Re: [v6ops] Our IPv6-only home network and future experiments

[Vasilenko Eduard <[e-mail]>]

  *   Rather, the issue is what happens when a modern, well-written dual-stack application is running on an IPv6-only network, but needs to access an IPv4-only resource on the Internet.
“CLAT” is something additional to DNS64+NAT64”. “CLAT” is not needed for the above.

Ed/
From: Brian Candler <[e-mail]<mailto:[e-mail]>>
Sent: Monday, April 15, 2024 10:17
To: Vasilenko Eduard <[e-mail]<mailto:[e-mail]>>; Soni "They/Them" L. <[e-mail]<mailto:[e-mail]>>; IPv6 Operations <[e-mail]<mailto:[e-mail]>>
Subject: Re: [v6ops] Our IPv6-only home network and future experiments

On 15/04/2024 08:03, Vasilenko Eduard wrote:

The initiative looks strange for me.

It has an assumption that some *old* application insists to use IPv4 and this application *could not be changed* to use IPv6 properly.

No, that's not at all what the underlying driver of this is.

Rather, the issue is what happens when a modern, well-written dual-stack application is running on an IPv6-only network, but needs to access an IPv4-only resource on the Internet.

The proposed approach allows the application to do DNS lookups, find only an A record, open an IPv4 socket and make a connection (as it would do on a dual stack network) - but this is transparently rewritten to an IPv6 connection to a PLAT with the IPv4 address embedded in the destination IPv6 address - i.e. 464XLAT.

This approach is cleaner than DNS64, where you pretend that every IPv4-only resource has a AAAA record that points to your local PLAT. It also means that an application can make a connection to an IPv4 literal address (e.g. 1.1.1.1) and it works.