How to protect your data
Revision as of 22:55, 10 July 2019
There are many things out there that will eat your data. What cannot be stressed enough is that RAID is not a panacea. There is no point in having multiple copies of your data in your computer if your computer is then stolen or destroyed. The first thing you need us a threat analysis.
Firstly, backup, backup, BACK UP. And keep it off site. You can use raid for this if you wish, mirroring over the internet.
Then you need to make sure you have integrity. The problem is that drives lie. New file systems like btrfs have integrity management built in, but if you haven't enabled mirroring or some such, your data is lost.
Raid is designed to protect against a disk failure, and if the data is lost at a level below the raid array, it will recover it for you, but what it needs is for the level below to provide integrity and not corrupted data.
The new dm-integrity target (mid 2019) is designed to ensure that should the data be corrupt at the dm level, then this triggers a read error and lets the layer above know that something is wrong. This means that raid can then correct the problem. Otherwise you need to use an integrity-aware filesystem above it.