From owner-freebsd-stable Wed Sep 15 10: 8:44 1999 Delivered-To: freebsd-stable@freebsd.org Received: from out1.mx.skynet.be (out1.mx.skynet.be [195.238.2.36]) by hub.freebsd.org (Postfix) with ESMTP id DF6C814D03 for ; Wed, 15 Sep 1999 10:08:34 -0700 (PDT) (envelope-from blk@skynet.be) Received: from [195.238.1.121] (brad.techos.skynet.be [195.238.1.121]) by out1.mx.skynet.be (8.9.3/odie-relay-v1.0) with ESMTP id TAA01762; Wed, 15 Sep 1999 19:08:18 +0200 (MET DST) Mime-Version: 1.0 X-Sender: blk@foxbert.skynet.be Message-Id: In-Reply-To: References: Date: Wed, 15 Sep 1999 19:05:15 +0200 To: Adrian Filipi-Martin From: Brad Knowles Subject: Re: Vinum performance testing... Cc: freebsd-stable@FreeBSD.ORG Content-Type: text/plain; charset="us-ascii" ; format="flowed" Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG At 12:29 PM -0400 1999/9/15, Adrian Filipi-Martin wrote: > Nice. I can use these graphs when making arguments about whether > to buy DPT's or other hardware. Keep in mind that these tests were done with old DPT hardware, not their most current stuff. > My only question is have you thought about adding system CPU usage > to the graphs? It seems that under most cases CPU cycles are the only > thing the DPT's are saving, but I have no idea how many cycles are spent by > vinum. I was calling rawio with a particular set of command-line arguments as suggested to me by Greg Lehey (the author of vinum and rawio), so CPU utilization information was not captured. I believe that this data is available with different sets of arguments to rawio, but the output would be quite a bit messier and not nearly so easy to turn into a spreadsheet. My inclination would be to continue to run the tests in basically the same way I have been, but to instead also run some additional tests on certain specific test configurations of interest (e.g., vinum 4-Way stripe with 256KB stripe size and 256 sector average transfer size) to capture this information only for those specific tests. This way you've got the kind of CPU utilization information you want, but you didn't waste a lot of time and effort in attempting to collect that information for the entire battery of tests. I can say that the few tests I do recall running that provided more detail, CPU utilization from vinum was very minimal. One test I have a record of running on this same configuration reported the following: /usr/local/bin/rawio -s 45660241920 -v 1 /dev/vinum/data1 >Test name: anon >Transfer count: 32768 >Record count: 16384 >Process count: 8 >Device size: 2147483647 >Test ID Time KB/sec %User %Sys %Total Reads Writes >RR anon 781.799011 2700.7 0.1 6.7 6.8 131072 0 >SR anon 994.598389 4318.3 0.1 7.3 7.3 131072 0 >RW anon 2871.548584 736.3 0.0 1.9 1.9 0 131072 >SW anon 195.158478 22007.6 0.2 60.8 60.9 0 131072 Of course, this was with older versions of vinum and rawio, and I believe that the %Sys percentage reported was a bug that was fixed long ago. Except for the Sequential Write test, you can see that the overall CPU utilization was less than 10% across the board, and less than 2% for the Random Write test (although that number seems suspiciously low and might also have been the result of a bug which has almost certainly been fixed since). -- These are my opinions -- not to be taken as official Skynet policy ____________________________________________________________________ |o| Brad Knowles, Belgacom Skynet NV/SA |o| |o| Systems Architect, News & FTP Admin Rue Col. Bourg, 124 |o| |o| Phone/Fax: +32-2-706.11.11/12.49 B-1140 Brussels |o| |o| http://www.skynet.be Belgium |o| \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/ Unix is like a wigwam -- no Gates, no Windows, and an Apache inside. Unix is very user-friendly. It's just picky who its friends are. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message