At 21:15 +0000 12/10/06, Paul Vixie wrote:
>we know there won't be a new namespace. ever. but in addition to adding
>new kinds of names (idn) and securing the data (dnssec) we have sometimes
>tried to improve the protocol (edns). if i were to embark on dnsv2 it would
>be with the hope of completely forklift-upgrading the protocol while keeping
>the namespace as it is.

I agree with that. Where I see the discussion getting wrapped around
an axle is when we cross the divide between conveying the data in the
name space with regulating the data in the name space. I've trotted
this out before, there is a difference between talking about DNS and
talking about provisioning of domain names.

It may be hard to see, but every DNS zone is managed by some sort of
registry. Whether it is a formal one like the g/sTLDs of ICANN or a
ccTLD, or just a personal zone with three names in it, there is a
registry. The heart of the registry may be a distributed database or
a text file, the process of generating the zone file may be a
database difference report or the text file may already look like a
zone file.

Coming up with a new wire protocol for DNS would represent about as
much of a change as bring up servers on IPv6 to a registry. DNS is
just a publication mechanism, not what is regulated. Perhaps a new
protocol is more like the transition from LP's to CD's - the music
stayed the same but the cases got smaller. 'Course the analogy
breaks down when you get to MP3's and on-line stores, etc. And
DC-101 (local radio) had to change from "7 album sides at 7" to "7
CD-sides at 7" declaring each half of a CD to be a side.

A cleaned up DNS would present an issue to the regulators - a better
implementation of the CLASS concept. But that's their problem, not
Edward Lewis +1-571-434-5468

Dessert - aka Service Pack 1 for lunch.

to unsubscribe send a message to with
the word 'unsubscribe' in a single line as the message text body.