Re: [K12OSN] SCSI vs IDE
Steve Langasek (vorlon@dodds.net)
Sun, 13 Oct 2002 20:08:37 -0500
On Sun, Oct 13, 2002 at 05:50:14PM -0700, Eric Harrison wrote:
> You ever wonder if SCSI drives are really that much better than IDE?
> Suspecting the hard drive, we dropped in a roughly-equivilant SCSI drive
> in place of the IDE (both 7200 RPM Seagates) and the performance increased
> by a magnitude. We ran thirty clients, all with KDE and all running different
> applications, without a problem. It definately was slower with thirty active
> clients, but it was perfectly acceptable.
> Has anyone else tried to use IDE drives with a terminal server, especially
> on /home, where all the heavy I/O occurs?
It should be noted that both "IDE" and "SCSI" cover a broad range of
compatible drive technologies spanning nearly 20 years of computing, and
as such, a simple comparison of "SCSI is better than IDE" is meaningless.
It's true that in general, SCSI performs better than IDE from the same
period. The bigger advantage SCSI has, IMHO, is that the target market
for SCSI consists of users that expect the drives to be reliable -- so the
mean time between failures tends to be much higher than for their IDE
counterparts.
When you were using IDE drives, did you try to tweak the IDE settings at
all with hdparm? It seems no one has solved the problem yet in the Linux
kernel that attempting to automatically turn on IDE optimizations for
newer hardware can crash some older hardware, so usually there's quite a
bit that can be done to boost IDE performance by twiddling bits with
hdparm (-c and -d, in particular).
Cheers,
Steve Langasek
postmodern programmer
_______________________________________________
K12OSN mailing list
K12OSN@redhat.com
https://listman.redhat.com/mailman/listinfo/k12osn
For more info see