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

From: Ernst Heiri <heiri@dont-contact.us>
Date: Tue, 02 Sep 1997 10:36:16 +0200

On Tue, 02 Sep 1997 08:10:04 +0100 Henny Bekker wrote:

> Dear cachers,
>
> Can someone explain to me why first level Web caches does have a
> higher hit-rate then top-level web caching servers ??
> A first level cache get's all the nice cacheble objects which might be
> accessed again a lot in case of a homogeneous'user group.. All the
> unresolvable requests are routed through the Top-level cache.. However,
> if the top-level Web cache does not have any shortage of resources, the
> HIT rate of the top-level cache might be close to that of the first level
> cache..
>
> 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 ??

For first level caching servers for medium size universities we get
hitrates of about 30-35%.
For our top level caching servers we get hitrates of about 15%.
The top level caching servers are not used by "direct clients".

An object cached once by a first level cache will not be requested from
the parent anymore - at least for a certain time. During this time
the first level cache will maybe have several local hits for this object.

The top level cache at the other hand will have a hit only if an other
"child-proxy" is requesting this object too.

The hitrate of a top level cache havely depends on how many clients
("child proxies", "direct clients" or siblings it has.

Ernst
Received on Tue Sep 02 1997 - 01:40:57 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:36:54 MST