RFC 8305

Happy Eyeballs Version 2: Better Connectivity Using Concurrency

📅 published December 2017, from ietf

Happy Eyeballs Version 2: Better Connectivity Using Concurrency ✨

Internet Engineering Task Force (IETF) 🦋 · Request for Comments: 8305 · Obsoletes: 6555 · Category: Standards Track · December 2017 · ISSN: 2070-1721

Okay so, this RFC is the big "Happy Eyeballs v2" glow-up 💖 It was written by D. Schinazi and T. Pauly from Apple Inc. 🍎 It's a Standards Track document, which means it's the real deal, not just a vibe. It was published in December 2017 🎀 and it obsoletes RFC 6555, which was the original Happy Eyeballs. So this is basically the upgraded version, bestie 🫶

Status of This Memo

This is an Internet Standards Track document. 💅

This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 7841.

Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc8305.

Copyright (c) 2017 IETF Trust and the persons identified as the document authors. All rights reserved. 🌈

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

Table of Contents

1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
  1.1.  Requirements Language . . . . . . . . . . . . . . . . . .   3
2.  Overview  . . . . . . . . . . . . . . . . . . . . . . . . . .   4
3.  Hostname Resolution Query Handling  . . . . . . . . . . . . .   4
  3.1.  Handling Multiple DNS Server Addresses  . . . . . . . . .   5
4.  Sorting Addresses . . . . . . . . . . . . . . . . . . . . . .   6
5.  Connection Attempts . . . . . . . . . . . . . . . . . . . . .   7
6.  DNS Answer Changes during Happy Eyeballs Connection Setup . .   8
7.  Supporting IPv6-Only Networks with NAT64 and DNS64  . . . . .   8
  7.1.  IPv4 Address Literals . . . . . . . . . . . . . . . . . .   8
  7.2.  Hostnames with Broken AAAA Records  . . . . . . . . . . .   9
  7.3.  Virtual Private Networks  . . . . . . . . . . . . . . . .  10
8.  Summary of Configurable Values  . . . . . . . . . . . . . . .  10
9.  Limitations . . . . . . . . . . . . . . . . . . . . . . . . .  11
  9.1.  Path Maximum Transmission Unit Discovery  . . . . . . . .  11
  9.2.  Application Layer . . . . . . . . . . . . . . . . . . . .  11
  9.3.  Hiding Operational Issues . . . . . . . . . . . . . . . .  11
10. Security Considerations . . . . . . . . . . . . . . . . . . .  12
11. IANA Considerations . . . . . . . . . . . . . . . . . . . . .  12
12. References  . . . . . . . . . . . . . . . . . . . . . . . . .  12
  12.1.  Normative References . . . . . . . . . . . . . . . . . .  12
  12.2.  Informative References . . . . . . . . . . . . . . . . .  13
Appendix A.  Differences from RFC 6555  . . . . . . . . . . . . .  14
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . .  15
Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  15

1. Introduction 🌸

Okay, so the vibe of the modern Internet is that lots of apps use hostnames 💻 and those hostnames often resolve to multiple IP addresses, and each one can have different performance and connectivity. 😭 Specific addresses or address families (IPv4 or IPv6) might be blocked, broken, or just sub-optimal on a network. So clients that try multiple connections in parallel have a chance of connecting way faster 🏃‍♀️💨 This document lays out the requirements for algorithms that reduce this user-visible delay, and it gives an example algorithm, which is called "Happy Eyeballs" 👀✨

This document defines the algorithm for "Happy Eyeballs", a technique for reducing user-visible delays on dual-stack hosts. 🎀 This definition obsoletes the original description in [RFC6555]. Now that this approach has been deployed at scale and measured for several years, the algorithm specification can be refined to improve its reliability and general applicability. It's giving "we learned from the first try" 💅

The Happy Eyeballs algorithm of racing connections to resolved addresses has several stages to avoid delays to the user whenever possible, while preferring the use of IPv6. 🦋 This document covers how to handle DNS queries when starting a connection on a dual-stack client, how to create an ordered list of destination addresses to try, and how to race the connection attempts. Main character energy, honestly 👑

1.1. Requirements Language

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here. 📝 (Translation: the words only count as rules when they're in ALL CAPS, bestie. Lowercase "should" is just chatting 🙈)

2. Overview 🧚

This document defines a method of connection establishment, named the "Happy Eyeballs Connection Setup". 💖 It has several distinct phases:

  1. Initiation of asynchronous DNS queries [Section 3]
  2. Sorting of resolved destination addresses [Section 4]
  3. Initiation of asynchronous connection attempts [Section 5]
  4. Establishment of one connection, which cancels all other attempts [Section 5]

Note that this document assumes that the preference policy for the host destination address favors IPv6 over IPv4. IPv6 has many desirable properties designed to be improvements over IPv4 [RFC8200]. 💜 If the host is configured to have a different preference, the recommendations in this document can be easily adapted. ✨

3. Hostname Resolution Query Handling 🍓

Okay so when a client has both IPv4 and IPv6 connectivity and wants to connect to a named host, it needs to send out both AAAA and A DNS queries. 📬 Both queries SHOULD be made as soon after one another as possible, with the AAAA query made first and immediately followed by the A query.

Implementations SHOULD NOT wait for both families of answers to return before attempting connection establishment. 🙅‍♀️ If one query fails to return or takes significantly longer to return, waiting for the second address family can significantly delay the connection establishment of the first one. Therefore, the client SHOULD treat DNS resolution as asynchronous. Note that if the platform does not offer an asynchronous DNS API, this behavior can be simulated by making two separate synchronous queries on different threads, one per address family. 🧵

The algorithm goes like this: if a positive AAAA response (a response with at least one valid AAAA record) is received first, the first IPv6 connection attempt is immediately started. 🚀 If a positive A response is received first due to reordering, the client SHOULD wait a short time for the AAAA response to make sure IPv6 gets the preference (it's super common for the AAAA response to trail the A response by a few milliseconds 🙄). This delay is called the "Resolution Delay". The recommended value for the Resolution Delay is 50 milliseconds. ⏱️ If a positive AAAA response is received within the Resolution Delay period, the client immediately starts the IPv6 connection attempt. 💅 If a negative AAAA response (no error, no data) is received within the Resolution Delay period, or the AAAA response hasn't been received by the end of the Resolution Delay period, the client SHOULD proceed to sorting addresses (see Section 4) and staggered connection attempts (see Section 5) using any IPv4 addresses returned so far. 🏳️‍🌈

If the AAAA response arrives while these connection attempts are in progress but before any connection has been established, then the newly received IPv6 addresses are added to the list of available candidate addresses (see Section 6), and the process of connection attempts continues with the IPv6 addresses added, until one connection is established. 🦄 Iconic behavior, no notes 🫶

3.1. Handling Multiple DNS Server Addresses 🌷

If multiple DNS server addresses are configured for the current network, the client MAY have the option of sending its DNS queries over IPv4 or IPv6. 🤷‍♀️ In keeping with the Happy Eyeballs approach, queries SHOULD be sent over IPv6 first (note that this is not talking about sending AAAA or A queries, but rather the address of the DNS server itself and the IP version used to transport DNS messages 🧠). If DNS queries sent to the IPv6 address don't get responses, that address MAY be marked as penalized and queries can be sent to other DNS server addresses. 🚫

As native IPv6 deployments get more common and IPv4 addresses run out, it is expected that IPv6 connectivity will get preferential treatment within networks. 📈 If a DNS server is configured to be accessible over IPv6, IPv6 should be assumed to be the preferred address family. (lowercase "should" here, so it's just a vibe, not a rule 💀)

Client systems SHOULD NOT have an explicit limit to the number of DNS servers that can be configured, either manually or by the network. 🙌 If such a limit is required by hardware limitations, the client SHOULD use at least one address from each address family from the available list. 💖

Part 1 of 3 done, bestie 💜 Stay tuned for part 2 🎀✨

4. Sorting Addresses ✨

okay bestie, so before the client even thinks about dialing any of the resolved destination addresses, it should decide the ORDER it's gonna try them in 📋💖 once that order is set, it can race each option after a short little delay (see Section 5) 🏃‍♀️💨 and it's SUPER important that the ordered list includes every address from both families that's been received by this point 🫶 because that's what gives you the racing effect for the WHOLE list, not just the first IPv4 and first IPv6 address. main character energy for the entire list, no cap 👑

First, the client MUST sort the addresses received up to this point using Destination Address Selection ([RFC6724], Section 6). 💅

If the client is stateful and has a history of expected round-trip times (RTTs) for the routes to access each address, it SHOULD add a Destination Address Selection rule between rules 8 and 9 that prefers addresses with lower RTTs 🌸 and if the client keeps track of which addresses it used in the past, it SHOULD add another Destination Address Selection rule between the RTT rule and rule 9, which prefers used addresses over unused ones 🦋 this helps servers that use the client's IP address during authentication, as is the case for TCP Fast Open [RFC7413] and some Hypertext Transport Protocol (HTTP) cookies 🍓 BUT this historical data MUST NOT be used across different network interfaces 🙈 and it SHOULD be flushed whenever a device changes the network it's attached to. lowkey a must-do 💀

Next, the client SHOULD modify the ordered list to interleave address families 🎀 whichever family is first in the list gets followed by an address from the OTHER family. so if the first address in the sorted list is IPv6, then the first IPv4 address gets moved up to be second in the list ✨ an implementation MAY want to favor one family more by letting multiple addresses of that family get attempted before trying the other family 🧚 the number of contiguous addresses of the first address family is called the "First Address Family Count" and it can be a configurable value 💜 this is done so you don't have to sit through a long list of addresses from a family when connectivity over that family is impaired 😭

Note that the address selection described in this section ONLY applies to destination addresses 💅 Source Address Selection ([RFC6724], Section 5) is performed once per destination address and is out of scope of this document, so we're not talking about it here 🫶

5. Connection Attempts 🚀

Once the list of addresses received up to this point has been constructed, the client will start making connections 🔥 to avoid unreasonable network load, connection attempts SHOULD NOT be made simultaneously 🌈 instead, one connection attempt to a single address starts first, and then the others in the list follow, one at a time. starting a new attempt does NOT affect the previous ones, since multiple attempts may happen in parallel 🦋 once one of the attempts succeeds (generally when the TCP handshake completes), all other attempts that haven't succeeded yet SHOULD be canceled 🛑 any address that wasn't attempted yet SHOULD be ignored. at that point the asynchronous DNS query MAY be canceled, since new addresses won't be used for this connection 🙅‍♀️ HOWEVER, the DNS client resolver SHOULD still process DNS replies from the network for a short period of time (recommended to be 1 second) 🕐 because they'll populate the DNS cache and can be used for later connections. big brain move 🧠✨

A simple implementation can have a fixed delay for how long to wait before starting the next connection attempt 💖 this delay is called the "Connection Attempt Delay". one recommended value for a default delay is 250 milliseconds ⏱️ a more nuanced implementation's delay should match the time when the previous attempt sends its second TCP SYN, based on the TCP's retransmission timer [RFC6298] 🎀 if the client has historical RTT data gathered from other connections to the same host or prefix, it can use that to influence its delay. and note that this algorithm should ONLY try to approximate the time of the first SYN retransmission, not any further retransmissions that get influenced by exponential timer back off 🙈

The Connection Attempt Delay MUST have a lower bound, especially if it's computed using historical data 💅 more specifically, a subsequent connection MUST NOT be started within 10 milliseconds of the previous attempt. the recommended minimum value is 100 milliseconds, and that's called the "Minimum Connection Attempt Delay" 🍓 this minimum value is REQUIRED to avoid congestion collapse when packet-loss rates are high 😭 the Connection Attempt Delay SHOULD have an upper bound, called the "Maximum Connection Attempt Delay". the current recommended value is 2 seconds ✨

6. DNS Answer Changes during Happy Eyeballs Connection Setup 🦋

okay so, if DNS answers change DURING connection setup, either by adding resolved addresses (for example due to DNS push notifications [DNS-PUSH]) or by removing previously resolved addresses (for example, when the TTL on that DNS record expires) 🕰️ the client should react based on how far along it is 🌸

If an address gets removed from the list and a connection attempt was already started for it, the connection attempt SHOULD NOT be canceled. it should be allowed to continue 🏃‍♀️ but if the removed address hadn't had a connection attempt started yet, it SHOULD be removed from the list of addresses to try 🗑️

If an address gets added to the list, it should be sorted into the list of addresses not yet attempted according to the rules above (see Section 4) 💜

7. Supporting IPv6-Only Networks with NAT64 and DNS64 🌐

ngl, while lots of IPv6 transition protocols have been standardized and deployed, most are transparent to client devices 🫠 the combined use of NAT64 [RFC6146] and DNS64 [RFC6147] is a popular solution being deployed, and it DOES require changes in client devices 🔧 one possible way to handle these networks is for the client device's networking stack to implement 464XLAT [RFC6877]. 464XLAT has the advantage of not requiring changes to user space software 🙌 however, it requires per-packet translation if the application uses IPv4 literals, and it doesn't encourage client application software to support native IPv6 😬 on platforms that DON'T support 464XLAT, the Happy Eyeballs engine SHOULD follow the recommendations in this section to properly support IPv6-only networks with NAT64 and DNS64 💖

The features described in this section SHOULD only be enabled when the host detects one of these networks 👀 a simple heuristic for that is to check whether the network offers routable IPv6 addressing, does NOT offer routable IPv4 addressing, and offers a DNS resolver address ✅

7.1. IPv4 Address Literals 🔢

If client applications or users want to connect to IPv4 address literals, the Happy Eyeballs engine will need to perform NAT64 address synthesis for them 🪄 the solution is similar to "Bump-in-the-Host" [RFC6535] but it's implemented inside the Happy Eyeballs library 💅

When an IPv4 address is passed into the library instead of a hostname, the device queries the network for the NAT64 prefix using "Discovery of the IPv6 Prefix Used for IPv6 Address Synthesis" [RFC7050] 🔍 and then synthesizes an appropriate IPv6 address (or several) using the encoding described in "IPv6 Addressing of IPv4/IPv6 Translators" [RFC6052] ✨ the synthesized addresses are then inserted into the list of addresses as if they came from DNS queries, and connection attempts follow the algorithm described above (see Section 5) 🫶

7.2. Hostnames with Broken AAAA Records 💔

At the time of writing, there exists a small but non-negligible number of hostnames that resolve to valid A records and broken AAAA records 😭 a broken AAAA record is one that contains seemingly valid IPv6 addresses, but those addresses never reply when contacted on the usual ports 🙈 this can be caused by, for example:

  • Mistyping of the IPv6 address in the DNS zone configuration
  • Routing black holes
  • Service outages

While an algorithm complying with the other sections of this document would correctly handle such hostnames on a dual-stack network, they will NOT necessarily work correctly on IPv6-only networks with NAT64 and DNS64 🫣 since DNS64 recursive resolvers rely on the authoritative name servers sending negative ("no error no answer") responses for AAAA records in order to synthesize, they won't synthesize records for these particular hostnames and will instead pass through the broken AAAA record 💀

To support these scenarios, the client device needs to query the DNS for the A record and then perform local synthesis 🛠️ since these types of hostnames are rare, and to minimize load on DNS servers, this A query should ONLY be performed when the client has given up on the AAAA records it initially received 🥺 this can be achieved by using a longer timeout, called the "Last Resort Local Synthesis Delay"; the delay is recommended to be 2 seconds ⏳ the timer starts when the last connection attempt is fired. if no connection attempt has succeeded when this timer fires, the device queries the DNS for the IPv4 address and, on receiving a valid A record, treats it as if the application had provided it (see Section 7.1) 🎉

7.3. Virtual Private Networks 🔐

Some Virtual Private Networks (VPNs) may be configured to handle DNS queries from the device 🕵️‍♀️ the configuration could cover all queries or a subset, like "*.internal.example.com". these VPNs can also be configured to route only part of the IPv4 address space, such as 192.0.2.0/24. however, if an internal hostname resolves to an external IPv4 address, this can cause issues if the underlying network is IPv6-only 😵‍💫 as an example, let's say "www.internal.example.com" has exactly one A record, 198.51.100.42, and no AAAA records 📝 the client sends the DNS query to the company's recursive resolver, and that resolver replies with these records. now the device only has an IPv4 address to connect to and NO route to that address 🚫 since the company's resolver doesn't know the NAT64 prefix of the underlying network, it can't synthesize the address. similarly, the underlying network's DNS64 recursive resolver doesn't know the company's internal addresses, so it can't resolve the hostname either 💀 because of this, the client device needs to resolve the A record using the company's resolver and then locally synthesize an IPv6 address, as if the resolved IPv4 address were provided by the application (Section 7.1) 🫶

8. Summary of Configurable Values ✨

okay so bestie, this is the cheat sheet 🎀 every value a client can set as a default for Happy Eyeballs, all in one place. the way these numbers were picked is so iconic: they were measured on a huge pile of real production devices 📱 to cut down the waiting users notice while not hammering the network 💖

Resolution Delay (Section 3): the time to wait for a AAAA response after you get an A response. Recommended to be 50 milliseconds. 🦋

First Address Family Count (Section 4): how many addresses from the first address family (like IPv6) get tried before you switch over to trying another family. Recommended to be 1. You can use 2 if you want to lean harder into one family 💅

Connection Attempt Delay (Section 5): the time to wait between connection attempts when there's no RTT data. Recommended to be 250 milliseconds. ⭐

Minimum Connection Attempt Delay (Section 5): the minimum time to wait between connection attempts. Recommended to be 100 milliseconds. MUST NOT be less than 10 milliseconds. 🫶

Maximum Connection Attempt Delay (Section 5): the maximum time to wait between connection attempts. Recommended to be 2 seconds. 🌸

Last Resort Local Synthesis Delay (Section 7.2): the time to wait after starting the last IPv6 attempt and before sending the A query. Recommended to be 2 seconds. 🧚

These numbers are not set in stone 🙈 networks keep changing, so the values are expected to change over time. Implementors are totally welcome to use different values without changing this spec. Also, IPv6 issues are expected to be less common, so the delays SHOULD be increased with time as client software gets updated 🌈

9. Limitations 💀

Happy Eyeballs handles initial connection failures at the TCP/IP layer 🧦 but other failures or performance issues can still mess with the connection you ended up with.

9.1. Path Maximum Transmission Unit Discovery 🍓

Happy Eyeballs only runs during the initial handshake, and TCP doesn't pass that initial handshake, so MTU problems can get masked and go totally unnoticed during Happy Eyeballs. Fixing this is out of scope here. One possible solution is "Packetization Layer Path MTU Discovery" [RFC4821] ✨

9.2. Application Layer 🎀

If DNS returns multiple addresses for different application servers, the application might not actually work on all of them. Common examples are Transport Layer Security (TLS) and HTTP 🙈

9.3. Hiding Operational Issues 👀

It's been seen in practice that Happy Eyeballs can hide issues in networks. Like, if a misconfiguration makes IPv6 consistently fail on a network while IPv4 still works, Happy Eyeballs might make it harder for the operator to notice 😭 It is recommended that network operators deploy external means of monitoring to make sure all address families are functional 🔥

10. Security Considerations 🔐

okay this part is SERIOUS bestie 💀 applications should NOT rely on a stable hostname-to-address mapping to get any security properties, since DNS results may change between queries. Happy Eyeballs may make it more likely that later connections to a single hostname use different IP addresses. Don't build your security on that mapping staying put 🙅‍♀️

11. IANA Considerations 📋

This document does not require any IANA actions ✨

12. References 📚

12.1. Normative References 💜

[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/RFC2119, March 1997, <https://www.rfc-editor.org/info/rfc2119>.

[RFC4821] Mathis, M. and J. Heffner, "Packetization Layer Path MTU Discovery", RFC 4821, DOI 10.17487/RFC4821, March 2007, <https://www.rfc-editor.org/info/rfc4821>.

[RFC6052] Bao, C., Huitema, C., Bagnulo, M., Boucadair, M., and X. Li, "IPv6 Addressing of IPv4/IPv6 Translators", RFC 6052, DOI 10.17487/RFC6052, October 2010, <https://www.rfc-editor.org/info/rfc6052>.

[RFC6146] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful NAT64: Network Address and Protocol Translation from IPv6 Clients to IPv4 Servers", RFC 6146, DOI 10.17487/RFC6146, April 2011, <https://www.rfc-editor.org/info/rfc6146>.

[RFC6147] Bagnulo, M., Sullivan, A., Matthews, P., and I. van Beijnum, "DNS64: DNS Extensions for Network Address Translation from IPv6 Clients to IPv4 Servers", RFC 6147, DOI 10.17487/RFC6147, April 2011, <https://www.rfc-editor.org/info/rfc6147>.

[RFC6298] Paxson, V., Allman, M., Chu, J., and M. Sargent, "Computing TCP's Retransmission Timer", RFC 6298, DOI 10.17487/RFC6298, June 2011, <https://www.rfc-editor.org/info/rfc6298>.

[RFC6535] Huang, B., Deng, H., and T. Savolainen, "Dual-Stack Hosts Using "Bump-in-the-Host" (BIH)", RFC 6535, DOI 10.17487/RFC6535, February 2012, <https://www.rfc-editor.org/info/rfc6535>.

[RFC6555] Wing, D. and A. Yourtchenko, "Happy Eyeballs: Success with Dual-Stack Hosts", RFC 6555, DOI 10.17487/RFC6555, April 2012, <https://www.rfc-editor.org/info/rfc6555>.

[RFC6724] Thaler, D., Ed., Draves, R., Matsumoto, A., and T. Chown, "Default Address Selection for Internet Protocol Version 6 (IPv6)", RFC 6724, DOI 10.17487/RFC6724, September 2012, <https://www.rfc-editor.org/info/rfc6724>.

[RFC7050] Savolainen, T., Korhonen, J., and D. Wing, "Discovery of the IPv6 Prefix Used for IPv6 Address Synthesis", RFC 7050, DOI 10.17487/RFC7050, November 2013, <https://www.rfc-editor.org/info/rfc7050>.

[RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, May 2017, <https://www.rfc-editor.org/info/rfc8174>.

12.2. Informative References 🦋

[DNS-PUSH] Pusateri, T. and S. Cheshire, "DNS Push Notifications", Work in Progress, draft-ietf-dnssd-push-13, October 2017.

[RFC6877] Mawatari, M., Kawashima, M., and C. Byrne, "464XLAT: Combination of Stateful and Stateless Translation", RFC 6877, DOI 10.17487/RFC6877, April 2013, <https://www.rfc-editor.org/info/rfc6877>.

[RFC7413] Cheng, Y., Chu, J., Radhakrishnan, S., and A. Jain, "TCP Fast Open", RFC 7413, DOI 10.17487/RFC7413, December 2014, <https://www.rfc-editor.org/info/rfc7413>.

[RFC8200] Deering, S. and R. Hinden, "Internet Protocol, Version 6 (IPv6) Specification", STD 86, RFC 8200, DOI 10.17487/RFC8200, July 2017, <https://www.rfc-editor.org/info/rfc8200>.

Appendix A. Differences from RFC 6555 📝

"Happy Eyeballs: Success with Dual-Stack Hosts" [RFC6555] mostly focuses on how to stagger connections to a hostname that has both an AAAA and an A record. This document also covers:

  • how to perform DNS queries to get these addresses
  • how to handle multiple addresses from each address family
  • how to handle DNS updates while connections are being raced
  • how to leverage historical information
  • how to support IPv6-only networks with NAT64 and DNS64

Note that a simple implementation of the algorithm described in this document is still compliant with the previous specification [RFC6555]. Implementations should take the new considerations into account when they apply, to optimize their behavior 🫶

Acknowledgments 🙏

The authors thank Dan Wing, Andrew Yourtchenko, and everyone else who worked on the original Happy Eyeballs design [RFC6555], Josh Graessley, Stuart Cheshire, and the rest of the team at Apple that helped implement and instrument this algorithm, and Jason Fesler and Paul Saab who helped measure and refine this algorithm. The authors would also like to thank Fred Baker, Nick Chettle, Lorenzo Colitti, Igor Gashinsky, Geoff Huston, Jen Linkova, Paul Hoffman, Philip Homburg, Warren Kumari, Erik Nygren, Jordi Palet Martinez, Rui Paulo, Stephen Strowes, Jinmei Tatuya, Dave Thaler, Joe Touch, and James Woodyatt for their input and contributions 💖

Authors' Addresses 💌

David Schinazi Apple Inc. 1 Infinite Loop Cupertino, California 95014 United States of America

Email: dschinazi@apple.com

Tommy Pauly Apple Inc. 1 Infinite Loop Cupertino, California 95014 United States of America

Email: tpauly@apple.com

📦 where this page comes from