Biggest mistake for IPv6: It's not backwards compatible, developers admit

Benny Amorsen benny+usenet at amorsen.dk
Wed Apr 1 17:00:04 CEST 2009


Jeroen Massar <jeroen at unfix.org> writes:

> This new option field sets the version to 6, the NAT devices can parse
> this new version field and.....
>
> The moment you have to touch something to upgrade it, you need to
> upgrade them all.

No, you don't. If packets with that proposal encounter a NAT device
which doesn't like the option, the option will just be stripped and
traditional NAT occurs. Thus connectivity is preserved, with the
limitations of traditional IPv4 NAT.

> What you propose can also be accomplished by tunneling all IPv6 packets
> inside IPv4 ala 6to4/proto41. The "Extra bit" then is that the protocol
> field is set to 41 aka IPv6.

No, if you tunnel v6 inside v4 you need some way of detecting that the
party you are talking to understands v6. Also, you can't actually
establish that tunnel if both parties are behind NAT.

> If they are upgraded like that, you can also upgrade them to support
> IPv6. Unfortunately those things get replaced, not upgraded.

Replacement is fine too. Upgrading to support v6 doesn't help, because
you have noone to talk v6 to (until all the ISP's wake up). If two
parties with upgraded IP stacks behind upgraded/replaced NAT-boxes need
to talk, it will just work, no additional configuration necessary.


/Benny



More information about the ipv6-ops mailing list