I actually run with multiple hard drives in my primary computer, and deliberately chose 2 TB 7,200 drives rather than 3 TB ones.
I did this because I would rather split my sample libraries over 2-3 drives than have them all on one drive. This is because splitting them to separate devices increases the parallel streaming capability which will potentially greatly improve load times for samples when the samples are from different instruments/libraries.
It IS still dependent on which libraries are most frequently accessed, using the above approach, but even though I still end up pulling a lot of samples from one drive or the other, I DO see a general performance improvement overall, from splitting across multiple drives.
I have nearly 3 TB of samples, and currently have them split across 3 different 2 TB 7,200 drives.
Bob Bone