[arin-ppml] What will be the end result of IPv4 exhaustion?

[michael.dillon at bt.com]

>  To put it another way, it 
> > makes it possible to extend the dual-stack phase of the v4-v6 
> > transition so that we aren't forcing some organizations to 
> turn off v4 
> > before everyone else has turned on v6.
> >
> You do realize that's a pipe dream, right?  There simply isn't enough
> IPv4
> and time to make that realistic, 

There are also sharks circling around. Some ISPs have MPLS in their
core and they don't need to dual stack. They can grow their v6 
infrastructure at the edges using 6PE and therefore they need
fewer IPv4 addresses overall to sustain growth.

If you don't have an MPLS core, and can't transition to one,
then it would be risky to rely on dual-stack when one half
of the addresses that you need, may not be available any more.

There are other, more creative ways of adding IPv6 at the edges
that roughly emulates what 6PE does. It is safer to only dual-stack
a fixed portion of your core, and use it to connect IPv6 islands
at your edge. The IPv6 islands can then grow without the need for
additional v4 addresses. Dual-stack is far from a panacea.

--Michael Dillon