Re: [squid-users] Performance question

From: John Halfpenny <jhalfpenny@dont-contact.us>
Date: Wed, 29 Jun 2005 11:21:25 -0400 (EDT)

i did read on a forum that someone advised the use of raid0. although presently i'm not at liberty to advocate it over the use of configuring squid for multiple drives as i haven't completed my testing yet.

what i +did+ say was that it beats a single ata in terms of general drive performance. you only have to perform a format to see (at a very basic level i grant you) that it is faster.

your comments about hw raid as not being particularly better are confusing, as hw raid on a scsi setup will almost certainly beat any ide configuration you care to mention. this isn't what i've done here, but if i had the money for a scsi raid card i wouldn't be digging out old kit to install squid onto. :)

the downside of a lost cache through raid0 is a risk that we have to face through lack of funds, and i'm prepared to mount another drive in place of the raid should it go down, which would only take a few minutes.

you are right in that you can install dansguardian and squid on the same machine (in fact, you need a proxy like squid to install dansguardian successfully anyway)- but we need ntlm authentication, which isn't currently included with dansguardian.

in the past i have setup two instances of squid on a box, one for the authentication and one for dans to 'sit on'. although it does work, try throwing a few hundred users at it- unless you've got *serious* kit, you get a serious meltdown in terms of performance. i would therefore always advise dans on a seperate machine if you need authentication. this also stays true to your opinions on redundancy, as a squid failure would still mean you can access the web through the dansg server & vice versa until you get the faulty one replaced.

 --- On Wed 06/29, Matus UHLAR - fantomas < uhlar@fantomas.sk > wrote:
From: Matus UHLAR - fantomas [mailto: uhlar@fantomas.sk]
To: squid-users@squid-cache.org
Date: Wed, 29 Jun 2005 17:03:18 +0200
Subject: Re: [squid-users] Performance question

On 29.06 03:42, John Halfpenny wrote:<br>> i'm in a similar situation. i'm replacing an isa box (something like a<br>> 2ghz machine with 1 big ata disk) with squid over the summer.<br>> <br>> we have a dual p2-400 server with 3 scsi disks in. i'm in the process of<br>> putting squid on it now using a 686 smp kernel to try and squeeze every<br>> ounce of speed from it!<br><br>SQUID itself won't benefit of multi CPU configuration, but the rest of OS<br>and diskd/aufs spool (using threads or different processes) will.<br><br>> one thing i can tell you at this stage, is that if you set up software<br>> raid0 to take care of the striping (rather than configuring squid to use<br>> the different drives) the access is way ahead of the performance of a<br>> single ata drive. and this is from a server which is around 6 years old.<br><br>Where did you get this? Your speed will be smaller when using RAID0, and<br>when you loose one of disks, you will loose whole cache with RAID0,
instead<br>of just part of it.<br><br>So, please tell me who provided you the informations that RAID0 (HS or SW)<br>is better?<br><br>> incidentally, we use a separate box for dansguardian with a fast processor<br>> in it. you really need that for dg.<br><br>It doesn't need to be separate box, if the CPU's are enough. If you have 2<br>CPUs in squid machine, you'll get one of them used only a bit, so you can<br>put DG onto the same machine, if the CPU is fast enough and you have enough<br>of memory.<br><br>-- <br>Matus UHLAR - fantomas, uhlar@fantomas.sk ; http://www.fantomas.sk/<br>Warning: I wish NOT to receive e-mail advertising to this address.<br>Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.<br>Spam = (S)tupid (P)eople's (A)dvertising (M)ethod<br>

_______________________________________________
Join Excite! - http://www.excite.com
The most personalized portal on the Web!
Received on Wed Jun 29 2005 - 09:21:30 MDT

This archive was generated by hypermail pre-2.1.9 : Fri Jul 01 2005 - 12:00:03 MDT