[arin-discuss] IPv6-only website

[Robert E. Seastrom]

Steve Bertrand <steve at ibctech.ca> writes:

> On 2010.04.23 00:25, Paul Carlson wrote:
>> 
>> 
>> I think that was part of what Steve was exploring.  A new network
>> would not necessarily have access to any IPV4 addresses after v4
>> exhaustion.  If v4 addresses were required to function in a v6
>> world then it would certainly change allocation rules for the last
>> of the addresses.
>
> Thanks for clarifying.
>
> You are 100% correct.
>
> This exploration test is not designed to be a simulation of what might
> happen... in fact, it is purely a real-world non-simulation
> configuration setup of what *will* happen.

I'm not sure this is a valid simulation or configuration setup, unless
you're trying to show what will happen when naive sysadmins fail to
configure DNS zones for greatest reachability, for the greatest number
of customers (which can obviously be done today if one wishes!).

This zone is, however, quite useful for sussing out whether a
particular cacheing recursive server you're talking to is dual-stack
(as suspected, BIND anyway will return SERVFAIL instead of NOERROR as
it tries to follow the delegation chain into address space that it
doesn't know about).  So thanks for setting it up.  OpenDNS, Google
and Level(3) folks, please pick up the white courtesy phone...

Consider two things:

1) Even if your ISP doesn't have IPv4 addresses at all, it is
reasonable to postulate the evolution of public resolvers along the
lines of 208.67.220.220, 208.67.220.222, 4.2.2.1 and 8.8.8.8 (none of
which is apparently currently dual-stacked) to be dual-stacked and so
be queryable from the v6 side as well as the v4 side and able resolve
records where the authoritative nameservers are in either or both
address spaces.

2) If you're looking for a place to host your own DNS records and have
no v4 address of your own, there are already places that are dirt
cheap to free (perhaps your registrar already offers this service?) to
host DNS records.  Just make sure that they offer at least one v6
address in the mix and voila, all is well; you are reachable even for
people whose cacheing resolver is v4-only.

Could you elucidate how this relates to a post-depletion scenario,
except by way of illustrating what is possibly a future common error?

Thanks!

-r

PS:  works fine for me:
onlyv6.com.         3600        IN      SOA     ns1.onlyv6.com. admin.onlyv6.com. 2010031102 7200 3600 1728000 172800