Re: Performance question

On Thu, 17 Jun 1999, Jens-S. Voeckler wrote:

> On Wed, 16 Jun 1999, Scott Hess wrote:
> ]On Tue, 15 Jun 1999, Simon Rainey <srainey@rmplc.net> wrote:
> ]>[...]
> ]>I think everyone realises that Squid's achilles heal is the fact that it
> ]>relies on the undelying OS to manage disk files. While the tests I've
> ]>done are admittedly not very scientific, they do suggest that effort
> ]>spent on the rumoured Squid-FS (or similar) would be worthwhile.

I keep hearing about the problems with using the unix file system for
caching. I understand that using the unix file system semantics to try and
keep a database(cache) is not the best solution in the world and will not
scale well at all. I have not yet seen any description of where the time
is being spent.

I assume that the conversion from URL-filename is trivial in terms of
time. Where is most of the time being spent? In the file open or in the
reading of the data or is the problem caused by forcing the file opens to
happen synchronously?

Alvin Starr || voice: (416)585-9971
Interlink Connectivity || fax: (416)585-9974
alvin@iplink.net ||
Received on Thu Jun 17 1999 - 06:40:46 MDT