I'm getting bit tired of hearing that we're screwing up
Microsoft if we
don't do "blah", "blah", "blah" when in fact its the other way around.
But lets solve this all anyway.
Microsoft wants to implement MARID/SPF/Caller-ID, right?
Microsoft wants to do it in the way that is IETF compliant, right?
What Microsoft wants and what Microsoft can do are two very
different things. Microsoft is larger than the governments of
some fair sized countries.
Building large scale commercial software is a very different
prospect to writing a small piece of code.
Deploying an Internet application is very different from getting
a specification through the IETF.
If we are going to go ask the Microsoft DNS group to do something
it had better be something really big, like implement DNSSEC so
we can secure MARID zones and associated accreditation zones.
It would be much easier to pitch DNSSEC than a new RR, even though
there is more engineering work - but not as much as you would
think since the big issue is not implementation, its regression
etc. At least with DNSSEC there would be a major functionality
improvement to point to.
I think we probably can get Microsoft to provide support for a
new RR type - in the next release of the server platform, which
given that its the off cycle upgrade with XP would put it after
Longhorn. I don't think that RR support that starts to become
available in 2007 and might be widely deployed in say 2009 is
very relevant for our purposes.
The IETF process has many pathologies. The most persistent failure
has been in designing new protocols that cannot be deployed, either
because they are incompatible with deployed architectures or
because they make demands that are unrealistic.
It may be possible to persuade an IETF WG to adopt a spec that
has the strong disagreement of a major stakeholder. OK, yes we
KNOW that it this happens. That does not mean it is a good idea
to do so, or that they will accept the aggreement.
This is a case where we have pragmatism in opposition to an idealized
concept of the architecture. Compared to the hacks that got slipped
by to make DNSSEC etc viable the proposed name hack is nothing.