At 14:31 09/22/2005 -0700, Tracy R Reed wrote:
>Gus Wirth wrote:
>> ATA drives have about twice the
>> seek times as SCSI drives and an order of magnitude worse error rates than
>> SCSI drives.
>> 
>> I'm being a bit pendantic here to help out the lurkers. I know you (Tracy)
>> already know this stuff. Maybe it's time I gave my hard drive lecture
again.
>
>Thanks for the info. Actually, I wasn't aware of ATA having twice the
>seek times as SCSI and an order of magnitude worse error rates. Why is
>that? Seek time is just a matter of the mechanics of the head mechanism,
>right? Do the heads differ between ATA and SCSI?

The seek time is determined by several factors. One is the servo loop that
places the heads on the tracks. The ability of the head to position itself
on the track with minimal settling time depends on the head mass, the
driving force of the servo, the damping factors, the size of the track, and
the feedback sensor that measures where the head is on the track. SCSI has
a superior set of servo controls compared to ATA drives.

>And where do the higher
>error rates come from? From read errors on the media (I have heard as
>yet unsubstantiated rumors that SCSI devices get the media that tests
>higher for quality but that it all comes from the same production
>process) or errors introduced in the ATA bus?

The higher error rates come from a combination of areal density (SCSI is
lower) and servo/head design. With a lower areal density, you get fewer
bits per square inch but you have a stronger magnetic field per bit.
Therefore you have less likelyhood of misreading the bit as it goes flying by.

All the hard drive manufacturers have white papers on their web sites that
provide pretty decent detail on how all this stuff works. IBM used to have
the best; I'm not sure what Hitachi has done with it since they bought IBM
out. Seagate also has some good tech data, since they are one of the
premier makers of SCSI drives.

Gus


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to