From owner-freebsd-scsi Tue Apr 14 21:09:57 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id VAA11059 for freebsd-scsi-outgoing; Tue, 14 Apr 1998 21:09:57 -0700 (PDT) (envelope-from owner-freebsd-scsi@FreeBSD.ORG) Received: from sendero-fddi.simon-shapiro.org (sendero-fddi.Simon-Shapiro.ORG [206.190.148.2]) by hub.freebsd.org (8.8.8/8.8.8) with SMTP id EAA11053 for ; Wed, 15 Apr 1998 04:09:54 GMT (envelope-from shimon@sendero-fddi.simon-shapiro.org) Received: (qmail 29886 invoked by uid 1000); 15 Apr 1998 04:19:13 -0000 Message-ID: X-Mailer: XFMail 1.3-alpha-040798 [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <199804150054.SAA24735@narnia.plutotech.com> Date: Tue, 14 Apr 1998 21:19:13 -0700 (PDT) Reply-To: shimon@simon-shapiro.org Organization: The Simon Shapiro Foundation From: Simon Shapiro To: "Justin T. Gibbs" Subject: Re: RAID performance/benchmarking Cc: scsi@FreeBSD.ORG Sender: owner-freebsd-scsi@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On 15-Apr-98 Justin T. Gibbs wrote: > In article you wrote: >> >> Depends on array size, type of controller, amount of cache, type of >> access. >> RAID arrays are not a good benefit for sequential access. > > This really depends. For Pluto's application, using RAID 3 not only > gives > us reliability, but also the ability to have one of the drives in a > stripe > "return late", but still maintain low latency by replacing the data > through > parity reconstruction. You are one of the few who still build RAID-3 arrays. Interesting... Since we are a realtime system where being even a > little late is unacceptable, the use of RAID gives us a big advantage. > Almost all of our accesses are sequential. I did not say that RAID arrays are not functionaly correct for Sequential access. Just that perfromance in RAID arrays is les than optimal when doing sequential access. In your case you ``do not have a choice'' but to use a RAID (3, 5 or otherwise), as you cannot afford data disruption. Simply, there is nothing free. If you want the reliability and availability that a redundant array provides, there will be a performance and capacity penalty. BTW, the next generation DPT cards does all the parity computations in hardware. Makes some difference in WRITE perfrormance and in degraded READ. ---------- Sincerely Yours, Simon Shapiro Shimon@Simon-Shapiro.ORG Voice: 503.799.2313 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message