Data backup; raid or multiple copies?

xoiio

Honorable
Nov 18, 2012
119
0
10,690
Hello all, so I have an hp proliant dl380 g6 coming to replace by g4 as a network storage server and misc. server (things like ts, uploads/downloads that can take a while, video processing, etc).

This will be a nice upgrade since I will have 8 sas/sata slots available (which i can upgrade to 16 with a cage and sas card) that I can ad storage space to as I need it. It likely will be slowly over time, so using a raid config would require even more storage to back up what I have, create the array, and then copy it back.

However, on the subject of data backup, and making sure you don't lose your data, in my case particularly, I was wondering what would be a better option, in general. Either a raid config, or multiple copies on multiple drives.

Say you have a raid array of 6 1tb drives in raid 6, this would give you one singe drive with a 4tb capacity, and allow for two drive failures, and if a drive failed, the main array would continue functioning, any programs using data in this location would still be able to access it, then you just pull out the drive, get a suitable replacement, and then pop it in to rebuild the array.

However, for things like home use that don't need access to data all the time, I'm curious if having each 1tb drive set as an individual, no raid configuration, would perhaps be better.

This would give you 6 terabytes of storage space total, and, if you backed up a computer with a 500gb drive, you could take this backup image and copy it across all the drives, or two or three, allowing for two drive failures, before you were at risk of losing your data.

This would also mean it is much easier to upgrade storage as you go on, replacing a 1tb drive with 2tb, 3tb, etc. (though I don't even think there are 3tb 2.5 inch drives yet).

Now, if you had one drive as the primary one, you could set a program up to copy the data over to other drives, but if the primary drive failed, programs that were using it actively would not be able to access the files, so you would need to change the file path settings, but then it would be back to normal use.

So, it would seem that a raid configuration would be ideal in an IT environment where uptime was a priority, but individual drives and manual copying (or semi-automated copying, with checksum verification) would possibly be the better solution for the home user, although in my case that would end up with 8, and later possibly 16 network drives to go through, which would be a tad annoying, but offer more storage space.

So, I'm wondering, given the choice, if you had 8 2tb hard drives, would you set them up individually to have 16tb of storage, or would you use a raid 6 configuration and have 12tb of storage, a considerable loss, but essentially constant uptime.

Another factor to consider is emergencies, with individual hot swap drives, though less than ideal, in a rush you could just pull them all out if you really needed to get out quickly, but I think that such an action would corrupt a raid array, unless you shut down the computer entirely and then pulled them out, (not sure if maintaining their order is critical, but I know I make sure to do that when dealing with raided drives).
 
Solution
RAID can protect against physical disk failure. It can't protect against deletion, file system corruption, software bugs, malware, ... the list goes on. Backups are the best answer. Having multiple copies of the same data on multiple devices. Tape gets a bad reputation, but for enterprise backup it is still cheap and easy with a small tape library.

kanewolf

Titan
Moderator
RAID can protect against physical disk failure. It can't protect against deletion, file system corruption, software bugs, malware, ... the list goes on. Backups are the best answer. Having multiple copies of the same data on multiple devices. Tape gets a bad reputation, but for enterprise backup it is still cheap and easy with a small tape library.
 
Solution
As above ^, RAID is not "backups", it's quick recovery from a failed drive. If I could only do one thing, I'd have a single disk, with a second single disk that I backup my files to often. I'd much rather do that than have a RAID system where both disks are clones of each-other or even in a larger RAID 5 array or something. That's what I do at home, have a main computer with a shared single hard drive, and an external drive I make backups on, along with a second internal drive that has a 3rd set of backups of stuff I can't easily replaces in case the backup drive fails.