On Tue, 18 Jul 2000, you wrote:
> hmmmm...we run 3 proxy servers(microsoft, sorry na lang), with over 1k
> computers here in the Ateneo. our average hit ratio is 30% roughly. if
> you're to think of it, hit ratio really is not a good metric for proxy
> cache. it basically says that your users go to the same sites often. that's
> human behavior dependent. i believe the better metric is average time to
> answer a request...which we clearly see in a graph in performance manager. i
> dunno if squid has this.
The value you will get in the "average time to answer a request" will give you
results directly proportional to the speed of your leased line, as this will be
the biggest delay, not the proxy response time.
You're right the Hit ratio is supposed to measure the amount of 'trips the
proxy went to the internet' to fetch a page. You should also look at stats on
the amount of bytes saved due to the fact that the page was storedin the local
cache You will see that although squid has a hit ratio of 40%, the actual
amount of bytes saved by this is a low 18-20%. This to me is the real value of
a proxy, not the actual hit ratio.
-
Philippine Linux Users Group. Web site and archives at http://plug.linux.org.ph
To leave: send "unsubscribe" in the body to [EMAIL PROTECTED]
