[squid-users] analyze squeezer.pl stats

Hi,

I have the following from one badly performing proxy from squeezer :

                        No of req Xfer (kB) Xfer
speed(kB/s) Xfer % Times to direct
Total traffic 48726 235280 1.317 100.00
0.78
Direct fetches 189 607 1.679 0.26
1.00
This server 11271 13558 3.404 5.76
2.03
Other servers 47944 24683 0.475 10.49
0.28
Cache hierarchy 59215 38241 0.683 16.25
0.41
Is it correct to translate those stats like this :
In general all traffic is 0.78 times faster than direct, so using no proxy
and going direct would be faster.
This server is 2 times faster than going direct.
Cache hierarchy is 0.41 times faster than going direct. Which means in this
case it is twice as fast to go direct than
getting the page from cache.

Am I right? If so, I do not understand what's going on.
Users indeed claim slow performance, however their hit rate is about 42%.

It is configured like this :
client PC -> this local proxy -> parent + trend interscan -> internet

The system is a Celeron 533 with 256Mb RAm and 20Gb IDE Disk
OS is RedHat Linux v6.2, squid 2.5DEV-pre4 is being configured with 64mem
and
cache_peer 138.249.118.136 parent 8080 8081 no-digest no-netdb-exchange
round-robin
cache_peer 138.249.118.131 parent 8080 8081 no-digest no-netdb-exchange
round-robin
cache_peer 138.249.17.25 parent 8080 8081 no-digest no-netdb-exchange
round-robin

I do also have a huge amount of page faults (see my previous mail of today
in the list).

Any comments would be very much appreciated.

regards,
./koen
Received on Thu Apr 04 2002 - 08:30:07 MST