*BSD News Article 88333


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.mel.connect.com.au!news.access.net.au!news.netspace.net.au!news.mel.aone.net.au!grumpy.fl.net.au!news.webspan.net!newsfeeds.sol.net!hammer.uoregon.edu!news.mathworks.com!EU.net!Ireland.EU.net!maths.tcd.ie!maths.tcd.ie!not-for-mail
From: ajudge@maths.tcd.ie (Alan Judge)
Newsgroups: comp.unix.bsd.freebsd.misc
Subject: Re: FreeBSD 2.1.6 Striping Performance
Date: 5 Feb 1997 10:28:39 -0000
Organization: Dept. of Maths, Trinity College, Dublin, Ireland.
Lines: 35
Sender: mathnews@gosset.maths.tcd.ie
Message-ID: <5d9ncn$hv@gosset.maths.tcd.ie>
References: <5d87kv$sin@news.pcslink.com>
NNTP-Posting-Host: gosset.maths.tcd.ie
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.freebsd.misc:35039

ryan@pcslink.com (Ryan Mooney) writes:
>I was just playing with striping under 2.1.6 and I got fairly poor performance 
>results... 

You don't say what results you actually got, but I've a few suggestions
anyway.  First of all, a 10M test file is way too small to get realistic
numbers.  You're more likely to measure the performance of your memory
system.  Also, I'd try a larger interleave.  I doubt that the problem
is related to the version of FreeBSD or of the ahc driver, but I'll
leave better informed people to address that.  However, I can provide
some numbers for a 2.2-BETA system.

Hardware:
	PPro 180 (128M RAM, PCUI)
		System disk on internal SCSI
		Adaptec 2940UW with 3 Quantum Atlas II 4GB
		UltraSCSI
Striping: Full 4GB partition on each drive, interleave as below.
Performance tested with iozone (2GB file, 64K reads and writes)

Interleave	Read perf	Write perf
16		21MB/s		11MB/s
32              23.9            15.2
64              24.5            18.8
128             24              18.7
256             21.5            19.9
512             20              20
2048            11.7            20

You'll note that there is a slight trade off between read and write speed.
All of this is sequential of course.  Your application might vary.
-- 
Alan Judge 

Like punning, programming is a play on words.   -- Alan Perlis