[SQU] Hit rate vs. bandwidth saving

Hi,

we are running squid-2.3.STABLE4 on a Linux box with 512 MB RAM and about
36 GB disk space which are nearly full (about 95%, although I set cache_dir
to only 28000, but that's another problem ...).
   
We have about 250000 requests per day yielding a client data transfer of
about 1.5 GB per day. What puzzles me is that the average hit rate
(including all TCP.*HIT) is greater than 50% while the average bandwidth
saving is only at 10-20%.
   
A typical scenario for one day:

                    requests bytes
  -------------------------------------
  TCP_HIT 25959 151.24M
  TCP_REFRESH_HIT 47279 59.89M
  TCP_IMS_HIT 56264 18.29M
  TCP_MISS 107679 1241.85M
  TCP_REFRESH_MISS 6407 60.46M

Is this normal or can it be improved?

Thank you,
Erik

-- 
| Erik Heinz, IKS GmbH Jena * erik@iks-jena.de * privat: erik@jena.thur.de  |
+---------------------------------------------------------------------------+
--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html