[ppml] Random v6 discussions (was Re: Policy Proposal: IPv4 Transfer Policy Proposal)

[David Williamson]

On Thu, Feb 14, 2008 at 08:49:52AM -0600, Kevin Kargel wrote:
> 	As customer connections are the primary consumer of IP's for an
> ISP, utilizing IPv6 for this will allow IP's to be reclaimed for use for
> global connections.

That's one of the key problems, isn't it?  The place you least want to
disrupt anything is the very edge.  We're talking about mucking up
layer 3.  That's the network layer...that bit that's supposed to
provide end-to-end connectivity.  Everything below isn't worried about
reaching beyond the local network.  Everything above is assuming the
network layer has done its job.  You can muck about in the middle of a
connection (by encapsulation, labelling, etc.), but the end points are
going to care about that end-to-end connectivity.

Much as I like the idea of "fix the edge so that we have more time to
fix the middle", I think it's going to be much more practical to fix
the middle, and dual-stack the edge until v6 is viable end-to-end on
its own.  Oh, and there's no way that's going to happen without massive
impact to the whole network, since getting the edge fixed will
definitely take longer than the remaining time before pool exhaustion,
no matter how you calculate it.  That implies some quality time with
protocol translators...pick your favorite 4-to-6 conversion method.

It stuns me that there are serious networking folks who don't think
we'll run out of v4 addresses.  But this is now waaaaaay off the topic
of policy, which is the purpose of this list.

-David