RAID - Redundant Arrays of Inexpensive Disks

I was talking with my brother Andy this morning about the wisdom of buying aBlue-Ray player, which sent my mind off on a tangent about how fast you could stream data from a hard disk drive. Hard drives can read and write data pretty quickly once the data has been located on the disk. However, locating the track where the data is stored can be a very slow process, relatively speaking. I just looked up a Western Digital disk that has a average seek time of 7 milliseconds, which is an eternity in a computer. However, track-to-track seek time is only 0.7 milliseconds. Still slow, but considerably faster. However, if you had a bunch of disks all lined up and you could switch between them, you could read (or write) from them successively. While you are reading a track from one, which should only take 10 microseconds for a 10,000 RPM disk, all the rest of the disks would be in the process of moving their read/write heads to the next track (seeking). Dividing the 0.7 millisecond seek time by 10 microseconds gives you 70, which means you would need 70 disks to be able to maintain the maximum data transfer rate for this particular model of disk. 70 disks times 500 gigabytes is 35 terabytes. Given the posted data transfer rate of 3 GB/second, it would take 12,000 seconds, or 200 minutes, or 3.3 hours to transfer all of the data that these disks could hold. That's ten terabytes per hour.

What you would do with all that data is another issue.