Date: Mon, 21 Aug 2006 19:25:05 +0200 (CEST) From: bsdml@werner.st To: "Chris Knipe" <savage@savage.za.org> Cc: Dan Nelson <dnelson@allantgroup.com>, freebsd-questions@freebsd.org Subject: Re: sluggish disk performance. Message-ID: <15985.85.126.91.51.1156181105.squirrel@webmail.werner.st> In-Reply-To: <003701c6c540$d0744f20$0a01a8c0@superman> References: <000701c6c539$bbb33710$0a01a8c0@superman> <20060821162156.GB45306@dan.emsphone.com> <003701c6c540$d0744f20$0a01a8c0@superman>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi >> In the last episode (Aug 21), Chris Knipe said: >>> Disks ad0 ofod intrn >>> KB/t 16.83 %slo-z 35456 buf >>> tps 103 4 tfree 414 dirtybuf well, with 100% random I/O at 7200rpm (or 120rpm per second = maximum number of theoretical IOPS) that is a typical number I see everywhere. The 16KB average I/O-Size are OK but could be higher - but definitely your I/O-Pattern an not the avg. transactions size are the cause of your 1.7MB/sec >>> MB/s 1.70 20988 desiredvnodes >>> % busy 98 5247 numvnodes >>> 4223 freevnodes >>> >>> Got 1 ATA100 Seagate 120GB disk in there at the moment.... 1.7MB/s at >>> 98% busy? Surely, that figure is WAY low??? I'd expect atleast >>> about 10MB/s on ATA100. >> >> That number's about right for random I/O and small blocksizes, which is >> what the KB/t field shows. If you were doing sequential I/O, the KB/t >> field would be at or near 128. Are you also running a "du", "cvs >> update", or other command likely to be doing random disk accesses? > > ALTER TABLE on a 200MB mySQL table? > > I guess its time for a dedicated disk then.... Depends on the number of random I/O's you want to be able to handle - could be that you might want 2 or more aswell :-) Cheers, Martin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15985.85.126.91.51.1156181105.squirrel>