Re: Can I *not* have an on-disk cache?

At worst, put the cache on a ramdisk...

[Not a hardware ramdisk, I mean a software ramdisk. You'd have to do the
squid -z thing to create the directories at startup, before starting squid,
but that should be easy enough to work into the startup script. If you're
lucky, your OS will even be able to directly suck data from the ramdisk
blocks to wherever it belongs, so you don't end up with duplicate in-memory
copies (one in ramdisk, one in fs buffers, one in squid :-).]

Later,
scott

----- Original Message -----
From: Steve Willer <willer@interlog.com>
To: Squid Users Mailing List <squid-users@ircache.net>
Sent: Tuesday, July 13, 1999 5:45 AM
Subject: Can I *not* have an on-disk cache?

>
> I'm guessing from the posts that most of you are using Squid in its
> regular mode. Perhaps I have needs that Squid doesn't really address,
> but...
>
> I'm running Squid in http-accelerator mode (in front of a web site). All
> of the pages are dynamic and therefore not cacheable. The cacheable items
> (buttons and stuff) probably only total 20MB or so.
>
> The Squid box is hitting its capacity limit, and I'm looking for ways to
> squeeze some more life out of it so we survive until we replace the box.
> Turning off all logging helps, but I would really like to avoid the system
> calls and directory tree walking involved with an on-disk cache.
>
> This is for Squid 1.1, although suggestions for v2 are welcome. I would
> like to have a nice-sized RAM cache to keep in-process data (say 200MB),
> but I want to eliminate any filesystem access by Squid. Any ideas?
>
Received on Tue Jul 13 1999 - 10:27:52 MDT