From owner-freebsd-performance@FreeBSD.ORG Wed Apr 16 07:23:26 2003 Return-Path: Delivered-To: freebsd-performance@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id E5E4237B401 for ; Wed, 16 Apr 2003 07:23:25 -0700 (PDT) Received: from svaha.com (svaha.com [64.46.156.67]) by mx1.FreeBSD.org (Postfix) with ESMTP id F042143F75 for ; Wed, 16 Apr 2003 07:23:24 -0700 (PDT) (envelope-from meconlen@obfuscated.net) Received: from FUSIONBOX ([64.156.25.5]) (AUTH: LOGIN meconlen, TLS: TLSv1/SSLv3,128bits,RC4-MD5) by svaha.com with esmtp; Wed, 16 Apr 2003 10:23:23 -0400 From: "Michael Conlen" To: "'Alain Fauconnet'" , "'Jason Stone'" Date: Wed, 16 Apr 2003 10:23:17 -0400 Message-ID: <000a01c30423$bd0f2a40$2b038c0a@corp.neutelligent.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.4510 Importance: Normal In-Reply-To: <20030416024844.GC7867@ait.ac.th> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106 cc: freebsd-performance@freebsd.org Subject: RE: tweaking FreeBSD for Squid using X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 16 Apr 2003 14:23:26 -0000 A lot of what you face doing a Squid server is backplane and other bus issues, though it's dependant on what you call "high performance" A pair of Sun E220R's (2 SPARC II processors) for example handled 1 = million requests a day on a pair of mirrored 72 GB drives each. (Granted they = were very nice 72GB drives). The thing about the Sun boxes was that they = could get information out of memory really really fast, and the NIC cards = could work to their full potential. Every device that did IO was on it's own = PCI bus.=20 It used to be that IDE drives took more processing power from the host = to perform it's operations, where as SCSI does not. If that's still true = I'd use that as a reason to stay away from IDE.=20 The other advantage of SCSI, if you need great disk IO, is that you can = have a lot of spindles. On a large SCSI system in a Sun for example I can get = a single drive array to look like one SCSI device (with 14 disks in it) = and put a lot of arrays on a channel. If I buy small, fast SCSI disks I can = take full advantage of the 160 MB/sec array, where as I've seen a big fast = IDE disk push no more than 10 MB/sec. The arrays can do RAID before it gets = to the controller card, so you don't need the RAID in the box at all.=20 Speaking of which, does anyone know of SCSI disk arrays with hardware = RAID that work with FreeBSD?=20 I've moved out of the Sun world and in to the FreeBSD world = professionally and have no idea what's out there for PC hardware. -- Michael Conlen -----Original Message----- From: owner-freebsd-performance@freebsd.org [mailto:owner-freebsd-performance@freebsd.org] On Behalf Of Alain = Fauconnet Sent: Tuesday, April 15, 2003 10:49 PM To: Jason Stone Cc: freebsd-performance@freebsd.org Subject: Re: tweaking FreeBSD for Squid using On Tue, Apr 15, 2003 at 07:31:24PM -0700, Jason Stone wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 >=20 >=20 > > You didn't mention whether is it SCSI or IDE. >=20 > Based on the fact that it's 15k rpms, I'm guessing scsi - do ide disks > faster than 7.2k rpms exist? I don't know. Seems that IDE disks evolve too fast for me nowadays. That's also why I was writing that I'm not even sure that the old stance "don't use IDE for servers" is still valid. OOTH, I've had a lot of trouble with busy IDE-based (ASUS P4* = m/b) FreeBSD servers lately (hard hangs, see bug kern/44867). >=20 >=20 > > Don't do RAID. Bring up a filesystem on each disk (with soft = updates > > of course), mount them "-o noatime" and configure Squid to = use > > multiple cache dirs. >=20 > Why is this preferrable to striping with raid-0? Well, because Squid does load balancing over multiple cache dirs quite well by itself, and (presumably) in a smarter way than just spreading raw disk blocks, so adding another layer of software for RAID-0 doesn't bring anything, and wastes CPU cycles. I'm not even sure that hardware RAID-0 is a good idea. According to my own experience (admittedly on a Linux box), removing software striping and using multiple cache dirs on physical volumes gave me a significant performance boost and lowered the load average of the server by about 30%. Since then, I've always stayed away from RAID on Squid boxes, whatever the O/S. Er. This is getting off topic maybe. I'll follow up privately if needed. _Alain_ _______________________________________________ freebsd-performance@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-performance To unsubscribe, send any mail to "freebsd-performance-unsubscribe@freebsd.org"