[5350] in linux-scsi channel archive
Re: Dumb question: Which is "better" SCSI or IDE disks?
daemon@ATHENA.MIT.EDU (Leonard Zhang System Administrator)
Tue Dec 8 03:03:51 1998
Date: Tue, 8 Dec 1998 09:27:29 +1000 (AEST)
From: Leonard Zhang System Administrator ISD RVIB <leonardz@rvib2.rvib.org.au>
To: Robert Minichino <rmini@joni.pasture.net>
cc: linux-scsi@vger.rutgers.edu
In-Reply-To: <199812071137.GAA09566@joni.pasture.net>
No one can totally imitate the real world. That is why the first release
of software has a lot of bugs :-)
To satify your curiosity, copy is done on real machines, with the whole
directory, reproducible. The hardware of SCSI is better than IDE. I don't
want to invest 2000 DPT to a XT :-)
Leonard
On Mon, 7 Dec 1998, Robert Minichino wrote:
> > > > I think currently spinning SCSI hard disks on the world, either in
> servers
> > > > or workstation, either in Unix/Linux or NT, is 50-pin old guys, no DPT
> disk
> > >
> > > And as my stats showed for real work even old 5400 rpm fast scsi on a
> > > now discontinued adapter (the BT946) beats current UDMA IDE for real
> world
> > > compiles. You "think". I've "measured"
> > >
> >
> > Open server 5.0.4p, DPT RAID-1, 32 MB cache, PCI, 4.2 GB (A cable)
> >
> > copy 9,177 K in 9 seconds.
> >
> > Red hat 4.2, 8 GB IDE
> >
> > copy 9,397 K in 2 seconds.
> >
> > both run in shell script.
>
> And many tape streamers are good at bulk transfer, too. Besides that, there
> is not nearly enough information here to conclude anything about either
> configuration. Are both operating systems run on the same system? Where are
> the partitions located on the disk (data transfer rates are different
> depending upon the track)? What are the specifications on the drives? Was
> the cache clean? What was the filesystem used on each OS? Was the
> filesystem set to do synchronous writes? Was the DPT cache written through?
> How much RAM was in each system? What was the processor speed on each
> system? Is the test reproducible?
>
> For a more accurate measure of disk throughput, grab Bonnie from
> http://www.textuality.com/bonnie/ and run with a file size of three times
> either the physical RAM size or controller cache size, whichever is greater.
> Post those results, along with the answers to the questions above, and then
> you might have a case. Or time some reproducable real-life tasks on setups
> that differ only in the drives/controller used.
>
> --
> Robert Minichino
> Chief Engineer
> Denarius Enterprises, Inc.
> http://www.denarius.com/
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
