[ppml] PIv6 for legacy holders (/w RSA + efficient use)

[Paul Vixie]

> there will be five stages at the edges, i will steal from the yet
> released next slide set in the series.
>   0 Denial, from both ‘sides’:
>     . We can ignore brain-dead IPv6
>     . IPv6 is perfect and those greedy fools just have to deploy it
>   1 Dual stack with IPv4 Dominant
>   2 Dual stack with both widely used
>   3 Dual stack with IPv6 Dominant
>   4 The IPv6 Internet (getting ready for IPv10 transition:)
>
> i assume dual stack core before we move an inch, i.e. out of stage 0.  and
> we are getting the dual stack core now.  transit providers who don't make
> that move will see it on their bottom line in one or two years.  one problem
> is that (some) router vendor support is still mediocre, so stalling
> increases the value of your capital.

i don't understand that final sentence.

> in stages 1 and 2, there will be massive use of v4/v4 nat and v4/v6 nat.  we
> can hope that v4/v4 slowly dies away in stage 3, as v6 technologic barriers
> are significantly lower financially, inter-operation is widespread, and
> acquisition of new ipv4 space becomes a more and more expensive proposition,

other than completing buildouts which were already planned and in progress
when IPv4 depletion occurs, i'm not sure what force will drive IPv4 space to
a higher price.  the value of an address is that you can reach other people
with it, and if other people can't grow in IPv4, then why would you care to?
(other than as i said, if it's V4-only infrastructure that you already had in
the deployment queue before the upcoming depletion event.)

> but at no time will growth be inhibited.  as we all know, the internet
> routes around blockage.  and growth will find the currently least cost path.

i am now completely off the rails of this message.  for one thing you seem to
be assuming that overcoming alternative cost is itself a cost-free event and
that it will be instantaneous and that the "market" will have perfect
knowledge.  for another thing you seem to assume that the step function in
value (customer's reachability) and cost (training, equipment, operations)
for those who choose an IPv6 alternative _as a result of_ IPv4 "costs" rather
than having planned for it, will be so low as to not be called a catastrophy.

> the key thing is to reduce the _technology_ expense of choosing v6 over v4
> in stages 1 and 2.  and for that, you have to wait for me to finish the next
> preso :).

"we're not going to win this with torpedoes, chief."  but lay it on me anyway.