Re: [v4v6interim] Single namespace

On Wed, Oct 01, 2008 at 11:34:30AM -0700, james woodyatt wrote:

> The "gross indecency" here is, apparently, admitting in public that we 
> cannot escape the need to amend the DNS standards to recognize, for the 
> first time, that we need more than one public horizon, i.e. the V4V6COEX 
> public horizon and the IPv6 public horizon.

While I understand that suggestion, and even have a certain amount of
sympathy with it, I note that there's a practical reason that many
DNS-heads don't like multiple versions of the namespace: they tend to
leak into other contexts.  Since we have experience with this
already, it would be prudent to plan that such leaks will be part of
any future deployment of "split-brain" approaches.

This is bad enough in the context of corporate NATs with RFC 1918
addresses inside, but at least one has a hope of spotting this (by
virtue of the 1918 address itself).  I appreciate what Fred said
earlier, that applications don't always care about the distinction
between 1918 and non-1918 address spaces; but the distinction is an
old one, and applications (or just help desk workers) that need to
deal with these cases can use that distinction to help them.  Since by
definition, in this case, what we're trying to address are
applications that can't learn new things (or they'd use IPv6
natively), I'm pretty worried that we'll create new modes of failure
that will be very hard to troubleshoot, because we'll have inserted a
new device in the network whose whole purpose is to intercept DNS
packets and change them into something else.

To be clear, I'm not saying, "Disaster! Bad! Evil! No! DNS purity!"
I'm just raising a caution that there seems to be a boom on for
split-brain DNS answers, and such answers come not only with DNSSEC
problems, but other issues as well.

A

-- 
Andrew Sullivan
ajs@commandprompt.com
+1 503 667 4564 x104
http://www.commandprompt.com/
_______________________________________________
v4v6interim mailing list
v4v6interim@ietf.org
https://www.ietf.org/mailman/listinfo/v4v6interim