[AusNOG] "stateless TCP" for DNS

[grenville armitage]

All,

I appreciate the feedback regarding the Intro text to our statelessTCP tech
report. In retrospect we probably should've said very little in the Intro,
as it was not our goal to write an authoritative discussion of pros and
cons of using DNS over UDP. The tech report was really aimed at helping
interested parties understand and experiment with our statelessTCP patches
to FreeBSD 9. (Our assumption being that whoever would want to try our
patches would already have their own reasons for thinking DNS over TCP
could be a future direction to go.) That being said, I accept the editorial
blame for allowing our Intro to make overly broad claims regarding the
possible reasons for a move to DNS over TCP, and our failure to acknowledge
the role of EDNS0.

As previously noted, it is a potentially-crazy idea to solve what might be a
non-problem. Nevertheless, we'd still be interested in hearing from anyone who
tries to use our patches and replicate our performance tests -- yes, we're
suckers for punishment :)

cheers,
gja