[dns-operations] Signaling client protocol to authority

[Paul Vixie]

> From: Mark Andrews <marka at isc.org>
> Date: Mon, 17 Jan 2011 14:57:47 +1100
> ...
> Well ask Apple/MS/RedHat/Google/Mozilla etc. to send out updates
> for the broken applications even if they are deprecated.  Grandma
> won't have turned off the update service.  It also much easier than
> getting every ISP to update their nameservers to support a DNS
> extension of questionable value.
> 
> This isn't about the OS doing things right.  The OS isn't broken.
> It's the applications running on the OS and only a handful of them
> in most cases.

no matter what layer the problem is at, the solution is either for the
purpose of improving in-quarter or in-year results vs. multi-year results.

to me, having no investment in v4 where i'll see hard costs or lost
revenue from people who get the v6 transition wrong, it's obvious that
we should waste no time at all optimizing the mixed v4/v6 case so that
people who are not ready for v6 won't accidentally be exposed to it.

to others, whose in-quarter revenue depends on v4 working well until
the moment it's no longer necessary, many kludges will seem necessary,
including those which change the protocol for everybody, or which ask
lots of server-side vendors and operators to behave differently, or
which make that kludge-like behaviour on the server side will be a
competitive differentiator.  gack.

let's go for the gold.  we know where we end up.  let's get there faster
and spend less time agonizing about the transition.