djbdns: An alternative to BIND

Dean Anderson dean at av8.com
Mon Apr 11 17:27:26 UTC 2005


On Mon, 11 Apr 2005, william(at)elan.net wrote:

> Surely, you aren't saying that is somethig wrong with that or that they
> are making non-compliant product just because they choose to use different
> "proprietary" protocol when two of their products interact with each other
> (while still supporting standard protocols for other systems)? Otherwise 
> if we do use your rationale tha product is bad when it does it, then all 
> my cisco equipment would be considered bad!

The way that the went about it is certainly irresponsible:

1) They first thought they had to change AXFR to accomodate IXFR (they
didn't really, but due to BIND architecture, it may have been easier)

2) Then they discovered a vagueness in the RFC describing AXFR.

3) Then, instead of clarifying the RFC first, they altered their protocol.
Knowing that other implementations would not interoperate, they included
code to detect BIND9, and code to turn off non-standard protocol.

4) Then they deployed BIND9 on an unsuspecting user community. 

5) Then, after criticism, finally decided to try to clarify the draft,
assuming that their employee who was a Working group co-chair would breeze
through the change. As justification for the change, they asserted it
would be easier for the 6 or 8 other DNS implementations to change their
installed base than to change BIND9. (holy cost-shifting, batman)

They should have realized that a non-interoperable change was going to be
tough to push though: That the likely outcome where 6 or 8 other
implementations (including their own) had read the draft a particular way
originally, would probably result in a clarification that the original
particular way was what was really meant.  They used bad judgement, to put 
it mildly.

Cisco has done some non-standard things, and it has gotten some
well-deserved criticism for them, as well.  But Cisco hasn't done anything
that is quite this blatantly irresponsible. In most cases that I can think
of (cisco hdlc, skinnystation--which technically wasn't cisco), standards
and implementations were far and few between and they formed up as best
they could to actually have a working product in an immature market.  
That isn't the same as what BIND9 did. DNS is a mature and pretty
well-defined protocol.

> > Its not a real big problem, though the BIND9 detection might be dicey. 
> > An implmentation that pretends to be BIND (but not using the proprietary
> > protocol) might have a problem. But so far as I know, there are no such
> > implemenations at present, so its not a big problem, at least, not right
> > now, anyway. It could be a problem later, if someone introduces a server
> > that pretends to be BIND9, but isn't.
> 
> Nobody should be producing product that "pretends" to be something else,
> that itself would be a problem and may even be illegal if BIND name is
> trademarked (and even if its not if somebody makes different product
> that is using bind name and that product does not work or works differently,
> it creates dillusion and bad reputation for makers of bind and so its
> something ISC could legally demand to be stopped).

BIND is an acronym of Berkeley Internet Name Daemon.  I've heard that
Vixie claims a trademark on this, but it seems rather like the linux
trademark issue of a few years ago. I didn't hear that they purchased the
copyright from the University of California.  So, I don't think it is his
to trademark, and it was a common term in use well before ISC existed.  
ISC didn't write BIND, but has only maintained and modified it over the
years.  They own modifications, at most.

But even if they did purchase the copyright from Berkeley, we are talking
about what amounts to packet signatures. Fair use allows one to create
interoperable products. [DMCA 1201(f), I think].

For example, many web browsers "pretend" to be other browsers. There are
also examples of packet signature changing, so that your linux box appears
to be a netbsd box, for instance.  One can always interoperate if one can
figure out how. Trademark issues are only relevant to marketplace
branding, not to protocol details. (I'm LPF president for something like
13 or 14 years, I know a little bit about copyrights, trademarks and
patents.)

		--Dean

-- 
Av8 Internet   Prepared to pay a premium for better service?
www.av8.net         faster, more reliable, better service
617 344 9000   





More information about the NANOG mailing list