[Prev][Next][Index][Thread]

Re: Looking for pointers



(Apologies if multiple copies of this slipped out - Neil)

>We are about to put into effect a very large proxy/cashe server. This one will
>not be just for one campus, but for all campuses in the country (no joke and 
>this is not being sent on the 1st).

We have been operating a scheme such as this at the HENSA Unix Archive in
the UK for more than a year. In fact we allow connections from anywhere in
the world, but obviously it is most sensible for people close to us to use
our services. A poster at the recent WWW95 conference discussed a similar
scheme in Germany.

>I am looking for anyone that may have some expertice with handling a
>very large area. At this time, I do know that I will have 7 to 8Gb of
>disk space for cashe.  I may have to cut it down just a little, a Gig
>or so, for the local non proxy/cashe server.

I would suggest that this is a reasonable amount of disk space. For
most of our lifetime we have been running on a 2Gb partition. In the
near future I hope to move this to 8Gb. Based on our experience, and
some simulations that we have run, we have found that it is a case of
diminishing returns. Even quite small caches can give good performance,
but as the caches get bigger and bigger the performance does not
follow.

>The only thing that we have not picked as of this time, is the machine and op
>for it. We will be running the Cern V3.0 server. 
>
>Looking for any ideas or pit falls.
>
>Al

I would say that the choice of caching server directly affects the machine
that you need to use. The CERN server is a forking server and with any
significant traffic will destroy all but the fastest machine. I have
recently been hearing about the Harvest cache, but have not had time to
evaluate it myself.

Get in touch if you have more questions.

Neil G. Smith,
HENSA Unix,
The University of Kent at Canterbury.

Follow-Ups: References: