0
I have a 4 disk (2.0TB each) FreeNas setup and I want to maximize the space available for storage. From what I've read that would involve using the stripe configuration, but that also increases the strain on the disks and provides no redundancy. I don't particularly care about redundancy but I don't want the disks to fail faster than necessary when I don't need the increased speed.
Is there a way to combine the disks into a single large volume but have each write to the volume be contained to a single disk rather than striped across them? If this were possible my thinking is it would reduce the strain on the disks. Additionally, in the case of drive failure I would only lose files on the failed disk and could simply replace it; if it were not recoverable.
With my setup, the apparent best option that is not striped is RAID-Z1 with 3 disks and the 4th being a spare, but this severely reduces the available space in order to provide redundancy, which is not something I'm too worried about for the files I plan to store there.
Alex
Posted 2014-01-06T22:19:50.117
Reputation: 101
I don't believe that zfs provides a concatenated pool like what you're describing. Though even if it did, I don't think you can count on restoring any files if a single drive failed. I wouldn't worry about the extra load on the disks when using striping - unless you have unusual I/O activity, the drive will likely die of old age than extra seek time from stiped I/O. – Johnny – 2014-01-07T00:14:10.100
In that case, can you point me to any additional information on ZFS concatenated pools and/or why using ZFS in that way wouldn't allow recovery on any non-damaged disks in the event of failure? I've been searching for information on this topic without much luck. – Alex – 2014-01-07T02:09:21.023
Since Zfs doesn't support simple concatenated volumes, I can't point you to any zfs docs that describe it. Check your zpool man page for supported RAID levels (mirror, raidz, raid1 (same as raidz), raid2, raid3). Likewise, I don't have any docs that support zfs filesystem recovery for a failure mode of a raid type it doesn't support.
– Johnny – 2014-01-07T03:31:52.480I was thinking more in theoretical terms for concatenated arrays, since you said that, even if zfs did support concatenation, you didn't think I could count on restoring files from a concatenated array if a single drive failed. I'm just wondering why that would be since on the face of it I don't see a reason why that wouldn't be possible. – Alex – 2014-01-07T05:38:04.407