Re: HIT rate first level Web cache compared to Top-level Web cache..

On Sep 2, 9:11, Henny Bekker wrote:
> Can someone explain to me why first level Web caches does have a
> higher hit-rate then top-level web caching servers ??
...
>
> When I look at the statistics of a Desire/Web-caching partner, it seems
> to be the way around... The HIT-rate of the "Second level cache" (the
> Top-level Cache) is much higher (abount 50% in March) then the HIT-rate
> of a First level Web caching-server (on an average32% during the same
> period)??
>
> can this be true or am I mistaken ??

A couple of ideas:

 - first, note that 50% is very close to the optimal cache hit rate.

 - you mention "no cache size limit" for the second level cache. With
   both size and use of the web exploding, our first-level cache has a
   turn-over rate of approximately 1.5 days; anything not queried for
   longer than that will be recovered from the no-limits second level.

 - first-level caches should filter out non-cacheable requests and do
   those "DIRECT". That includes all script requests, secure HTTP,
   POSTs, etc. Thus, these "spoil" the statistics of a first-level
   cache, but not those of the second-level one.

 - Although you can do much of that for first-level caches, too, via
   auto-configs, first-level caches are often used in a firewall
   that effectively prohibits that kind of optimization.

 - Consider, too, that first-level caches may choose to use second-level
   caches 'proxy-only', at least for some domains, to more efficiently
   use the joint disk capacity.

Best, Andreas

-- 
Andreas Strotmann       / ~~~~~~ \________________A.Strotmann@Uni-Koeln.DE
Universitaet zu Koeln  /| University of Cologne   \
Regionales Rechenzentrum| Regional Computer Center \
Robert-Koch-Str. 10    /|    Tel: +49-221-478-5524 |\   Home: -221-4200663
D-50931  Koeln        __|__  FAX: +49-221-478-5590 |__________~~~~~~~~~~~~