DNS question from a network guy

[Pete Ehlke]

On Tue, Jan 21, 2003 at 03:26:38PM -0500, Kevin Darcy wrote:
> 
> The decision of which nameserver to query is usually made on the basis of
> which nameserver  has answered recent queries faster. Chances are, many if not
> most of the queries will be sent from nameservers which don't have a good
> "scoresheet" of response times from your nameservers, so this tends to inject
> a large random element into the nameserver-selection algorithm.
> 
That's BIND's behaviour. Other implementations are known to randmly or
pseudo-randomly query all the listed servers, to query all of them in
parallel, and to query all of them in parallel the first time and latch
on to the first one that responds, forsaking all the others.

-Pete