Menu

RAID is not a backup solution

March 16, 2011 - Programming

Recently I had a rather short discussion (more conversational then anything) about RAID. Yes you know who you are. And I am talking about that skype conversation that coincidentally occured during minecraft gameplay.

The topic? RAID and backups. At some point I remarked that “RAID is not a backup solution” or “isn’t a replacement for backups” to which the reply was “RAID-1 is”. I couldn’t remember the specific reasons why RAID wasn’t a backup solution but I have since regathered them.

The main reason RAID 1 (or any RAID level, mirroring, striping, or otherwise) isn’t a substitute for backup is because there are a lot of risks that it can’t protect against. If you accidentally delete a file, it will instantly be removed from both mirrored copies. Sure, you can recover it via the recycle bin; but what if you delete it- or it is deleted by- other software programs? Or you don’t notice? My point here is that a backup would have that file for you to recover if you found out you did need it either way; for RAID, the file will be gone. A backup is in effect a way to save the “state” of your work or filesystem at a given time; wether it’s a drive image or just a backup of the folders you predominantly work in, it stores the complete state of things, and you can go back to that state. It’s like a photograph. RAID, however- is a duplicate. The “mirror” drive is always the same as the first drive. RAID is designed to reduce the downtime of a server or other high-usage machine from the failure of one of it’s hard disks. It’s not meant as a replacement for data backup. Because it’s a duplicate and not a “saved state” so to speak, If your disk is corrupted by a software bug or virus, the corruption will be done to both mirrored copies simultaneously. If you’re hit by a bad enough power surge, it’ll probably fry both disks at the same time; If someone breaks into your house, they’ll steal the box that holds both disks;If your house gets flooded or burned, both disks will be ruined, etc. Those last few are often regarded as corner cases or “things that will never happen”. And while it’s true they are unlikely, there is no such them as overpreparedness. If you value your data at all you will have it in several places at once; Just do a thought experiment about how much you would lose if your main computer were to explode and everything on and in it was unrecoverable.

Basically, although the chance of data loss due to drive failure is reduced, data loss due to viruses or user errors, e.g. overwriting a file or deleting it, is not. If a file is deleted or overwritten the RAID array cannot be used to retrieve it. The file is gone. When a file is deleted it is deleted from all the drives. When a file is overwritten it is overwritten on all the drives. This is why a RAID array is not a backup solution. And don’t pull he old “well I’m always careful” routine. You aren’t. One day you might be sleepy or tired or otherwise trance-like and do something stupid. We all do it. Don’t try to make excuses. This doesn’t even begin to cover the fact that a lot of RAID solutions are implemented in software, and software contains bugs; and it means you now have to carry this extra driver around in memory, and you are reduced in the various hard drive utilities you can use. And, occasionally RAID issues will occur that will simply prevent you from rebuilding the mirrored set.

The Basic difference is that RAID is redundancy- a back-up is a backup. The difference between Redundancy and a backup is pretty easily illustrated; If you accidentally overwrite your PhD thesis with garbage, redundancy ensures that you have multiple copies of garbage, in case one gets bad. A backup ensures that you can restore your PhD thesis.

Sometimes people will say “well, RAID-5 is different! it provides mirroring in the form of block-level striping AND distributed parity! Well, alright then; consider this. In a RAID-5 array consisting of disks over 400Gb, if you lose a disk there’s something like a 75% chance of having an unrecoverable read error while the array is being rebuilt.

Have something to say about this post? Comment!

3 thoughts on “RAID is not a backup solution

Tux2

I strongly agree with this. I had a situation with a server that had 5 SCSI hard drives in a RAID 5 configuration. Somehow a driver got corrupted on the system and the computer would bluescreen no matter what we did. The computer was also not very friendly to all the recovery CDs and other stuff we stuck into it, including several different Linux live CDs (yes, we had the RAID drivers on a floppy). The data on the hard drives HAD to get recovered, and although the data was there, we just couldn’t get to it. In the end we finally found another server with enough SCSI slots that sensed the RAID on the disks and we were able to recover all the data we needed with an Ubuntu 10.10 live CD, but the whole process took 5 days (troubleshooting and recovery), where we could have been doing something else more productive if we just had a backup of the data.

Craig

Hey BC, I think I know who your talking about.
Though, from what I remember of that conversation, in the end we all agree’d that it was only a very basic level of backup to the extent of hard drive failure, which is what most of my customers have.

I would completely agree with this to any other extent.

BC_Programming

Yeah, I don’t quite recall how the conversation ended. There are a lot of people who believe otherwise and I felt it would make a nice topic for a entry 😀

Comments are closed.