Single-drive RAID-0?

[Yamyam]

How difficult would it be to design a single-drive RAID-0 array? Here's how it might work: all read heads access the platters simultaneously, so a 3-platter design would operate as a 3-way (or 6-way) striped array. It would require a highly sophisticated controller, and it would be limited by the interface cable's bandwidth..... but it seems like a really convenient way to get very high performance out of small, moderately inexpensive product. Would noise or other cumulative factors prevent running several read heads in parallel?

Alternatively, would it be possible to implement two seperate sets of read heads and controller arms? Say, one for the top platter, and a second, independent system for the bottom platter?

Or (finally, last one I promise) how about two sets of read heads, 180 degrees apart, reading on two interleaved spiral tracks?

 

[Shalmanese]

Its been asked before, do a search back a couple of months.

Basically, it came down to cost/reliability/cost/sychronisation/cost issues.

some HD mfg came out with one ages back and it was a major flop.

 

[Gunbuster]

Separate arm/heads: Costs too much

Heads on the same arm writing at the same time: Thermal expansion/calibration hijinks ensue

 

[kpb]

I'm pretty sure segate had a baracuda way back in the day when they where thier top of the lin scsi drives that did that. Didn't last very long so i'd assume the difficulties associated with weren't worth the results.

 

[Yamyam]

Ok. Thanks for summarizing for an ignorant newbie

 

[Smilin]

In many ways, multiple platter drives used today already do what you mention.

There has been some research on hardware-based security for drives involving multiple heads - a normal one that is locked down and a separate one that is read-only.