RAID uses a collection of disks of the same type to provide data protection, spreading data across the disks in such a way as to maximize the recoverability of the data if there is a single disk failure.
In effect, the RAID controller aggregates the disks and presents a single disk image to host operating systems so that applications never have to know where or how the data are being placed on the storage media.
A RAID subsystem is an example of aggregation combined with virtualization. It presents a virtual volume to the computer host that "looks" like one disk, when in fact the volume is made up of many disks. The virtualization allows for easy interoperability between the host and the storage subsystem.
From Aggregation and virtualization: Beyond RAID, Network World Storage Newsletter, 01/03/01.
RAID arrays write data to multiple disks in units called stripes. Striping places only a small portion of any datastream in one location, then jumps to the next physical drive. How fast data is stored and recalled is determined by the stripe size and the location of the parity.
Data can be written to this array in several ways. In RAID Level 0, the data is striped across drives without any parity bits for error checking. This does not provide any capability for recovery but does yield maximum transfer rates.
RAID Level 1 is disk mirroring. Data is written to multiple disks simultaneously. This provides complete redundancy, but the trade-off is the loss of disk space for the complete second copy.
RAID Level 3 stripes data one byte at a time across multiple drives, with parity stored on an extra drive. The speed is good; but for large files, a small data stripe size slows the transfer.
RAID Level 5 stripes sectors of data across the multiple drives with the parity interleaved. The flexibility of RAID 5 is the key to optimizing for a specific application.
Compare to just a bunch of disks.
Additional resources
