file spread over multiple platters?

[J.Clarke]

In a typical IDE hard drive, is a large file spread over
multiple platter sufaces in parallel (perhaps allowing
more than one read/write head to be active at once),
or is the file spread over multiple cylinders sequentially
so that the head must jump from cylinder to cylinder
to access the entire file? The question has to do with
whether a large disk drive with multipl platters has
faster file access than a hard drive with just one platter
surface.

No IDE drive currently on the market, and as far as I know no drive of
any kind currently on the market, does parallel reads or writes--they
all use one head at a time.

 

[Timothy Daniels]

In a typical IDE hard drive, is a large file spread over
multiple platter sufaces in parallel (perhaps allowing
more than one read/write head to be active at once),
or is the file spread over multiple cylinders sequentially
so that the head must jump from cylinder to cylinder
to access the entire file? The question has to do with
whether a large disk drive with multipl platters has
faster file access than a hard drive with just one platter
surface.


*TimDaniels*

 

[Rod Speed]

In a typical IDE hard drive, is a large file spread over
multiple platter sufaces in parallel (perhaps allowing
more than one read/write head to be active at once),
or is the file spread over multiple cylinders sequentially
so that the head must jump from cylinder to cylinder
to access the entire file?

Yes, basically because switching heads is a lot quicker
than moving the heads from cylinder to cylinder.

The question has to do with whether a large disk
drive with multipl platters has faster file access
than a hard drive with just one platter surface.

Thats more complicated than it first looks, particularly
when the two drives have the same capacity.

Obviously the single platter drive would normally
have a lot more sectors per track to get the same
capacity out of a single platter. That determines
how fast the sectors move under the heads if the
RPM is the same with the two drives, and that is
another crucial factor in the speed of read and writes.

 

[Timothy Daniels]

Yes, basically because switching heads is a lot quicker
than moving the heads from cylinder to cylinder.

By "switching heads" you seem to be implying that
if a file spanned a track, the sectors from one track
would be read and then the subsequent sectors
would be found in the same cylinder, but on another
platter surface. That way, the arm wouldn't have to
be moved but merely a different head would be used.
Is that correct?

Thats more complicated than it first looks, particularly
when the two drives have the same capacity.

Obviously the single platter drive would normally
have a lot more sectors per track to get the same
capacity out of a single platter. That determines
how fast the sectors move under the heads if the
RPM is the same with the two drives, and that is
another crucial factor in the speed of read and writes.

If a file spanned platter surfaces before it spanned
cylinders, access would be faster because arm
movement wouldn't be needed. That would require
multiple platter surfaces. Does having multiple
platter surfaces available to do that make large file
accesses faster?


*TimDaniels*

 

[Rod Speed]

Rod Speed wrote

By "switching heads" you seem to be implying that if a
file spanned a track, the sectors from one track would
be read and then the subsequent sectors would be
found in the same cylinder, but on another platter surface.
That way, the arm wouldn't have to be moved but
merely a different head would be used. Is that correct?
Correct.

If a file spanned platter surfaces before it spanned cylinders, access
would be faster because arm movement wouldn't be needed.

Correct. BUT if you have say half the number of platters,
you'd have twice as many sectors per track, so you wouldnt
even need to switch heads at all when writing the larger track.

That would require multiple platter surfaces. Does having multiple
platter surfaces available to do that make large file accesses faster?

Like I said, thats more complicated than it looks, because its even
faster if all the sectors are in a single track than spread over two
tracks in a cylinder. Then you dont even need a head switch.

Another complication tho is the initial access to the first sector
in the file. Clearly once the heads are moved to the correct
track, you still have to wait for the sector you want to pass
under the heads. That also requires a substantial amount of time
to happen. Thats why drive RPMs have been increased over time.

 

[Timothy Daniels]

"Rod Speed" offerred his best thoughts in response to:>

Correct. BUT if you have say half the number of platters,
you'd have twice as many sectors per track, so you wouldnt
even need to switch heads at all when writing the larger track.

Your logic is impeccable. Are we always to expect this
level of expertise from you?


*TimDaniels*