It’s a fact that all hard drives will eventually fail. Traditional hard drives have moving parts, and will wear down over time. Even solid-state hard drives are not indestructible. To mitigate the issue of failing drives, any critical system needs fault tolerance, namely RAID (Redundant Array of Independent Disks).
RAID is implemented for both performance and fault tolerance. There are several different drive configuration options available. Here are some of the most common:
|RAID 0 (striping)||Performance only; NO FAULT TOLERANCE|
|RAID 1 (mirroring)||Good read/write performance; can sustain a single drive failure; good for OS and transaction databases|
|RAID 5 (striping with parity)||High read performance; can sustain a single drive failure; good for large data storage, web services, archiving|
|RAID 6 (striping with dual parity)||High read performance, can sustain two drive failures; used for backup-to-disk, high availability, data archives|
|RAID 10, 0 + 1, 1 + 0 (striped RAID1 sets)||High read/write performance; can sustain one failure in each subarray; used for fast databases, virtual datastores|