admin:raid

Redundant Arrays of Independent Disks (RAID)

Contrary to some beliefs, especially by CEOs, a RAID is not a backup. Even though per definition some raids have mirrored copies of your data and can recover from one or multiple drives failing, this is just to ensure the continuous operation of services.

A backup however is having multiple copies of files on multiple data storages which are not in the same system. A backup which you store on the same disk or RAID as the "original" data is useless, since it doesn't protect against natural hazards (flood, fire), theft or ransomware.

RAID level description space efficiency minimum amount of disks recommended amount of disks possible disk failures
0 striping 1 2 <5 0
1 mirroring 1/n 2 n-1
5 block-level striping with parity block 1-1/n 3 3+n*2 1
6 block-level striping with two parity blocks 1-2/n 4 4+n*2 2
RAID level description space efficiency minimum amount of disks recommended amount of disks possible disk failures
01 mirror of stripes 1 4 4*n 1 complete stripe
10 stripe of mirrors stripes/n 4 2*n m-1 per span

Usually when there are multiple disks mirrored (e.g. with RAID-1, RAID-5, RAID-6), only the capacity of the smallest disk is usable. Any excess capacity which the other disks have can not be used inside of that RAID. It is advisable to use partitions instead of whole disks as RAID members as you can still use the excess capacity this way.

Oftentimes, it's very expensive to set up a RAID. Sometimes, you can live with only one or two copies of data when you can restore from a backup. The solution to reduce costs from RAIDs is either using a RAID level where the storage is used more efficiently (e.g. RAID-5 or RAID-10 instead of RAID-1) or using a storage pooling technology like Greyhole.

  • Last modified: 2024-07-05 14:31