Lesson number one: a RAID is not a backup solution. RAID, which stands for redundant array of independent disks, is designed for improved performance and fault tolerance. In spite of the fact that you are using multiple disks for a single volume, you should never, ever consider a RAID a replacement for a backup system. “Fault tolerance” simply means that the RAID will continue working if one of the disks in the array fails. This is a benefit if you are running a server that can’t go offline while you swap out a failed hard drive. But if the whole RAID goes down and you didn’t have the data backed up, it will be an unmitigated catastrophe.

 One other way a RAID is not like a backup system is that it does not allow you to revert to a previous state. RAIDs don’t take snapshots of data. They work in real time and won’t provide any contingency if there is a virus attack or accidental deletion.

 In many ways, a failed RAID is worse than a failed independent disk in the same configuration. That’s because a RAID (except for RAID 1) splits your data across several different drives. It’s like the opposite of putting all your eggs in one basket. The RAID controller and file system keep track of which bits and bytes go where and how to piece them back together. Absent these, it’s like having a bunch of spare parts and no instructions for assembling them.

 All that being said, there are ways to recover data from a failed RAID. If it’s a RAID 1, you should be able to pull out one of the functional independent disks and perform a normal data recovery on it, since RAID 1 volumes simply mirror the data across all the disks.

 For a RAID 0, RAID 5 or RAID 6, you may be able to rebuild the RAID virtually. There’s a program I use called R-Studio that has a built-in RAID reconstruction module. Basically, what you do is take out all the independent disks and image them as-is. If the failed disk can’t be imaged, you can replace it with a Missing Disk or Empty Space object in R-Studio.

 Next, you rebuild the RAID from the disk images in R-Studio. Here, it’s critical that you replicate the RAID configuration exactly. You must choose the correct number of disks, offset, block size, and block order (stripe).