Re: [wg-c] Initial Numbers

[Kent Crispin <kent@songbird.com>]

On Wed, Dec 15, 1999 at 10:58:54AM -0500, Kevin J. Connolly wrote:
> Well, Christian Huitema presented a paper at ISOC-NY last Spring
> on this very subject.  The more TLDs there are, the greater the
> likelihood that one will need to query the root in order to resolve 
> a domain name.

This is undeniably true.

> This is (according to Dr. Huitema, who, I believe,
> knows a few things more than the average bear about the DNS) as
> fundamental as 1+1 = 2.  So, what's the flaw?

No flaw, but the problem is much more complex than 1+1.  I didn't
see the talk, but obviously, the number of hits on the root
depends ultimately on user behavior -- a TLD that is *never*
referenced by anyone will never cause any hits on the root server. 
So in order to make reasonable deductions you have to make some
assumptions about user behavior.  

I imagine that the assumption made was that the probability of a
SLD being referenced by a user has a uniform distribution across
all TLDs -- a reasonable guess.  From that assumption you could 
then make assumptions about average local resolver cache size 
dahdidahdidah mumble mumble wave hands, and then come up with a 
probability of a trip to the root server being necessary.

The key insight, though, may be that there is a high probability
that lightly populated TLDs will require a trip to the root
because they aren't in the cache, while heavily populated TLDs
like .com will be much more likely to be cached, and that these
two factors very roughly balance, so therefore there is a rough --
very rough -- linear relationship between the number of TLD zones
and the load on the root, regardless of the number of subdomains 
in the TLDs.  A conclusion of substantial significance...

I would love to see the paper -- do you have a reference?


-- 
Kent Crispin                               "Do good, and you'll be
kent@songbird.com                           lonesome." -- Mark Twain