For our solution we took a look at pricing and determined it would be
cheaper to just go with a single drive for each of our squid servers. Our
layout, in accelerated mode is like this:
Internet --> fw --> squid as a reverse proxy (rp) --> fw --> rp --> fw -->
back end web servers <--rp <-- network users
Two sites, 3 squid servers at each, which will give us reduntancy in case of
an ISP failure. A spare server at each site is cheaper than going with a
raid set up and redundancy failovers on each individual servers. The data
stored on the squid servers is not mission critical and can be easily
replaced. Every situation is different though. For us, an hour or even 4
hour's worth of downtime was not deemed critical to our business.
Chris
-----Original Message-----
From: A. Sajjad Zaidi [mailto:sajjad@iinix.com]
Sent: Tuesday, August 10, 2004 2:17 AM
To: Kvetch
Cc: squid-users@squid-cache.org
Subject: Re: [squid-users] RAID question
On Mon, Aug 09, 2004 at 12:36:28PM -0400, Kvetch wrote:
>
> I have about 30 sites running on these servers. Is setting up one
> cache_dir my best solution? How do I determine what size I should
> make my cache_dir?
From what I've gathered, setting up 2 or more caches on separate drives is
better than 1 big cache. This is what I did with my setup (2 x 36GB).
There is one 18GB partition for cache at the start of each drive (where the
transfer rate is better) and the rest is used for logs and the OS.
Ideally, I would have used four drives, making 2 RAID1 mirrors and dividing
the cache among those, but was limited by the available hardware.
-- A. Sajjad Zaidi GnuPG Key ID: 0xD7AD0E13 "Boolean values, so simple yet so easily confused."Received on Tue Aug 10 2004 - 05:07:10 MDT
This archive was generated by hypermail pre-2.1.9 : Wed Sep 01 2004 - 12:00:02 MDT