Most of the reliability features and issues discussed in this part of the site relate
to making drives themselves more reliable. However, there is only so much you can do to
improve the reliability of a single drive without the cost becoming exorbitant.
Furthermore, since most people aren't willing to pay for ultra-reliable drives,
manufacturers have little incentive to develop them. For those applications where
reliability is paramount, the quality of no single-drive solution is sufficient.
For these situations, many businesses and power users are increasingly turning to the use
of multiple drives in a redundant or partially-redundant array configuration. The common
term that refers to this technology is Redundant Arrays of Inexpensive (or Independent)
Disks, abbreviated RAID.
The principle behind RAID is "belt and suspenders": if you store redundant
information across multiple disks, then you insulate yourself from disaster in the event
that one of the disks fails. If done properly, you also improve performance--sometimes in
a substantial way--by allowing the drives to be accessed in parallel. And you can make it
so bad drives can be replaced without the system even being taken down.
RAID is a big topic unto itself; there are many different ways that RAID can be
implemented; various hardware and software considerations; and many tradeoffs to be
considered when implementing a system. I have therefore created a separate area that discusses RAID in detail. Check it out
if the subject interests you. RAID is rapidly increasing in popularity, and I believe it
will only be a few years before it starts showing up even in high-end home systems.
Next: Hard
Disk Warranty and Disaster Recovery Issues
