler and Paul 462 Saab who helped measure and refine this algorithm. The authors would 463 also like to thank Fred Baker, Nick Chettle, Lorenzo Colitti, Igor 464 Gashinsky, Geoff Huston, Jen Linkova, Paul Hoffman, Philip Homburg, 465 Erik Nygren, Jordi Palet Martinez, Rui Paulo, Jinmei Tatuya, Dave 466 Thaler, Joe Touch and James Woodyatt for their input and 467 contributions. 469 13. References 471 13.1. Normative References 473 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 474 Requirement Levels", BCP 14, RFC 2119, 475 DOI 10.17487/RFC2119, March 1997, 476 . 478 [RFC6052] Bao, C., Huitema, C., Bagnulo, M., Boucadair, M., and X. 479 Li, "IPv6 Addressing of IPv4/IPv6 Translators", RFC 6052, 480 DOI 10.17487/RFC6052, October 2010, 481 . 483 [RFC6146] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful 484 NAT64: Network Address and Protocol Translation from IPv6 485 Clients to IPv4 Servers", RFC 6146, DOI 10.17487/RFC6146, 486 April 2011, . 488 [RFC6147] Bagnulo, M., Sullivan, A., Matthews, P., and I. van 489 Beijnum, "DNS64: DNS Extensions for Network Address 490 Translation from IPv6 Clients to IPv4 Servers", RFC 6147, 491 DOI 10.17487/RFC6147, April 2011, 492 . 494 [RFC6535] Huang, B., Deng, H., and T. Savolainen, "Dual-Stack Hosts 495 Using "Bump-in-the-Host" (BIH)", RFC 6535, 496 DOI 10.17487/RFC6535, February 2012, 497 . 499 [RFC6555] Wing, D. and A. Yourtchenko, "Happy Eyeballs: Success with 500 Dual-Stack Hosts", RFC 6555, DOI 10.17487/RFC6555, April 501 2012, . 503 [RFC6724] Thaler, D., Ed., Draves, R., Matsumoto, A., and T. Chown, 504 "Default Address Selection for Internet Protocol Version 6 505 (IPv6)", RFC 6724, DOI 10.17487/RFC6724, September 2012, 506 . 508 [RFC7050] Savolainen, T., Korhonen, J., and D. Wing, "Discovery of 509 the IPv6 Prefix Used for IPv6 Address Synthesis", 510 RFC 7050, DOI 10.17487/RFC7050, November 2013, 511 . 513 13.2. Informative References 515 [DNS-PUSH] 516 Pusateri, T. and S. Cheshire, "DNS Push Notifications", 517 Work in Progress, draft-ietf-dnssd-push, March 2017. 519 [RFC6877] Mawatari, M., Kawashima, M., and C. Byrne, "464XLAT: 520 Combination of Stateful and Stateless Translation", 521 RFC 6877, DOI 10.17487/RFC6877, April 2013, 522 . 524 [RFC7413] Cheng, Y., Chu, J., Radhakrishnan, S., and A. Jain, "TCP 525 Fast Open", RFC 7413, DOI 10.17487/RFC7413, December 2014, 526 . 528 Appendix A. Differences from RFC6555 530 "Happy Eyeballs: Success with Dual-Stack Hosts" [RFC6555] mostly 531 concentrates on how to stagger connections to a hostname that has an 532 AAAA and an A record. This document additionally discusses: 534 o how to perform DNS queries to obtain these addresses 536 o how to handle multiple addresses from each address family 538 o how to handle DNS updates while connections are being raced 540 o how to leverage historical information 542 o how to support IPv6-only networks with NAT64 and DNS64 544 Authors' Addresses 546 David Schinazi 547 Apple Inc. 548 1 Infinite Loop 549 Cupertino, California 95014 550 US 552 Email: dschinazi@apple.com 553 Tommy Pauly 554 Apple Inc. 555 1 Infinite Loop 556 Cupertino, California 95014 557 US 559 Email: tpauly@apple.com