Re: SCSI disks

Torbjorn Lindgren (tl@funcom.com)
Mon, 2 Jun 1997 12:08:34 +0200 (MDT)


On Mon, 2 Jun 1997, W. Reilly Cooley wrote:
> I hate to continue to beat this subject into the ground, but exactly what are
> SCSI hard drives considered to the superior to IDE? According to the fellow
> who wrote "Upgrading and Repairing PCs" (who, as a PC technician, I don't
> wholly trust), SCSI hard disks a mostly IDE disks with an additional SCSI
> interface, (ie, the head-disk-assembly is the same, and most of the control
> electronics.)

The head-disk-assembly is the same, and so are the core control
electronic. This doesn't mean that it's an IDE disk with extra SCSI logic.

IDE disks also need extra interface logic, albeit slightly less involved
(which doesn't do much difference for cost nowadays).

This distinction is important since there *have* been cases where other
disks have been used to create SCSI disks. The example I know of died a
merciful dead long before the IDE interface was born.
They used MFM/RLL disks to create SCSI disks, and the performance of the
junk was quite bad :-)

Note that the high-end disk assembly are usually only available in SCSI
flavors, such as 7200+ RPM disks.

> It seems to be generally considered that SCSI is superior to IDE, but from
> this guy's writing (excepting for multiple devices on the SCSI bus or really
> fast HDs) SCSI provides no benefit. Pls. someone cursorily clarify.

With a good SCSI controller you get the same data to the CPU with much
less CPU overhead, even compared to a busmaster IDE controller.

For a small system this might not make that much difference, especially
given the cheap and quite good IDE busmaster interfaces available now. In
bigger installations, with many machines, other factors might tilt the
balance towards SCSI even for smaller machines with one/two disks.