BIND - how predominant?

[David R. Conrad]

Michael,

> However, what difference does it make to the original poster, since
> non-BIND and BIND DNS servers MUST be compatible or name resolution
> fails.

The original poster asked: 

>>> Can someone point me to an AUTHORITATIVE resource which states the
>>> percentage of DNS servers on the net (or in the public & private world)
>>> which run a flavor of BIND on a flavor of Unix?

which implies nothing about compatibility or interoperability.  It would
seem that Bill's answer would be pretty much in line with the request
modulo there is no distinction on the flavors of Unix (given BIND runs
on every version of Unix shipped that I'm aware of, I suppose this
number could be derived by figuring out how many units each vendor
ships).  Of course, calling Bill authoritative for something always
makes me nervous... :-)

> We have issues doing zone transfers with our non BIND server to or from
> non BIND servers.

ISC has in the past attempted to address this situation by trying to put
together an effort to develop a neutral protocol conformance test suite
for the DNS protocols, but to date, there have been only expressions of
interest (e.g., "that'd be nice..."), no one (including Cisco,
Microsoft, Lucent, Nortel, Checkpoint, and others) have committed to
funding such an effort.  I suspect that until such a suite is developed,
people will use BIND as the de facto conformance test (i.e., "it works
with BIND, ship it!") and you'll continue to see difficulties with other
non-BIND implementations.  I do not consider this a good thing. 
Fortunately, BIND version 9 will be much more RFC compliant so the de
facto approach may have more validity in the future.

Regards,
-drc