[H-GEN] partition advice

[Jason Henry Parker]

Martin Pool <martin.pool at mincom.com> writes:

> davidj at amh.com.au wrote:
> > The primary FAT, or primary superblock is in the center of the drive, and
> > using some form of tree, get to anywhere in the disk in minimum _average_
> > time. 
> > The theory behind this is the disk heads are on average, located
> > in the center of the platters.
> Is that really true?  It sounds a little statistically suspect to me.  I
> seem to remember that the current-best seek algorithms spread the
> head-presence-probability evenly over the disk.

Now, I wasn't going to bring this up, but...  Exactly HOW do partition
names relate to the physical position of the data on the disk
platters?  Correct me if I'm wrong, but most "modern" disks (for
pretty large values of modern, as far as I'm aware) do their own
mapping of C/H/S to physical locations in any event.

(a good example of this is that disk geometries are given in
Cylinders/Heads/Sectors, when in fact, the number of sectors varies as
you get further toward the "hub" of the disk.)

(And another thing---if hda1 is a "fast" partition, then doesn't that
mean that the end of hdan (where n is the last partition on the disk)
is also "fast"?  :)

IMHO, bothering about where your partitions are is a bit pointless as
I doubt you'll get much of a performance increase by not having your
head move too far too often.  As Martin indicated, it's all about
caching anyway, so you might as well just get more RAM and never have
to touch your disk except on boot or write.  :)

Jason

-- 
`...if you're arguing I must have paid.' | Jason Henry Parker
`Not necessarily. I could be arguing in  | henry at freezer.humbug.org.au
 my spare time.'                         | http://fox.uq.net.au/~zzjparke/
   We are Borg, of Debian.  Resistance was futile.  We were assimilated.

-
This is list (humbug) general handled by majordomo at humbug.org.au .  Postings
are accepted only from subscribed addresses of lists general or general-post.