Over the last 20 years, enterprise disk drives have shown a predictable growth when it comes to both capacity and throughput.
When I started out in this business, many IT shops had just upgraded to 360M-byte disk packs. As I recall, these disk packs contained about a dozen platters for the data, were about 18 inches high, were larger in diameter than a 33-rpm record and weighed about 30 pounds.
When you walked through an IT room you saw row on row of the washing machine-sized drives that the disks worked in - and often, sitting on top of the drives, you saw what appeared to be an assemblage of large cake containers. These were the covers for the disk packs. Those of you who go back as far as this may also remember that once a week, some time during the third shift, contractors came in with special vacuums to vacuum out the disk packs, getting rid of the dust that accumulated.
The disk pack vacuuming business fell on hard times with the invention of Winchester drive technology. By the time the first of these had arrived (160M-byte, rack-mounted devices - and yes, one device did take up the whole width of a 19-inch rack), it was clear that the disk pack vacuuming business was not going to be a good place for future investment.
Capacity has grown, and devices have gotten quite a bit smaller over the years.
Capacity growth has mostly been achieved due to significant improvements in both the media (the actual spinning disk within the drive) and the drive heads (the component inside the drive that writes and reads data to and from the disk).
Advances in media technologies have allowed for increased areal density - the number of bits of data that can be stored on a square inch of the media. As aerial density has increased over the years, storage capacities have grown, while the footprint of disks has gotten smaller and smaller.
Improvements in the media would have been of no value, however, if head technologies hadn't improved at the same rate; as the bits are packed closer together on the media a finer granularity is needed to ensure accurate read and write operations. New head technologies became available, of course, and head development continues to support the trend toward more data on less physical disk.
Whatever growth was enabled by improvements to the heads and media, however, would have been of limited value if we didn't at the same time come up with faster ways to move data between the disks and the applications that use it. In the next several columns, we will look at what has been done to improve data throughput, paying particular attention to the evolutionary trend toward serial I/O interfaces. It turns out that these not only offer increases in speed, but also quite a bit else.
Mike Karp is senior analyst with Enterprise Management Associates, focusing on storage, storage management and the methodology that brings these issues into the marketplace. He has spent more than 20 years in storage, systems management and telecommunications.