Re: [squid-users] optimal not too agressive refresh patterns

From: Joe Cooper <joe@dont-contact.us>
Date: Fri, 23 Nov 2001 18:32:15 -0600

Don't forget to note price/performance of the two Squid-based entries
compared to Cisco. (At the third cacheoff the price/performance of our
Squid entry went up by a pretty good margin--one must guess how Cisco
has developed in the same time period since they didn't show up. But
it's worth taking a look at those numbers as well, keeping in mind that
the numbers cannot be directly compared, because the 3rd cacheoff
workload was considerably tougher than the 2nd cacheoff workload.)

It is worth pointing out (as I think Vivek is doing subtly) that Squid
cannot travel in the >40Mbps networks where Cisco and several of the
other proprietary vendors can without clustering. Squid is coming close
to removing at least two of the problems that prevent moving up to that
level of scalability, but it's not there quite yet.

Price performance of Squid is still quite good, however, especially in
smaller networks (1.5-7Mbps).

Vivek Sadananda Pai wrote:

> khiz code wrote:
>
>> BTW has anyone done a comparison between a good h/w based squid box say like
>>having 512 MB ram and 4 scsc1 hard disks and Cisco's Cache engine ..
>>Cisco has yet to appear in a bakeoff i believe
>>
>
> See http://polygraph.ircache.net/Results/cacheoff-2/
>
> The smaller Cisco box had 7 disks and the larger Squid box had 6.
> Throughputs are 951 and 160, respectively.
>
> -Vivek

-- 
Joe Cooper <joe@swelltech.com>
http://www.swelltech.com
Web Caching Appliances and Support
Received on Fri Nov 23 2001 - 17:29:27 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:04:29 MST