But on the flipside, as the data is packed increasingly dense, more re-alignment would be required as temperatures differ throughout the entire drive stack. The old drives (ie: that old Barracuda) would not have required any re-alignment when reading off of different discs in the overall stack, because all of the servo data was read from a dedicated platter, and data densities were not yet high enough that differential rates of thermal expansion were a problem.
In the 'enterprise', the issue is that servers need to be able to pump out the io's. They need more actuators, more spindles. Not fewer. I personally don't believe coming up with an 11-platter, 500gb/platter drive, would be of any significant engineering challenge for a current producer of hard drives (in a 1.6" high package) such as Seagate. The problem becomes, would such a drive actually sell and be economical.
It would seem to be pretty hard, IMHO, to justify the cost of setting up a seperate assembly line, and to take lower manufacturing yields, simply to deliver what essentially would be a niche product only used by enterprise customers with large amounts of data but a low IOPS requirement. Can't see such a drive making any money without selling them at $1500+ a piece, obviously a no-go when 2Tb drives can be bought for $200.
I don't agree here; latency is dependent on RPM. Seek time is dependant on the performance of the linear actuator assembly, and the feedback control system that runs it. If you look at early 3600rpm and 5400rpm drives, especially, there are a wide range of different seek times amongst different models. Not so much in the 7200rpm drives, but most modern drives share basic mechanical designs that are 10-15 years old, hence, perform very similarily.
Last edited:
