[squid-users] Moving to squid from inktomi

From: Marshall <marshall@dont-contact.us>
Date: Thu, 27 Jun 2002 14:28:05 -0400 (EDT)

Hello,

I am currently using Inktomi Traffic Server for reverse proxy. I need to
add more cache servers, but the Intomi software license is a whopping
$4500 for a single processor. Beside the fact, the "appliance" it is
running on is going to hell since they have 4 disks in RAID 5 using the
Linux kernel. I am not a Linux user and don't want to babysit software
raid anyways.-
Back to the point, I am looking for a cheaper solution while trying to add
redundancy. My overall goal is to hit 600 req/sec. using a minimum of two
machines. My hardware budget is $7000. (The Inktomi appliance is
supposedly capable of 600 req/sec in reverse proxy mode.)-
So far, I am on this track for each server's hardware: Single 1.13Ghz
PIII, 1GB RAM, 2x18GB 10K Ultra 160 HDs (1 for OS/logs, 1 for cache). I
can get 2 1650s with this config from Dell for ~$4,300 which is well under
budget which is good. I am leaning towards FreeBSD. Based on old IRC
cache-off benckmarks (using hardcore guestimating), it 'seems' that each
server could handle over 300 req/sec if configured properly.-
These will be caching primarily images (70% of which are over 40K).
Considering my low requirements for cache storage space and that I will be
using squid in reverse mode does 600 req/sec seem feasible with the above
scenario? If not, can hardware upgrades get me there (15K drives, dual
proc, more RAM, etc)?
Any input would be appreciated.

-Marshall
Received on Thu Jun 27 2002 - 12:28:17 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:08:51 MST