Originally posted by DJ-CHRIS
When you have 3 of the drives, you are most likely running them in Raid 5 which is basicly a combination of Raid 1 and 0. Something like that, I dont know much about raid
No man. Raid 5 is a configuration where you have a "parity disk". This is less secure than Raid 1 but has room for more capacity. For example, in Raid 1, if you use 5x120gb disks, ur total capcity will be 120Gb since all disks will contain the redundant information. In Raid 5, 5x120Gb will give you 480Gb (120x4). The last drive is a parity disk. If one disk fails, you can retrieve it by XORing the other 3 disks and the parity disk. If 2 disks fail simultaneously, with Raid 1 you are good. With Raid 5 u are screwed!.. At least it seems that way to me.
Parity disk (disk N) = disk1 XOR disk2 XOR ... XOR .. disk N-1
So every write would entail you to write to the parity disk. 1 read of the current contents of disk X, compare the diffs, and 1 write to the parity disk, and of course 1 write to the disk X itself. At least, that's the best algorithm I can think of. If anyone can do better, post away!
As for the original question, it's difficult to say without benchmarks. Using 1) 3 drives simultaneously vs 2) using 2 faster drives. Certainly, 3 is better than 2, but the 2 are faster...