> >> Let's assume ... that a large part of the Internet is going to
> >> continue to be IPv4-only. ...
you've lost me. where will growth occur? two variables: "nat?" and
"stack?" so, six possibilities:
nat? stack?
---- ------
nat v4
nat v6
nat v4+v6
nonat v4
nonat v6
nonat v4+v6
given the relative ease of acquiring v6 address space and the relative
ease of deploying v4+v6 end hosts and either v4+v6 campuses or v6 tunnels
in v4 campuses, there is no incentive to do nat/v4 any more, and precious
little incentive to do nonat/v4. growth can be expected to occur only in
the other four tuples, with most falling in nonat/v4+v6 and nat/v4+v6.
not only growth of new hosts but metagrowth by upgrade of existing hosts.
therefore after a middle state of perhaps five more years, the majority
of services that anybody will want to access will be v4+v6 reachable, and
it will be realistic to consider provisioning first nat/v6 and then nonat/v6
endhosts.
very shortly thereafter, v4-only hosts will decline in value, both on the
server side and the client side (and both sides of peer-to-peer).
v4 will last as long as it's useful.
--
Paul Vixie
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf