On Monday 03 November 2003 09:10 pm, Ricardo Jose Guevara Ochoa wrote:
> The cache of my university that I work for gets slow if i have more
> than 200 active connections on it
> - 512 Mb
> - Cache_mem 96mb
Drop the cache_mem setting; the default of 8 MB is usually fine. Your OS will
use available RAM to cache files anyway, and reserving less for Squid leaves
more available for your system when needed.
> - Two 32 Gb SCSI Hard Disk Drives
> - Using ufs and one hard disk only for cache( 10gb of it )
Unless you need a full 10 GB of cache, dropping the total down to 5 GB or even
2 GB will reduce Squid's memory usage (see the Memory section of the Squid
FAQ for details). Regardless of total cache size, use aufs instead of ufs
(aufs has better performance). Split your cache between the two disks - you
will maximize your total disk throughput.
How may requests per second do you need/want the system to handle?
Adam
Received on Mon Nov 03 2003 - 21:27:59 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:21:04 MST