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--

Reply via email to