I recently saw an assertion in a technical forum again that RAID is backup. I thought the subject had been thoroughly covered in the past, but it seems that it still needs some attention.
RAID, except for RAID 0, provides redundancy which is different from backup. RAID, except for RAID 0 (just assume the rest of the document makes the RAID 0 exception when talking about RAID), protects against disk failure, but there are several reasons why even this is not to be relied upon 100% for disk failure. RAID is awesome in the following scenario:
- A single disk in an array fails, independently of others, and no other disk fails in the time it takes to rebuild and replace (or replace and rebuild) the disk.
That’s it. That’s the one thing RAID is good at for data protection.(*)
Backup attempts to mitigate against data loss including these additional scenarios, among others, many of which overlap:
- Data corruption. Including encryption by cryptolocker or similar.
- Data deletion. That oops scenario.
- Data center loss (disaster). Acts of God, terrorism, freak electrical accidents.
- Disk or storage system failure, including multiple disk failures or controller failures.
RAID is an idea that came after the idea of backup. It was designed to eliminate the need for a slower restore from backup process in order to recover from certain specific failure scenarios as determined by the RAID protection level selected. When designing a policy for data protection, the backup itself should be the first and foremost concern prior to worrying about RAID.
On a similar note, replication without longer-term recoverable snapshots (with sufficient retention) on the target media is also not backup. Replication was designed, among other things, to provide a faster recovery than a restore from backup for an additional set of scenarios, including multiple disk or controller failure in the primary RAID.
If you or your IT department got these technologies out of order and are not backing up today, we can help you find a solution that meets your needs.
(*) The technically minded reader will note that there are more RAID configurations that have not been addressed here. RAID6 which will tolerate 2 simultaneous disk failures. RAID1+0 and 0+1 which will tolerate certain failures of more than one disk, so long as they are the not the wrong second disk that fails. There are also other advanced nested RAID levels and software RAID which can tolerate additional numbers of simultaneous disk failures. The point is that none of these protect against data loss in all cases of corruption, deletion, disaster, etc.
