Re: [squid-users] 50 requests per second

From: Robin Stevens <robin.stevens@dont-contact.us>
Date: Fri, 28 Jun 2002 18:46:02 +0100

On Fri, Jun 28, 2002 at 01:34:39AM -0500, Joe Cooper wrote:
> I benchmarked this box at 240 reqs/sec for four hours on a full cache,
> and didn't have time to really push it beyond its limits. I wouldn't
> expect it to sustain that in an ISP environment however, and would
> recommend it for a significantly less demanding life (I'm comfortable
> calling it a 180reqs/sec machine, with ability to handle peaks over
> 220--it lives in an ISP where it peaks at 160 reqs/sec, and usually
> sustains around 120).
 
This is not dissimilar to typical loading on our machines - peaks last term
totalled around 2 million requests an hour across four machines. What
worries me is that this is 80% higher than the previous year's peak...

As regards how I measure throughput, I use the method of stick an entire
university behind the cluster and see how the machines cope with the
traffic generated :-)

> Another reason is that latency reduction is the prime motivation for
> caching in most of my clients environments. ISPs in the US are not
> strapped for bandwidth in most cases, but they do want a 'leg up' on the
> competition in their area, by providing a 'snappier' browsing experience
> (and if it saves them a few hundred bucks a month on bandwidth, that's a
> nice bonus).
 
Reducing bandwidth charges were our original incentive for implementing a
caching system, but this no longer applies. Management are very reluctant
to spend any more on cache hardware, and arguments for retaining the system
will come down chiefly to latency improvements, and additionally the
security benefits (we got to be experts on catching Code Red & Nimda
infected hosts).

> It's all about balance. And RAM being dirt cheap these days.

Unless you're buying it from Dell :-) Certainly when we purchased the
first machines, 512MB ECC DIMMs were much more expensive than two 256MB
modules, hence we went for 6x256. (Concerns over warranty are the main
reason we don't generally go for cheaper third-party RAM on the servers.)

-- 
--------------- Robin Stevens  <robin.stevens@oucs.ox.ac.uk> -----------------
Oxford University Computing Services ----------- Web: http://www.cynic.org.uk/
------- (+44)(0)1865: 273212 (work) 273275 (fax)  Mobile: 07776 235326 -------
Received on Fri Jun 28 2002 - 11:46:04 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:08:53 MST