RAID

RAID is a family of storage architectures that combine multiple drives to improve performance, increase usable capacity, or tolerate drive failures.

RAID 0: A RAID level that stripes data across drives without redundancy. Its parallel striping can improve throughput, but one failed drive can destroy the array.
Block-level storage: Data storage that exposes fixed-size addressable blocks rather than files or objects. RAID commonly combines drives below the filesystem at the block-storage layer.
JBOD: A storage arrangement that presents multiple drives individually or concatenates their capacity without RAID-style redundancy. JBOD combines drives without RAID's defined striping, mirroring, or parity schemes.
Mean time to failure: A statistical estimate of the average operating time before a device or system fails. Drive failure rates help estimate array reliability, though they do not predict an individual failure.
RAID 1: A RAID level that mirrors data across drives. It trades usable capacity for a complete duplicate that can survive a drive failure.
Data striping: A method of dividing data into chunks and distributing them across multiple storage devices. Striping lets drives handle different portions of a workload in parallel.
Erasure coding: A method that splits data into fragments and encodes redundant fragments for distributed recovery. It offers redundancy across storage nodes using a different scheme than drive-level RAID.
Rebuild time: The time required to restore redundancy after a failed storage device is replaced. Long rebuilds leave a degraded array exposed to further failures.
RAID 5: A RAID level that stripes data and distributed parity across at least three drives. Its parity lets an array reconstruct data after one drive fails.
Data mirroring: A storage method that keeps identical copies of data on separate devices. Mirroring provides RAID redundancy by preserving duplicate data.

Linked from 5 pages