Re: Process size..

From: Scott Hess <scott@dont-contact.us>
Date: Wed, 10 Nov 1999 08:45:51 -0800

Better to let the OS handle it. Two problems. One is that the OS might be
better tuned towards keeping disk blocks in memory than Squid is towards
keeping pages in memory. [Might not, of course, but the OS is _about_ this
type of thing, whereas Squid is not.]

The second problem is the killer. You never, ever, ever, ever want Squid
to swap. Since Squid is a select()-based system, it means that if any
request requires a page to be swapped in, all requests will wait. With a
cache_mem of 8M, you're likely to repeatedly touch all of that 8M much more
frequently than with 500M. The longer a page goes between touches, the
more likely it is to be swapped out.

Later,
scott

----- Original Message -----
From: <Tony_Melia@Dell.com>
To: <squid-users@ircache.net>
Sent: Wednesday, November 10, 1999 8:01 AM
Subject: RE: Process size..

> Is there a disadvantage to having thsi value higher? The normal reaction
> is, if you have the extra RAM, use it, or is it better to leave it for
the
> OS to use as disk cache?
>
> Regards,
>
> Tony Melia MCSE
> Dell Server Support Group
> Bray, Ireland.
>
>
> * +353 1 4772072
> * Tony_Melia@Dell.com
>
>
>
> -----Original Message-----
> From: Dave J Woolley [mailto:DJW@bts.co.uk]
> Sent: Wednesday, November 10, 1999 12:30 PM
> To: 'squid-users@ircache.net'
> Subject: RE: Process size..
>
>
> > From: Marc-Adrian Napoli [SMTP:marcadrian@cia.com.au]
> >
> > cache_mem 500 MB
> >
> The normal reccomendation is to leave this at the
> default 8MB; what's your reasoning in using such
> a large value?
>
Received on Wed Nov 10 1999 - 09:56:33 MST

This archive was generated by hypermail pre-2.1.9 : Wed Apr 09 2008 - 11:57:32 MDT