In message <5.2.0.9.1.20030526172452.055d4380(a)mail.igrin.co.nz>, Simon Byrnand writes:

The problem I can see with the introduction of ipv6 is that its such a fundamental change, a bit like converting an entire country from imperial to metric, except that you're trying to convert the WHOLE world :)

Which sounds bad, except when you realise that the process doesn't require a "flag day", and has been very carefully designed so that it can happen over the course of several years (dual stacks, tunnels, etc). The major thing which is leading to slow progress, IMHO, is the lack of a really compelling need for it. IPv4 + NAT is a 90% solution. (OTOH there's something approaching a compelling need ofr multicasting, and that hasn't exactly happened much since last year either.) I think you're probably right that it'll take a very long time before IPv4 will be completely gone. But there've been plenty of other network protocols and many of them are nearly extinct. ("Native" windows networking is nearly gone -- haven't seen it in ages -- and even IPX is looking endangered (Novell went IP a couple of versions back), without even looking at protocols which were much less widely deployed.) FWIW, I do still keep meaning to experiment some more with IPv6 given some spare time, especially as most (all?) of the operating systems I'm using support IPv6 out of the box (OpenBSD, Linux, Solaris, etc, all support it out of the box these days). Ewen