Return to BSD News archive
Newsgroups: comp.os.linux.misc,comp.os.linux.networking,comp.os.linux.setup,comp.unix.bsd.bsdi.misc,comp.unix.bsd.misc
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.mel.connect.com.au!news.syd.connect.com.au!phaedrus.kralizec.net.au!news.mel.aone.net.au!grumpy.fl.net.au!news.webspan.net!www.nntp.primenet.com!nntp.primenet.com!enews.sgi.com!arclight.uoregon.edu!su-news-hub1.bbnplanet.com!news.bbnplanet.com!cam-news-hub1.bbnplanet.com!howland.erols.net!surfnet.nl!news.unisource.nl!xs4all!plm.xs4all.nl!plm
From: Peter Mutsaers <plm@xs4all.nl>
Subject: Re: Linux vs BSD
X-Newsreader: Gnus v5.2.39/Emacs 19.34
Sender: plm@localhost.xs4all.nl
Organization: My Unorganized Home
Lines: 50
Message-ID: <873evtxn6t.fsf@localhost.xs4all.nl>
References: <32DFFEAB.7704@usa.net> <5c155c$p6u@raven.eva.net>
<5c19pg$rf6@lynx.dac.neu.edu> <5c341j$3dp@cynic.portal.ca>
<5c3k6o$qro@lynx.dac.neu.edu>
Date: Thu, 23 Jan 1997 00:14:18 GMT
Xref: euryale.cc.adfa.oz.au comp.os.linux.misc:153671 comp.os.linux.networking:65797 comp.os.linux.setup:93421 comp.unix.bsd.bsdi.misc:5681 comp.unix.bsd.misc:1971
>> On 21 Jan 1997 18:41:12 -0500, mkagalen@lynx.dac.neu.edu (Michael
>> Kagalenko) said:
MK> ]I see this claim regularly to, and I recently tried to investigate
MK> ]it. Unfortunately, I have found only one Linux user who is willing
MK> ]to give me any statistics on NFS performance, and his network and
MK> ]server were heavily loaded.
MK> That's right, there is a conspiracy amongst Linux users to deny you
MK> the data.
MK> ]Under Linux, last time I had direct experience with it I saw
MK> ]50-100K/sec read and write performance under NFS. The person I
MK> ]talked to more recently, with the heavily loaded network and servers,
MK> ]saw 150-200K/sec performance for read and write. If anyone out
MK> ]there has better figures (preferabily bonnie output, along with
MK> ]descriptions of the machines and whether this is NFS v2 or NFS v3),
MK> ]I would love to see that information.
MK> Clearly, you have no idea what you are talking about. Disk
MK> performance benchmarks are limited by the narrowest
MK> bottleneck in the system. Linux system in question is most
MK> likely to use IDE drives, whereas workstations (to which you
MK> misleadingly refer as "486-class machines"; no workstation is
MK> "486-class machine", even if CPU speeds are comparable) use
MK> SCSI-drives (and high-performance buses). I encourage you to
MK> read the documentation that comes with benchmark that you are
MK> using.
Todays IDE drives are not much slower than SCSI drives, but anyway:
I have a P90 with PCI SCSI and quite fast drives. I ran both FreeBSD
and Linux for a long time, until 2 weeks ago (when I deleted Linux).
About 3 weeks ago I did some tests for myself, using as client a P120
notebook, also both Linux and FreeBSD installed (current versions). I
only tried Linux<->Linux and FreeBSD<->FreeBSD, so I cannot tell
whether the difference was caused by the client or the server.
I can confirm the FreeBSD data, about 400k/s writing and just under
900k/s reading.
Linux did much worse: writing was 100k/s. reading was better, about
400k/s.
Straight FTP transfer did about 900k/s on both Linux and FreeBSD (so I
assume it is not a specific ethernet driver issue).
I don't know what NFS version I used. I think Linux only has version
two.
--
Peter Mutsaers | Abcoude (Utrecht), | Trust is a good quality
plm@xs4all.nl | the Netherlands | for other people to have