Basic RAID
RAID (redundant array of inexpensive disks) is a series of disk configurations to get more reliability or performance (or both) out of multiple disks. RAID typically consists of two operations - striping and mirroring. Striping involves writing disk blocks in a pattern across multiple drives. This is very fast, because modern controllers can read and write to multiple disks simultaneously, so you essentially double your disk throughput. But you also double your liability, because a failure on either disk causes data to be lost. Mirroring takes care of that by writing exact copies of the same data to multiple disks. RAID-5 is a combination of the two, and is the most common, balancing space efficiency with speed.
Here are a few common configurations. All of these can be extended to larger configurations of more drives.
1 Comment
RSS feed for comments on this post.
Leave a comment
Sorry, the comment form is closed at this time.
It’s interesting to me that you didn’t go into a comparison of RAID-3, 4, 5, which is where I think the really interesting tradeoffs come up. What you said about RAID-5 basically applies to all three, with each one representing a different approach towards what should be fast (affecting all three of read, write, CPU demands). Basically, in a nutshell RAID-3 would be more appropriate for working on large files (e.g. multimedia) whereas RAID-5 would be more appropriate for transactional systems; databases, basically.
I think a comparison among these is particularly well suited for diagramming, because you can illustrate the process of reading and writing, and what different kinds of parity calculation and different units of size (disk blocks vs. bytes) do to CPU use, disk head movement, etc. Perhaps you could a subsquent feature on those.
Comment by James Wetterau — 2/25/2005 @ 2:23 pm