Term: RAID [redundant array of independent disks]

RAID is an acronym that originally stood for redundant array of inexpensive disks. With disks now generally considered inexpensive the acronym is now more commonly meant to refer to redundant array of independent disks.

The concept is relatively simple: use two or more disks (an “array”) to hold data to either improve resilience or performance or both.

There are several configurations, numbered 0 through 6. The most commonly encountered are:

0: Striping – Data is divided up across two (or more) drives without duplication. This is a speed-only enhancement, and actually increases risk as any single drive failure will cause the entire collection of data to be potentially lost.

1: Mirroring – Data is identically duplicated on multiple disks when it is written. Any single drive can fail and the remaining drive continues to hold the data. Write performance can be somewhat slower, but read can be faster as it is spread across the two drives.

5: Striping with parity – Data is distributed across two or more drives, as with RAID 0, to which an additional “parity” drive is added. The parity information is, in essence, a checksum on the data, and is generally rotated among all drives in the RAID array. The goal is that the loss of any single drive can be compensated for by using the parity information and the remaining drives’ data to calculate the missing data. Speed is also generally improved as reads are spread across the entire array.

More information: Standard RAID levels (Wikipedia)