Re: [dnsext] afasterinternet.com trial and draft-vandergaast-edns-client-subnet-00

[Matthew Dempsky <[e-mail]>]

On Tue, Aug 30, 2011 at 8:12 PM, Andrew Sullivan <[e-mail]> wrote:
> I actually don't care about the IETF process.  What I do care about is
> the potential for interoperability headaches later because of
> undocumented collisions in EDNS0 option code interpretation.  Avoiding
> that sort of headache is the exact reason we have a registry.

I suggested in April 2010 that part of the EDNS0 option code space be
reserved for private use, but it received very little feedback
(positive response from Colm MacCárthaigh, slightly negative from Paul
Vixie, and some meta discussion about how strictly IANA follows its
instructions on allocating numbers):

http://www.ietf.org/mail-archive/web/dnsext/current/msg07768.html
http://tools.ietf.org/html/draft-dempsky-edns0-options-for-private-use-00

Reserving a code point for public (i.e., general, unnegotiated) use
doesn't make sense until the protocol has solidified.  For a private
test (even a large scale one conducted across the Internet between
multiple organizations), it seems appropriate to use a private use
code so long as the participants have agreed a priori what the code
means in transactions between their DNS servers.

I don't see anything wrong with (e.g.) privately negotiating to use
65300 to mean draft-vandergaast-edns-client-subnet-00 when querying
Google's DNS servers but to then mean draft-xxx-something-else when
querying someone else's.