From owner-freebsd-current Tue Aug 26 18:59:14 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id SAA25163 for current-outgoing; Tue, 26 Aug 1997 18:59:14 -0700 (PDT) Received: from sendero-ppp.i-connect.net (sendero-ppp.i-Connect.Net [206.190.143.100]) by hub.freebsd.org (8.8.7/8.8.7) with SMTP id SAA25141 for ; Tue, 26 Aug 1997 18:59:04 -0700 (PDT) Received: (qmail 7017 invoked by uid 1000); 27 Aug 1997 01:59:21 -0000 Message-ID: X-Mailer: XFMail 1.2-alpha [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit MIME-Version: 1.0 In-Reply-To: <199708261857.LAA23349@MindBender.serv.net> Date: Tue, 26 Aug 1997 18:59:21 -0700 (PDT) Organization: Atlas Telecom From: Simon Shapiro To: "Michael L. VanLoon -- HeadCandy.com" Subject: Re: IDE vs SCSI was: flags 80ff works (like anybody doubted it) Cc: "Jordan K. Hubbard" , current@freebsd.org, Ollivier Robert Sender: owner-freebsd-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi "Michael L. VanLoon -- HeadCandy.com"; On 26-Aug-97 you wrote: ... > You forgot a condition: Given unlimited CPU cycles, and a limited > budget, IDE is much ``better'' than SCSI. Absolutely true :-). However, if you have the money for unlimited CPU cycles, should you not have the money for unlimited - something + decent disk? Reminds me of all the Lose95 users haggling $1.00/month off their ISP bill to save money... Well, not so bad... :-) ... > And start loading it with processes, while accessing multiple drives, > possible for interleaved swap, various disk-accessing processes, > and/or striped partitions. You'll really wish you were using SCSI in > that scenario. But is it not what ``real'' computers do? My Heathkit H8 was faster than my fingers (definitely faster than my brain). IF we are building servers, toss away the single-user desktop mentality. This was supposed to be my point. ... > >Under 2.2, we see the saturation point at about 900 disk I/O ops/sec. > >Under 3.0 we see just over 1,400. Again, the test method was different, > >so these results are not meaningful. Our target was proven 800. We are > >happy. > > I would think the disk subsystem would be the primary limiting factor > here. What mix of controllers and drives were these tests run on? 1 DPT controller, 32MB cache, 6 Barracuda 4GB drives. Minimal RDBMS configuration. To saturate the CPU, on a twin P6, you will need about 100-200 disks, depends on how they are controlled. > It would also be interesting to run this simulation against a striped > set of SCSI drives. It would also be enlightening if you ran the same > test against your striped set of IDE drives. This is what the DPT did in these tests (RAID-0 across 2 busses, 32K stripes). If the IDE controller is capable of true DMA, rather than PIO, as long as multiple accesses to a drive and all drives are masters, I suspect it will (should) be faster than SCSI. Simon