RE: [squid-users] squid sucks? ;-(

I (and my users) have noticed a significant slowdown since using Squid.

My ADSL provider said we are using only 40% of our bandwidth. The Internet
seems much faster with out squid. But I have never tested with real numbers,
only that it "feels slower".

If you could give more details on how to test fps, I would be willing to try
it and send you our numbers.

Thanks
--AB

-----Original Message-----
From: RIP [mailto:rip@box43.pl]
Sent: August 16, 2004 3:08 PM
To: squid-users@squid-cache.org
Subject: [squid-users] squid sucks? ;-(

Hi!

Hard to bolive myself I am talking that after houngreds of hours spend on
watching/configuring/tunning/etc squid, but here it goes..

4x DSL
1090 Kbps download max per DSL, one squid box,

Configuration A
transparent redirect to squid,
cachemgr stats:
Cache Misses: ~0.10000
DNS Lookups: ~0.10000
CPU Usage: ~10-20%
Cache hits: ~None (cache_dir null /tmp <-- for testing)

HTB limit 950Kbit per DSL
WRR attached per DSL
acticve users: ~50 per DSL

Configuration B
some as A but no redirection to squid

1/4 clients to each DSL
HTB limit 950Kbit per DSL
WRR (weight round robin) attached per DSL
users get always all avaiable bandwidth (wrr)

testing procedure:

www.numion.com/yourspeed (20second test)
one computer attached to dsl1, second one to dsl2

squid off, users go directly (masquarade), htb wrr attached
hour computer1 computer2
19:49 528kbps 261kbps
19:52 594kbps 518kbps
19:55 431kbps 575kbps
19:59 582kbps 540kbps

squid on, transparent redirection, htb wrr attached
20:02 181 223
20:06 187 166
20:09 175 126

squid off, users go directly (masquarade), htb wrr attached
20:12 405 322
20:16 423 286
20:19 367 324

squid on, transparent redirection, htb wrr attached
20:23 92 126
20:25 144 164
20:27 200 186
20:28 157 160

squid off, users go directly (masquarade), htb wrr attached
20:29 679 378
20:32 542 249
20:35 542 435
20:39 545 216
20:42 309 537
20:45 471 350

average fps squid on, computer1: 164kbps
average fps squid off, computer1: 493kbps

some one can explan this?
some one can do test to confirm/negate that?

regards.

--RIP--
Received on Tue Aug 17 2004 - 14:58:11 MDT