I'm entertaining creating a largish (3.5TB+) backup server for my family.

I already own enough hardware, and I don't want to buy any more right now. The config I envision look like this:

  • Two 1TB drives
  • Two 750GB drives
  • RAID 5?
    • Distributed parity
    • Can tolerate one HDD failure
  • BTRFS or LVM + thin provisioning
    • BTRFS heavily preferred
  • Can add more drives in the future

Is this possible? I know BTRFS can fuse a bunch of drives together into one filesystem without parity, but I'd like to sustain at least one drive failure (I don't trust anything with moving parts).

Worst case I can just make 750GB partitions on the 1TB drives, but it's not trivial to make them exactly the same, and besides it's wasteful.

  • If you want to know if it's possible, then try it. – Ƭᴇcʜιᴇ007 Sep 30 '14 at 18:12
  • Trying it would require a lot of time (I'd have to build the NAS from components lying around, setup the drives and software and test). I'd rather know now if this is a waste of time. – PythonNut Sep 30 '14 at 20:38

After experiencing ZFS, I would not recommend anything else. It is SOOOO easy and reliable.

My recommendation is, create ZFS pool with two mirrors in the pool. After you have pooled them, you can slice and dice it any way you like.

One draw-down of ZFS is that you cannot downsize. You can add more devices, mirrored or stand-alone, at any time.

P.S. RAID5 requires 3 drives. You have pair of each, so RAID5 is not ideal. And it is slow. Mirror is much simpler to manage and offers read speed improvement because of load potential sharing.

  • But then I only get 1.75TB of space, VS 4x0.75TB = 3TB with a RAID. I don't care about performance very much. Time I have (waiting for backups in the background), money I do not. (Figuratively speaking) – PythonNut Sep 30 '14 at 20:40
  • Actually, I take that back, you're right on the space issue. – PythonNut Dec 4 '14 at 5:41
  • btrfs raid5/6 isn't yet mature. According to their wiki it is still vulnerable to write-hole problems. Besides, they're introducing a new library for higher numbers of parity disks. So unless you go the mirror way I'd abstain from btrfs. – Álex Aug 21 '15 at 15:49
  • Incidentally, that's the reason I still run my raids with zfs. If only zfs allowed reconfigurations on the fly like btrfs, there would be no reason not to use zfs without second thoughts until btrfs raid matures. I'm hoping that when I finally run out of space in my zfs pools btrfs raid will be there to replace it. – Álex Aug 21 '15 at 15:51
  • 1
    Here is the btrfs status on raid wiki: btrfs.wiki.kernel.org/index.php/RAID56 – Álex Aug 21 '15 at 15:52

From this SO answer:

Btrfs distributes the data (and its RAID 1 copies) block-wise, thus deals very well with hard disks of different size. You will receive the sum of all hard disks, divided by two – and do not need to think on how to put them together in similar sized pairs.

So btrfs will deal with heterogeneous disk sizes in a reasonable manner.

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.