Just FYI:
This problem was caused by the caching itself.
When I disable caching for all intranet hosts, the speed is like it should be......(+/- local lan speed)
So it seems the my Squid disk performance is not like it should be.
rgrds,
Bart
Schelstraete Bart wrote:
Hello,
I 'm facing a problem with my Squid proxy server, and I hope maybe of you can help me. (the problem is maybe not squid related)
I installed Squid 2.5 on a Linux server which is currently live (RH 8.0). This machine is a 2 CPU machine, with 1 GIG RAM, 2 x 100 NIC. (in fact this are 1G bit cards, but it's currently connected on a 100Mb switch).
And I also installed Squid 2.5 on a HP-UX machine , which I'm testing now. (HP-UX 11i). This machine has 3 CPU's, and 1x100 NIC. (also 1Gbit card)
(completely the same Squid configuration)
Everything works fine, and Squid itselfs works quite fast. Because we have a lot of servers in our lan, which doesn't need to use the 'cache_peer', I configured Squid that he goes directly to those machine, so without the cache_peer.
Now yesterday I did some tests, and I downloaded a file via ftp through the proxy. (proxy will bypass cache_peer for lan, and I configured Squid).
If I used the Linux Squid proxy , I got a download speed of +/- 300 Kb/s. But If I'm using the HP-UX Squid proxy I get a download speed of 1000 / 1200 Kb/s..
I double check the configuration, and I checked the access.log file, and Squid not using the cache_peer for that download.
So I don't really understand why the Linux machine is that slow comparing with the HP-UX.
Does somebody has a clue what I can do to improve this kind of connections?
Bart
