From owner-freebsd-questions Thu Jan 7 19:20:58 1999 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id TAA20845 for freebsd-questions-outgoing; Thu, 7 Jan 1999 19:20:58 -0800 (PST) (envelope-from owner-freebsd-questions@FreeBSD.ORG) Received: from mail.HiWAAY.net (fly.HiWAAY.net [208.147.154.56]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id TAA20836 for ; Thu, 7 Jan 1999 19:20:56 -0800 (PST) (envelope-from dkelly@nospam.hiwaay.net) Received: from nospam.hiwaay.net (tnt3-1.HiWAAY.net [208.147.146.1]) by mail.HiWAAY.net (8.9.1a/8.9.0) with ESMTP id VAA32744; Thu, 7 Jan 1999 21:20:21 -0600 (CST) Received: from nospam.hiwaay.net (localhost [127.0.0.1]) by nospam.hiwaay.net (8.9.1/8.9.1) with ESMTP id UAA43955; Thu, 7 Jan 1999 20:38:15 -0600 (CST) (envelope-from dkelly@nospam.hiwaay.net) Message-Id: <199901080238.UAA43955@nospam.hiwaay.net> X-Mailer: exmh version 2.0.2 2/24/98 To: Forrest Aldrich cc: freebsd-questions@FreeBSD.ORG From: David Kelly Subject: Re: Clarion RAID subsystems on FreeBSD ? In-reply-to: Message from Forrest Aldrich of "Thu, 07 Jan 1999 14:20:56 EST." <4.1.19990107141906.009f0c50@206.25.93.69> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 07 Jan 1999 20:38:14 -0600 Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Forrest Aldrich writes: > Where I work, we just met with a vendor that resells Clarion > RAID arrays, also some tape libs. When we mentioned that > most of our servers are FreeBSD-based, we got the ol' sales > chat about free systems and support issues, blah blah. > > Firstly, we would need specialized drivers. I wondered if anyone > out in the FreeBSD world might have some experience in using > these arrays with a FreeBSD system. I have not used my StorageTek (a.k.a. Clariion, two eyes, right?) RAID with FreeBSD but your salesman is full of hot air and not worthy of the sale. The Clariion RAID hardware I have has a serial port that speaks plain old vt100 @ 9600 baud and can be used to anything the whiz-bang fancy GUI sorry interface they provide. Field service personnel prefer the serial interface. I got mine working before reverting to the serial interface myself, but have seen it used. Once configured the RAIDs you have set up simply appear as whatever LUN you have configured them on. The box remembers and doesn't have to be told at each power cycle. And in normal use all of the "special" software is turned off unless you want to read statistics off the thing. In my case they got the fonts so badly wrongly sized it is almost impossible to use their GUI anyhow. Instructions provided for changing the X default font size resource haven't worked either. Now where super special software comes into play is where they support SCSI bus failure and fallback to a redundant SCSI bus. My racks have (2) RAID processors (100 MHz PowerPC 603e's) with one external SCSI bus each, but both of these CPU's are connected to the internal 5 SCSI buses where the drives connect. Ie: either RAID CPU can access any of the 30 drive bays in the package. Catch-22: when I last bothered to ask, they didn't support the redundant fallback SCSI bus stuff on SGI Irix. Doubt I'd have complicated my system with it even if it was available. My SGI system already has 13 SCSI buses. Final-nail-in-coffin: The drives come in a neat hotswap caddy. You can't buy these caddies without a Clariion HD inside at just over twice the street price of similar drive. Plus the drives I have are a model of narrow FAST IBM drive exactly the twin of the FAST Wide drive in my FreeBSD box. But with Clariion custom firmware. Found out the firmware 1) is hard coded for 520 byte blocks, and 2) doesn't honor the standard SCSI bad block mapping commands. 3) a low level format doesn't eliminate or map out the bad blocks that are there. And they make a big deal about the RAID CPU boot firmware being stashed somewhere on only a few of the drives in the rack. We have a security phobia at work. Standard Procedure is a new HD gets its bad block list recorded before being used. If the HD is defective and has to go elsewhere it doesn't leave if its BB list has grown. If I can't get to the bad block list, then it goes to the crusher. All this means our Clariion HD's will never leave, alive. I didn't chose this hardware. The Powers That Be Decided We Needed It, and one day a semi truck trailer full of stuff arrived. (3) Cygnet 1803 jukeboxes were the bulk of the volume. Somewhere out of the blue they proclaimed we had a requirement for "sustained 100 MB/sec thruput" (this was to feed a T1). Did I mention we have (3) 30 drive bays each with (2) RAID CPUs? But only 8 HD's each (total of 24). To meet the performance requirement the installation crew configured (via serial ports) each rack of 30 to have (3) mirrored drive pairs (RAID 0?), one single drive, and one hot swap (total of 8). Once they left, I reconfigured the drives into 4 sets of RAID 5, two of these were in the same rack but each of those was on a separate CPU. Left one hot swap per rack, and one cold swap on the shelf. Only used 4 of my 6 RAID SCSI CPU's and only 4 of the 6 wide differential SCSI buses going to my SGI 2 CPU R10k Challenge L. Each HD is 9G (actually 8.3, if I recall). So to see how outrageous I could get (plus the user's were crying for disk space) in Irix I configured those (4) RAID 5's into one large striped logical volume of 130G. It was now fairly easy to light 20 HD access lights. But I only got 20 to 25 MB/sec thru the filesystem using dd to write from /dev/zero to a file. :-( If it were my choice I believe I would buy high quality disk drive racks and fill them with quality generic HD's. Pay SGI for their plexing option (software support for hot swap and other things). And build the RAID in software. As a fallback there are those who sell RAID CPU's in a 5-1/4" HH form factor that you could add. If the host were FreeBSD, I'd be talking to Greg Lehey about the commercial options in his vinum driver. Cost a whole lot less. Would get something I stand to make sense of when it fails. Forgot to mention one of the RAID CPU's forgot its config and I had to re-bind it? And rebuild the entire 130G fs from zero? Have to admit the Clariion hardware is first class. Each rack has redundant RAID CPU's (see above, the software wasn't available for this under Irix), redundant power supplies, lots of air flow, and at least 60 seconds of built in UPS so the thing can shut itself down cleanly. -- David Kelly N4HHE, dkelly@nospam.hiwaay.net ===================================================================== The human mind ordinarily operates at only ten percent of its capacity -- the rest is overhead for the operating system. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message