Ga'ash Soffer van Voodoo Extreme heeft in dit artikel wat basic info neergeplant over de werking van harde schijven. Hieronder een stukkie:
When copying large files, throughput is the most important statistic. There is a lot of confusion about throughput of a drive. Here, throughput will refer to the internal transfer rate of the hard drive, the amount of data which the disk head can read into its cache per unit time.In math terms, throughput is (rpm * 60 * density * 512 ) / 1 MB, where density is usually replaced by sectors-per-track, rpm refers to the spindle speed in RPM (i.e. is it a 5400RPM, 7200RPM, or what have you speed drive) and 1 MB is whatever unit of size you want, i.e. if you want how many real MB/sec can be transferred, we use 1,048,576bytes. We'll use 1,000,000bytes to simplify calculations.
One problem with this is that there is usually more sectors on the outer tracks than the inner tracks, because they are longer. Drives which employ this technique are said to use Zoned Bit Recording, or ZBR, and encompasses almost all new drives. The problem with this is that throughput varies drastically depending on the track used to record the data on. The reason it is called zoned bit recording has to do with the way tracks are arranged. Since there are many tracks on a drive with little space in between them, the efficiency loss is marginal if instead of optimizing the number of sectors for each track, which is difficult, to group together a certain number of tracks and then, for each track in this grouping, or zone, use a set number of sectors per track. The number of zones may vary from drive to drive. The Maxtor DiamondMax Plus 40 has 16 zones.
It is also important to note that throughput does not take into consideration fragmentation and track switch times. That stuff is encompassed in seek time. Therefore, while throughput is, in my opinion, a more useful statistic than seek time, both are essential to accurately estimate performance on specific types of data.