From owner-freebsd-amd64@FreeBSD.ORG Tue Apr 19 21:39:31 2005 Return-Path: Delivered-To: freebsd-amd64@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 72C9716A4CE for ; Tue, 19 Apr 2005 21:39:31 +0000 (GMT) Received: from error404.nls.net (error404.nls.net [216.144.36.24]) by mx1.FreeBSD.org (Postfix) with ESMTP id 5BEE443D3F for ; Tue, 19 Apr 2005 21:39:30 +0000 (GMT) (envelope-from ketrien@error404.nls.net) Received: from error404.nls.net (ketrien@localhost [127.0.0.1]) by error404.nls.net (8.13.1/8.13.1) with ESMTP id j3JLiFdN069056; Tue, 19 Apr 2005 17:44:15 -0400 (EDT) (envelope-from ketrien@error404.nls.net) Received: (from ketrien@localhost)j3JLiENT069055; Tue, 19 Apr 2005 17:44:14 -0400 (EDT) (envelope-from ketrien) Date: Tue, 19 Apr 2005 17:44:14 -0400 From: "Ketrien I. Saihr-Kesenchedra" To: Ken Gunderson Message-ID: <20050419214414.GD63598@bahre.achedra.org> Mail-Followup-To: Ken Gunderson , freebsd-amd64@freebsd.org References: <20050419202419.GB63598@bahre.achedra.org> <20050419151900.29d0ee57.kgunders@teamcool.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20050419151900.29d0ee57.kgunders@teamcool.net> User-Agent: Mutt/1.4.2.1i X-Virus-Scanned: ClamAV 0.83/841/Tue Apr 19 12:44:10 2005 on bahre.achedra.org X-Virus-Status: Clean cc: freebsd-amd64@freebsd.org Subject: Re: designing new freebsd server for amd64 arch X-BeenThere: freebsd-amd64@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Porting FreeBSD to the AMD64 platform List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 19 Apr 2005 21:39:31 -0000 On Tue, Apr 19, 2005 at 03:19:00PM -0600, Ken Gunderson wrote: > I loved IBM drives years ago, but started having too many issues when > they started cutting a few too many coreners. The things just started > dying prematurely. Hitachi drives also have a fairly bad rep amongst > serveral consultants I know. I'm reminded of a Demotivators poster; "There's good money to be made in prolonging the problem," to paraphrase slightly. e.g., yes, I'm calling them out. Observed and demonstrated statistical failure rates on the Hitachi 10K174 and 15K73 line are typically lower than Seagate and Fujitsu. Except when you get some kid slapping parts together and calling it a box. At which point, all failure rates go out the door, because the whole thing is a ticking time bomb. > To summarize, for the K8S Pro, it seems your suggestion would be to stay > away from the pci- x adapters all together?? Which leaves something > like the Adaptec 2120 or the LSI 320-1LP (for a 2U rackmount system). > Can anyone give me some guidance if either of these has advantages/ > disadvantages under fbsd amd64? No, PCI-X adapters on the A (133MHz) bus, which is where Tyan jammed all the onboard peripherals. This creates major timing migraines, especially at a 133MHz clock. The B bus (66MHz) does not have this problem, which leaves you with anything that can downclock to 66MHz. If it's PCI-X compliant, it can downclock to 66MHz. Hell, the Ultra320-2X can step down to PCI32/33. (I've actually done this for testing.) The catch is that you lose the bus bandwidth you may need for extremely intensive disk I/O. Of course, you really aren't going to be in any trouble until you get up into 1Gbit-2Gbit FC-AL. The 320-2 here does over 100MB/s read and 100MB/s write on 15K73's with RAID1. If you need more than that, you should probably start looking at an Iwill H8501 and it's triple 8131 setup. As far as advantages/disadvantages - last I'd heard, aac(4) is not 64-bit clean, nor is asr(4). Which means you aren't going to be running them on amd64. amr(4) is 64bit clean and >4GB safe, so really, it's the only choice. Well, besides Qlogic FC (isp(4)), but I don't know if anyone's even tested that. -ksaihr