From: Niall O Broin (niall at domain linux.ie)
Date: Mon 13 May 2002 - 12:27:37 IST
On Mon, May 13, 2002 at 10:42:44AM +0100, Breathnach, Proinnsias (Dublin) wrote:
> You could, of course, count the cache hits, and multiply their size, by
> their frequency etc ... but that doesn't really give an answer, as there's a
> human factor ... bandwidth usage may well have _increased_ because you're
> using squid (or any other cache), people will use the internet more often if
> it's responsive, when it's not, they go back to doing things the old way !
I think what I want must be eminently measurable. The total bytes served
from squid is the sum of the sizes all the hits + all the misses. If there'd
been no squid, every cache hit would have resulted in a fetch from the
internet, so the bandwidth gained is =
cache item size X number of hits for that item
> The only real way is to turn off the cache for a week, keep stats, turn it
> on and keep more stats ... and even that won't answer you !
No, that of course won't answer me because I'd not be comparing like with like.
This archive was generated by hypermail 2.1.6 : Thu 06 Feb 2003 - 13:16:47 GMT