RAID

A server stores important data on one physical disk. The disk fails. Everything on it becomes unavailable.

RAID stands for Redundant Array of Independent Disks. It combines disks for performance, fault tolerance, or both. Common levels balance usable capacity against disk-failure protection: RAID 0: spreads data across at least two disks for parallel access and full combined capacity; one disk failure loses the array. RAID 1: keeps identical copies, typically on two disks; one can fail, with half the total capacity usable. RAID 5: uses at least three disks and distributed parity—recovery information that reconstructs missing data. It survives one disk failure and uses one disk's worth of capacity for parity. RAID 6: uses at least four disks and two sets of parity. It survives any two disk failures and uses two disks' worth of capacity for parity. RAID 10: spreads data across mirrored pairs, typically at least four disks. Half the capacity is usable; one disk per pair can fail while data remains available. These capacity examples assume equal-sized disks; parity calculations add write overhead.

Data
 ↓
Disk A
Disk B

Depending on the RAID level, data can survive certain disk failures.