From owner-freebsd-hackers Wed Mar 31 18:59:50 1999 Delivered-To: freebsd-hackers@freebsd.org Received: from allegro.lemis.com (allegro.lemis.com [192.109.197.134]) by hub.freebsd.org (Postfix) with ESMTP id 3838114D25; Wed, 31 Mar 1999 18:59:44 -0800 (PST) (envelope-from grog@freebie.lemis.com) Received: from freebie.lemis.com (freebie.lemis.com [192.109.197.137]) by allegro.lemis.com (8.9.1/8.9.0) with ESMTP id MAA10046; Thu, 1 Apr 1999 12:29:24 +0930 (CST) Received: (from grog@localhost) by freebie.lemis.com (8.9.3/8.9.0) id MAA90659; Thu, 1 Apr 1999 12:29:23 +0930 (CST) Message-ID: <19990401122922.Q413@lemis.com> Date: Thu, 1 Apr 1999 12:29:22 +0930 From: Greg Lehey To: FreeBSD Hackers , FreeBSD SCSI Mailing List Subject: CCD and Vinum compared with new performance measuring tool Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.93.2i WWW-Home-Page: http://www.lemis.com/~grog Organization: LEMIS, PO Box 460, Echunga SA 5153, Australia Phone: +61-8-8388-8286 Fax: +61-8-8388-8725 Mobile: +61-41-739-7062 Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG In the past few weeks, I've been bitching about the fact that bonnie doesn't do what I want in measuring storage device performance. I've now solved the problem: I've written another program. You can pick it up at ftp://ftp.lemis.com/pub/rawio.tar.gz. From the man page: DESCRIPTION rawio tests the speed of the low-level character I/O device special in a concurrent environment. It is intended for comparisons of storage de- vices on a single system, and is not suited for cross-platform perfor- mance testing. By default, rawio spawns eight processes, each of which performs the same test. Four tests are available: Random Read The random read test reads varying length records from the speci- fied device special, starting at random positions within the file. Sequential Read The sequential read test reads constant length records from the specified device special, starting at the beginning of the file. Random Write The random read test writes varying length records to the speci- fied device special, starting at random positions within the file. Sequential Write The sequential read test writes constant length records to the specified device special, starting at the beginning of the file. Here is some sample output measuring vinum volumes and straight disks (remember, these are ancient hand-me-down pre-SCSI-1 CDC drives; only the comparison counts). da2 is the raw disk, ccd0 is a striped ccd, and s128k is a striped vinum volume with the same geometry (128 kB stripes): Test ID K/sec %User %Sys %Total I/Os RR da2 576759 0.0 0.7 0.7 800 RR ccd0 1224133 0.1 2.0 2.1 800 RR s128k 1220979 0.4 2.2 2.6 800 Test ID K/sec %User %Sys %Total I/Os SR da2 899264 0.0 0.4 0.4 800 SR ccd0 1903052 0.0 1.5 1.5 800 SR s128k 1925672 0.1 1.8 1.9 800 Test ID K/sec %User %Sys %Total I/Os RW da2 589874 0.0 0.8 0.8 800 RW ccd0 1380529 0.1 2.2 2.3 800 RW s128k 1370186 0.3 2.6 2.9 800 Test ID K/sec %User %Sys %Total I/Os SW da2 901745 0.0 0.5 0.5 800 SW ccd0 2114599 0.1 1.6 1.7 800 SW s128k 2116235 0.0 2.1 2.1 800 Not surprisingly, the performance figures for vinum are pretty much the same as for ccd; the somewhat higher CPU time is probably due to the debug aids I still have in vinum. Compared to bonnie, the results are pretty reproducible. Comments welcome Greg -- See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message