[squid-users] squid cache Large rock with aufs optimization for bandwidth saving

[Amos Jeffries]

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

On 16/11/2014 8:07 p.m., Ahmed Allzaeem wrote:
> Hi developers .
> 

The developers are not here. This is the user mailing list.

...
> But still wondering , why after im using squid3.head that support
> large rock and using aufs with 4 processes with object size about 4
> M configured in each prcess .the mean size is about 38 KB ???


1) while your Squid supports large objects in rock stores you are not
actually using that. So your Squid workers are still stuck with only
the memory cache shared, and disk caches separate.

2) you have several different cache_mem values spread across the
workers. Only the last cache_mem matters so you have only 6GB used by
any worker. Not the 25GB you may think some are using.

3) you are doign similar things with minimum_object_size. But again
oly the last configured value matters. So its 0 for all workers.


The mean object size is just an average of all the cacheable traffic
your proxy is processing. Bandwidth savings is more closely related to
*total* cacheable traffic than an average of individual object sizes.

Amos

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

iQEcBAEBAgAGBQJUZ+uqAAoJELJo5wb/XPRjzZ4H/3wMteFcHint7RRowwZvtR5x
lwo0Z/u2DsWAfF+lXmvH/PpDhyzEKDzBJh9tVZR1xB5fQ3bT9ydhQrnkoXClq8mC
lqYwmMJy57JG4YGCZnaw8XXt1D248r+RywlNzPXNukyNDTGFBDWkiTBuT2b+Jlxn
ofwj0mjFMuiyxC7Oe7o/f3uqOD+5lpS9spBX8DD8weQzcfKFb/7IrZnb9Hpv9FBY
LhMTWFu8sOBndyaYedelLcUriUL4kthg59sqwp68oDg2SLBUv//UdX5If9RAWm4G
AAYacYMrsOn+OX8wEjscwinY3TNhH+QjAAb1oOZKSfJEtDbIvLp9I2qprDi34o8=
=bgGV
-----END PGP SIGNATURE-----