...
I run a pair of 10k rpm sata drives...
The characteristics of the drive are the .... bottle neck.
I look at two aspects and am required to assume the 3rd as they are either no longer published ar more difficult to find than I care to expend the effort to find.
Rotational Delay - One of the most important aspects. This is generally looked at exclusivly as the rpms, but that can be a very big misleading thing to do without the details (were that pesky devil lives).
The amount of On board memory. The larger the better is my motto as at mechanical/physical world any drive the average person can buy simply can not transfer data to and from the platters / disks at a rate that is significant compared to the rate at which the on board controller can transfer data to / from its interface to the (assumed) cpu. Of course this to with out the details (were the pesky devil lives) can be very misleading as there are proofs that MS's caching and the on board memory can work at odds with each other and produce worse results than if one or the other were absent. Here too one needs to look at the details, when one does one sees the test being performed are not even remotely resembling what a typical home computer user would be doing.
The Third leg and the only one in my control, is the disk usage and the interface connection. If one were to buy a 7200rpm, 8mb 80gb Western Digital, Maxtor, Seagate hard drive (should be able to be bought for under $60US) and put it in a good USB2.0 / Firewire / ESATA drive case and connect that with the appropriate cable AND dedicate that connection to that drive (ESATA does this as it is a point to point communication protocol) then in my opinion and experience one is going to have a superior transfer rate when compared to ... lets say the C: drive that is a ... lets say Free for all Head Movement AND data transfer environment.