[squid-users] mgr:info question

[Amos Jeffries]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 13/11/2014 4:11 p.m., Alberto Klocker wrote:
> Looking at the squidclient mgr:info command output I was wondering
> what the difference between these two entries are?
> 
> Cache information for squid: Hits as % of all requests:      5min:
> 0.7%, 60min: 0.3%z Hits as % of bytes sent:        5min: 51.3%,
> 60min: 25.5%
> 
> 
> I can guess the first one means all requests but I'm stumped on the
> wording of the second entry.

HTTP messages consist of a variable amount of bytes each. A HIT on a
1KB object saves less bandwidth than a HIT on a 1MB object. The only
(relatively) predictable thing based on message count is consumed CPU
and network sockets.


"% of all requests" is counting by messages.
 - higher values means less CPU consumed/wasted upstream (server
latency), less contention for TCP sockets (HTTP pipeline congestion).

"% of bytes sent" is counting by bandwidth used.
 - higher values means less bandwidth consumed/wasted upstream (TCP
congestion).

So ... in the past 60min 0.3% of your clients were responded to using
the cache, which reduced your upstream bandwidth usage/cost by 25.5%.

Amos

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (MingW32)

iQEcBAEBAgAGBQJUZCouAAoJELJo5wb/XPRjzJIIAM499MGEXNvMGI60Vul/jShC
2eP5W7O/6rDygQQBkSpI/h9ph0xcgRc4eG+F/jbvUIY828fYOkIZirAOknr8NxPD
hV3xB3nZw859SyOfB6+Iz0X2Fp7VWIZZiRg27XxGuWcWsLGF7qmA7Kjh6S0jy21c
mCSZYKU5Rgx6FMNNBac7NxztogD0ZD2z2JQwJsSK1JI4JFkw2eImu+NTPPU6aQUG
pdiRBGTzATTL0lDXOBaM96s7o3XwkKift4OzZgetBoG+b3LQ2XX4Iw9caBnK/Fr8
MS2nLsvISUI+XrvpqFTYFdTlRVOEYjzfOGPzOwxgNdl86TlzTMh1ExOTmWd92RE=
=pN5q
-----END PGP SIGNATURE-----