[dns-operations] co-operative cacheing

[Tony Finch]

On Thu, 8 May 2008, Simon Waters wrote:
>
> I suspect the results will always be specific to the site in question to
> be of much use elsewhere.

That would be surprising. I'd expect different query traffic from DNS
servers with MTA clients or web cache clients or web server clients or
workstation clients, but I'd expect similar aggregate behaviour from
similar classes of clients at different sites.

> Since some clients will send the same query to multiple recursive
> servers much more readily than others.

It's essentially the case that stub resolvers only use one recursive
server. This is certainly true for my MTAs (hence running a resolver
on each MTA machine has useful fate-sharing properties).

But it isn't really stub resolver behaviour that I'm concerned about. I
want to know if you can reduce recursive lookup latency by querying other
reolvers in a cluster, which depends on the query data not the resolver
implementation.

Your point about most latency coming from brokenness is a good one.

Tony.
-- 
f.anthony.n.finch  <dot at dotat.at>  http://dotat.at/
FISHER GERMAN BIGHT: VARIABLE 3, BUT EASTERLY 4 OR 5 IN SOUTH GERMAN BIGHT.
SLIGHT. FOG PATCHES. MODERATE OR GOOD, OCCASIONALLY VERY POOR.