RAID 1 backups.. better to get different drives?

[Tom B]

I'm planning on using two 1tb SSDs for backup purposes in RAID 1. I want to use SSDs because I want a silent solution as backups will be taken automatically overnight while I'm trying to sleep.

My question is, am I better using different brands? The OCD in me says I should get two identical drives, but the pragmatist in me asks: "Are two identical drives with identical data being written to them more likely to fail at the same time?" as the purpose here is backup, I wonder whether it would be better to get two different brands/models as they'd be less likely to fail at the same time?

Any performance overhead caused by mismatched drives is irrelevant here as they're just for backup and reliability is the important factor here.

 

[neil_g]

RAID is not a backup method.

deletion, corruption and some types of controller failures will all destroy your data.

 

[Tom B]

Isn't the entire point of RAID 1 to have duplicated data?

edit: I'll be using git or rsnapshot to manage versioning of the data so deletion isn't really the problem, I just want the data written to both drives in case one fails.

 

[Armageus]

Isn't the entire point of RAID 1 to have duplicated data?

Yes, but duplicated data isn't the same thing as a backup

Corruption may get duplicated between drives, viruses may get duplicated between drives, and deleted/overwritten files get duplicated between drives.

RAID generally provides high availability - i.e. if a drive (or multiple drives in different raid levels) fails, you carry on as if nothing has happened.

If you accidentally delete a file, the RAID set immediately mirrors that change.


A second drive can be a backup, but in that case you don't want it it RAID, rather use some backup software (or even something as simple as a batch file and Robocopy), to copy files at a scheduled interval (e.g. could be automated everytime you shut down, or once a week etc.)

A second drive in the same PC, also doesn't protect against e.g. electrical fault or fire, which could damage both - removable drives, a NAS or cloud backups, are options you should be considering.

If your data is important, then you need to have several different backups, and of several different ages (e.g. daily, and maybe monthly).

 

[Rroff]

Mirrored RAID's main benefit is in convenience in that you can replace a failing member disc and be back up and running with minimal disruption (or even no disruption) - I've seen entire arrays fail about as commonly (neither is that common) as a single drive failure.

It shouldn't be relied on as a sole backup mechanism.

 

[Tom B]

Yes, but duplicated data isn't the same thing as a backup

Corruption may get duplicated between drives, viruses may get duplicated between drives, and deleted/overwritten files get duplicated between drives.

Perhaps I should have explained what I'm doing. I want to back up my desktop to another computer in my flat and I want to protect the backup server from drive failure. My /home directory is tracked via git and i'm going to push all changes to the file server nightly in a cronjob. Perhaps you're right, I'll manually push changes to both drives. (I may change this to use rsnapshot as it's probably a bit more suitable than git)

 

[Armageus]

Perhaps I should have explained what I'm doing. I want to back up my desktop to another computer in my flat and I want to protect the backup server from drive failure.

Having explained it - this sounds fine, having a RAID mirror in your backup server increases the availability of the backup server, and reduces the chance of data loss from the backup server. Your server is the "backup" of your data.

However the other option with 2 drives in the backup server, is to do something like backup to 1 of the drives daily, and then backup to the other drive weekly (giving you an option to restore older files if not noticed immediately)

(or to alternate between drives daily - giving you e.g. today on 1 drive and yesterday on the other)



Possibilities are endless, all depends on how critical your data is, and how quickly you may notice any issues (e.g. if you only take daily backups, and you don't notice the same day that something has been deleted then you have no way of recovering)