Re: [squid-users] Squid sizing..

From: Don Campbell <don@dont-contact.us>
Date: Thu, 22 Mar 2001 11:28:21 -0500 (EST)

So maybe a better question would be.. Given a T1 of outgoing connectivity,
how much drive space should I allocate, which leads to the amount of memory
needed.. I arbitrarily picked a 40Gig drive because its raw performance is
50% higher than the 20Gig 7200 rpm drives I have, but that doesn't mean I
really need to use all the space.. Maybe I only need to cache a weeks worth
of data, maybe two weeks, I don't know..

I do know, that for another $88 I could bring the box upto 512M ram.. But is
that the right solution.. Basicly, I'm looking for a best practice sizing
solution..

-Don

>
>
> On Thu, 22 Mar 2001, Don Campbell wrote:
>
>> Hello,
>>
>> I'm in the process rolling out a production squid cache.. The
>> server specs are: Pentium II 266mhz, 256 Meg Ram, 40 Gig 7200 RPM IDE
>> disk.. Running FreeBSD 4.2 and using WCCP v1 to a cisco router..
>> Outgoing connectivity is a T1, and this is in an ISP environment with
>> a mix of dialup, isdn and frac T1 clients.. I'm thinking a
>> 30gig slice for squid and only give 20 gig to squid for use. Based on
>> suggestions I've found in the archives I'm planning on 46 level one
>> directories.. Does this sound like a reasonable configuration?
>
> You probably won't have enough memory. I'd suggest doubling your RAM.
> Hopefully memory prices are still really low....
Received on Thu Mar 22 2001 - 08:28:27 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:58:47 MST