Tony;
> > The only real problems are that with IPv4 we
> > allowed the IP addresses to be configured everywhere ...
>
> If *configuration* were the 'only' problem it might be possible
> to fix renumbering. The fact that applications expect they need
> to know about the addresses in use will compound the problem
> such that it will be very difficult if not impossible to make
> renumbering completely transparent.
Are you talking about ancient whois implementation?
> I have no doubt we can find
> a way to completely automate renumbering, but I seriously doubt
> that we can 'fix' all of the application developers, and their=20
> products. Given this state, the end user will be exposed to
> renumbering events. We either accept this and find a way to
> scale routing without renumbering, or accept that NAT will
> persist.
Or, are you saying an NAT server can be detached from old ISP and
attached to new ISP keeping the existing connections?
Masataka Ohta
PS
With end-to-end multihoming, it is possible that application/transport
program periodically check DNS to be smoothly renumbered.
