A co-worker and I are in the early stages of planning and implementing a Squid cache server. In the past, with some clients, I generally took what hardware (and drive space) was available and made use of it without much regard to performance tuning etc. (ie: customer said "here's what I have... make it a Squid cache server" <grin>) I'd rather not do that in this case,and so here I am asking for some thoughts based on our system needs.
Our network is connected to the Internet with a single T1. We support about 550 students and 250 faculty/staff. The students have network/Internet connectivity from their dorms, and many of the faculty do as well.
My preference is to set up Squid on a Linux/Intel based server since I am very familiar with this type of a configuration. My question is, with the above network user base to support, what would be a good starting point for an Intel/Linux based Squid proxy caching server?
Processor Speed?
Memory Size?
Should we bother considering mulit-link trunking (bonding) two 100Mbps ethernet connections to the switch, or is that just plain silly?
Also, regarding the cache volume:
What is a reasonable size?
RAID 1, or RAID 5? Seems like RAID 1 is the way to go for speed. Or maybe a RAID 1+0 for speed and redundancy.
Can I get away with software-based RAID, or should we not play games and go for a hardware raid solution?
Did I miss anything?
Thanks in advance.
--- Bill Arlofski Unix/Novell Systems Administrator The Hotchkiss School waa@hotchkiss.org 860-435-3140Received on Mon Nov 05 2001 - 11:34:36 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:03:54 MST