DragonFly kernel List (threaded) for 2004-01
Re: Background fsck
Matthew Dillon wrote:
Second, on a modern hard
drive there isn't all that much of a difference between seeking a few
tracks and seeking half the disk because it actually takes as long for
the disk heads to settle on the target track as it does to actually seek
Er, this may or may not be true.
A friend of mine back at Seagate was the servo lead for one of the
Cheetah X15 products (LP144? anyway, I've tried my best to drown out the
acronyms and silly codes from back then). As I recall, they had it down
so that seeks under a certain distance happened much faster than those
above this threshold. Other drives (other Seagate models, various
competitor models, etc.) would show a roughly constant seek time.
The server drives were always much better than commodity (IDE) drives;
they would spend more time tuning those (the build of the parts as well
as the servo algorithms).
Take a look at the datasheet for the X15, which quotes both a
track-to-track and "average" seek time:
Also note how the commodity drives (80GB Barracuda IDE, in this case) don't: