From owner-freebsd-hardware Wed Feb 21 17:39:40 2001 Delivered-To: freebsd-hardware@freebsd.org Received: from mass.dis.org (user-uinjtfm.biz.mindspring.com [165.121.245.246]) by hub.freebsd.org (Postfix) with ESMTP id D16A837B65D for ; Wed, 21 Feb 2001 17:39:36 -0800 (PST) (envelope-from msmith@mass.dis.org) Received: from mass.dis.org (localhost [127.0.0.1]) by mass.dis.org (8.11.1/8.11.1) with ESMTP id f1M1f8T02912; Wed, 21 Feb 2001 17:41:08 -0800 (PST) (envelope-from msmith@mass.dis.org) Message-Id: <200102220141.f1M1f8T02912@mass.dis.org> X-Mailer: exmh version 2.1.1 10/15/1999 To: The Hermit Hacker Cc: freebsd-hardware@freebsd.org Subject: Re: IOPs ... what exactly are they? In-reply-to: Your message of "Wed, 21 Feb 2001 14:56:03 -0400." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 21 Feb 2001 17:41:07 -0800 From: Mike Smith Sender: owner-freebsd-hardware@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org (this should not have been sent to -smp, redirected) > Searching google isn't coming up with anything that explains it ... can > someone point me to some docs, or provide an explanation? An "IOP" is an I/O operation. IOPs are the MIPS of the I/O world. > For instance, Tom stated that the 170 does 4000IOPs, vs the 352 at > 7000IOPs ... yet, the 352 will do twice the sustained transferof the 170 > (200MB/s vs 100MB/s) ... so, IOPs don't == transfer speed ... That's correct. The IOP rating is a function of the CPU speed and available I/O; the 352 has twice as many busses as the 170 (hence 200MB/ sec vs. 100), but it's CPU-limited to less than twice on total I/Os per second. > And, I imagine (can't get to the Mylex web site right now for some > reason), the 2000 has an even higher rating then both of those ... Yes; it has 4 channels and more CPU. > So, when evaluating what one needs, how do you determine? are there > thresholds one can work with? 4000IOPs will comfortably handle n xgig > drives, but if you go above n, then you really need to jump to the 7k IOP > level else performance drops substantially? You ignore the IOP rating entirely. > We're looking at this for a database server, so will most likely be going > to the 64MB cache level automatically ... but beyond that, I'm virtually > uneducated :( Adding cache isn't really useful; it's like a rubber band - unless you have enough real bandwidth, eventually it'll stretch as far as it can go. Work out how much I/O your server is going to do, and then scale your disks accordingly. -- ... every activity meets with opposition, everyone who acts has his rivals and unfortunately opponents also. But not because people want to be opponents, rather because the tasks and relationships force people to take different points of view. [Dr. Fritz Todt] V I C T O R Y N O T V E N G E A N C E To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hardware" in the body of the message