On 10-okt-2007, at 15:31, Thomas Narten wrote:
If there was a magic "trivial" transition/upgrade strategy, we would have done it years ago.
How about this: we magically go back 10 years in time and define "IPv4.6" which is the first 32 bits of the IPv6 address in the IPv4 address fields and the other 96 bits in a new header that sits between the IPv4 header and the upper layer protocol (TCP/UDP/etc). IPv4 stacks are updated to copy back the extra header with the lower 96 bits whenever sending packets in reponse to ones with those 96 bits in them.
I'm pretty sure we'd still have issues with unupgraded TCP stacks and firewalls, but just implementing the simple copyback mechanism and adjusting firewalls accordingly would be a much simpler way for existing IPv4 users to be able to talk to IPv6 users than adding IPv6 to their networks. Applications on the IPv4 host that are unaware of the fact that they're talking to an IPv6 host would still have NAT issues, but it would also be possible to run applications that know about the IPv6 address of the correspondent so the communication would be end-to-end clean.
(Ok there would have to be some extra bit shuffling to avoid using up the IPv6 address space as fast as the IPv4 space but that's just details. Maybe copy bits 4 - 31 and set bits 0 - 3 to 1111 in the IPv4 address if they are 0010 in the IPv6 address.)
_______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf