From owner-freebsd-current Sun Sep 7 23:24:50 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id XAA11221 for current-outgoing; Sun, 7 Sep 1997 23:24:50 -0700 (PDT) Received: from MindBender.serv.net (mindbender.serv.net [205.153.153.98]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id XAA11215 for ; Sun, 7 Sep 1997 23:24:45 -0700 (PDT) Received: from localhost.HeadCandy.com (localhost.HeadCandy.com [127.0.0.1]) by MindBender.serv.net (8.7.5/8.7.3) with SMTP id XAA10617; Sun, 7 Sep 1997 23:23:57 -0700 (PDT) Message-Id: <199709080623.XAA10617@MindBender.serv.net> X-Authentication-Warning: MindBender.serv.net: Host localhost.HeadCandy.com [127.0.0.1] didn't use HELO protocol To: Greg Lehey cc: Drew Derbyshire , Ollivier Robert , current@freebsd.org Subject: Re: lousy disk perf. under cpu load (was IDE vs SCSI) In-reply-to: Your message of Mon, 08 Sep 97 11:48:06 +0930. <19970908114806.25653@lemis.com> Date: Sun, 07 Sep 1997 23:23:52 -0700 From: "Michael L. VanLoon -- HeadCandy.com" Sender: owner-freebsd-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Warning: In continuing this discussion, I risk going off into a tangent. I feel this is still information relevant to people planning on new hardware purchases. However, I think we have firmly established the players and the facts in this matter, and if you don't care to read more, skip this thread now. >> I think it's more because EIDE is cheaper, and there are a helluva lot >> of cheap EIDE drives out there. Plus, some people aren't willing to >> spend the extra on SCSI, but still want the performance. [...] >> IDE has always been an exercise in compromises. EIDE has just hacked >> on a lot of turbo-chargers and accelerators, making it much better at >> competing with SCSI on the low end. However, even UltraIDE (or [...] >The annoying thing is that the price isn't that much higher. It's >just because the volume market is (E)IDE that the prices difference is >exaggerated. Not entirely. Yes, there are _some_ drives where you can find identical models in EIDE and SCSI form. And they usually end up being slightly more expensive in SCSI form, where the largest part of the difference may simply be due to the relative manufacturing volues. But the majority of the SCSI market tends to be "higher-end" drives, in completely different models. For example, more mass-market oriented drives (read higher tolerances/ lower build quality) that spin 4500-5400 RPM at the max, making up most of the EIDE drive market. In the SCSI market, 7200 RPM is becoming common, and 5400 RPM is almost the bottom-line, with 10000 RPM drives now available. What's more, EIDE's potential is commonly compared with SCSI using EIDE hardware that exists in a laboratory, or is just starting to hit the market, and comparing to SCSI hardware that is widely available. For example, UltraIDE, with 33MB/s. Almost none of the motherboards currently in existance support this mode. Yet, drives are already being marketed this way. Compared with Ultra-Wide SCSI, at 40MB/s, which is widely available, and has been for a few years. I'm not blaming anyone, or saying this is an industry consiracy -- this is just the way things have played out, and it's good to be educated on what's going on. My final note is that over the last four years or so, I have had more (E)IDE drives fail (3) than SCSI drives (1) in my own systems. And I have owned more SCSI drives (11) than (E)IDE drives (5). Holding the drives in my hand, I can tell the difference in build quality. My experience is that there truly is a difference in the quality of components that go into each kind of drive. So, to end my little dissertation, I just want to say that I _don't_ think EIDE drives are "evil". I don't think this has to be a religeous issue. Which is also to minimize those who smugly say "Now, who said SCSI was better?" or "Now, who said SCSI was faster?" I just did. But if you fully understand the merits of each, and you still think EIDE is the best use of your money, then good for you. Personally, I only put EIDE drives in my Win95/DOS/Game machine, because SCSI just wouldn't give me any advantage there. But my NT box is entirely SCSI driven, just as my NetBSD box is. I don't currently have a FreeBSD box running, but when I did, it was also SCSI. One final ironic note of interest: for some time, I had one EIDE drive in my NetBSD system, which was used as an emergency backup boot, in case something on my main SCSI setup got toasted, or some other catastrophy. I have five SCSI drives in that system, three of which are older than the EIDE drive. The SCSI drives get used for all system activity, so are tickled constantly by cron, daily scripts, email, software builds, etc., with several minutes of non-stop load at least once per day. The EIDE drive just sat there, only having to receive a tarball once each day, and rotate old backup tarballs. I just had to send the EIDE drive back for repair/replacement because it developed a severe mechanical failure. All the SCSI drives are still running strong. The standard qualifiers apply. This is one guys' experience, on a single system, with a small number of drives. Your mileage and karma may vary. ----------------------------------------------------------------------------- Michael L. VanLoon michaelv@MindBender.serv.net Contract software development for Windows NT, Windows 95 and Unix. Windows NT and Unix server development in C++ and C. --< Free your mind and your machine -- NetBSD free un*x >-- NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3, Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32... NetBSD ports in progress: PICA, others... -----------------------------------------------------------------------------