Why don't HDD manufacturers read from all heads simultaneously?

[badass]

A quick look at the performance of HDD's all in the same family, only with more or less platters makes it obvious that they only read from 1 or 2 heads at once. Why is this?
Discuss

[Funkstar]

It's a good quastion. For years I assumed they did, but then worked out they didn't when I got to know more about them.

I have no idea, but there must be a fundamental reason why they don't as it would be an easy, massive gain in performance if they could.

[BobF64]

The heads dont move independantly, they move as a set.

Having them move independantly would require more motors and more logic or fewer heads and platters.

So I guess in theory they could be made that way, and therefore get faster reads.


They can, I believe, read from multiple platters at the same time, provided the head is over correct location.


Mind you, I have been known to be wrong before.

[CaptainCrash]

Apparently someone has filed a patent for a design which can do this, if you fancy a little light bedtime reading.

No idea if it's actually been made to work in practice, but the difficulty is explained in paragraphs 7/8/9 of "Background to the invention".

[arthurleung]

Feel free to prove me wrong, but if you use multiple heads, the magnetic field of each head would interfere with each other, preventing data to be written onto the disk reliably.

If you have two heads writing the same spot, one want to write a 1 (Lets say upward) onto disk 1, and one want to write a 0 (Downward) onto disk2. The two fields would cancel each other. You get a valid write on disk2, but not on disk1. There is also possibility that the two heads attracts to each other, causing the head to hit the disk surface.

--------------Head 1
------------------------------Disk 1
--------------Head 2
------------------------------Disk 2

You really DON'T want any force other than aerodynamics acting on the head. Otherwise the head will just crash onto the disk.
It will certainly work with optical media though.


Or if you place 2 actuator arms on the harddrive each reading off each end of the disc (Which means double the number of heads). It will work, but from the mechanical complexity and reliability point of view, why not just get 2 harddrives with single head.

[switchmode]

Feel free to prove me wrong, but if you use multiple heads, the magnetic field of each head would interfere with each other, preventing data to be written onto the disk reliably.

I can't prove it but I remember reading that the magnetic field to write to disk platter media has to be very very strong (Technical Term) Its confined to a tiny microscopic area just below the head its not penetrating. otherwise it would affect the data adjacent or on the other side of the platter. rubbing a powerful magnet over a platter is not enough to wipe it properly.

Its why residual magnetism is left from previous writes to the disk and can be read with more accurate reading heads in forensic examinations. the heads on subsiquent writes which may not be alligned exactly and leave the old data on the edge of the track or a bit deeper in to the track.


Its the head technology / magnetic media thats mostly made these massive capacity drives available in such a small form factor.


Storage review has an article on this, they state that the elecrtonics to process the data from the heads and send to the PC are not up to the job "on the drive" with ready available controllers (processing bandwidth etc). It may well be able to be done now though, it might just be a cost complexity decision now as electronics in Solid State drives prove they can be faster transfering data, sata for example can have very high sustained speeds.

[badass]

The heads dont move independantly, they move as a set.

Having them move independantly would require more motors and more logic or fewer heads and platters.

So I guess in theory they could be made that way, and therefore get faster reads.


They can, I believe, read from multiple platters at the same time, provided the head is over correct location.

They can't and that is what I am asking about.
Why not read from all heads at once or at the very least all top or all bottom heads. Access time wouldn't be affected much but sequential reads and writes would go through the roof!

For Proof that only one head (or possible 1 platter - i.e. 2 heads) is reading/writing at once, compare the sequential read/write speeds of drives from the same family that differ only in numbers of platters. Often the single platter drives are faster

[badass]

Feel free to prove me wrong, but if you use multiple heads, the magnetic field of each head would interfere with each other, preventing data to be written onto the disk reliably.

If you have two heads writing the same spot, one want to write a 1 (Lets say upward) onto disk 1, and one want to write a 0 (Downward) onto disk2. The two fields would cancel each other. You get a valid write on disk2, but not on disk1. There is also possibility that the two heads attracts to each other, causing the head to hit the disk surface.

--------------Head 1
------------------------------Disk 1
--------------Head 2
------------------------------Disk 2

You really DON'T want any force other than aerodynamics acting on the head. Otherwise the head will just crash onto the disk.
It will certainly work with optical media though.

Sounds plausible. Heat assisted recording may help with this

[badass]

Apparently someone has filed a patent for a design which can do this, if you fancy a little light bedtime reading.

No idea if it's actually been made to work in practice, but the difficulty is explained in paragraphs 7/8/9 of "Background to the invention".

That explains it
The tracks are only "Quasi Aligned" for various reasons and can be several microns out.