Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I did a bit of research already and didn't come across any filesystems so far that have the following features:

Expandability, Redundancy, and the ability to utilize the full capacity of different disk drive sizes.

Basically, I'd like to get a large chassis that will support 20+ 3.5" drives and just slap what ever I come across in it. The drives could vary from 1TB to 4TB+. I would like to be able to just have 1 volume/partition if possible as well. It's just to store misc. media/scratch drives/temp. Nothing mission critical.

Any idea how I could accomplish this?

share|improve this question
7  
No. Don't do it. slap whatever I come across is a recipe for disaster. –  ewwhite Dec 11 '13 at 20:52
1  
Have a look at Ceph. Adding OSD's (one per disk) expands the filesystem automatically. Should perform well enough for media storage. For scratch/temp drives you want to use a local ssd for that. –  Matt Dec 11 '13 at 21:09
1  
As it is for scratch/temp non mission critical data, why asking for redundancy ? and how do you concile "redundancy" with "utilize the full capacity of the drives" ? You can't have both. –  jlliagre Dec 12 '13 at 21:03
    
Get a Drobo. –  ewwhite Dec 12 '13 at 21:16
add comment

2 Answers 2

zfs sounds like just the thing!

Expect to lose all of your data at any time if a drive fails.


Oh... I missed 'redundancy'.

Put all of your disks into a single LVM VG and then create logical volumes in that pool with LV mirroring enabled. And then test.


But really, @ewwhite is right - take the time and do it right.

share|improve this answer
1  
ZFS people don't recommend that mess. –  ewwhite Dec 11 '13 at 21:43
    
It's a use case mentioned on the ZFS wiki (well… the Ubuntu wiki, so take with a MASSIVE hunk-o-salt) and will work. For a scenario where you're not mirroring/raidzing and just want to use it as a may-disappear-anytime-scratch-space, what's wrong with it? –  MikeyB Dec 11 '13 at 21:50
    
Because it's lazy. It's not difficult to find disks of equal size and plan/commit to doing so. The reason it may be difficult to find something with all of the OP's desired attributes may be because it doesn't exist or doesn't make sense for professionals to do. Momentum matters in IT, and if you find yourself on the fringes of a solution, it makes sense to evaluate whether you're going down the wrong path. –  ewwhite Dec 11 '13 at 21:59
add comment

Redundancy does not work the way you would like it to!

It depends on having the same amount of diskspace on all the drives.....so be prepared to only be able to have (size of the smallest drive)*((number of harddisks)-1) in the best case (RAID5)...but if you have big sized drives and a weak box (considered IO and CPU) you should look for RAID6...chances are that you have a faulty drive while rebuilding!

If the data is valuable, consider doing it the right way (RAID5/6 with matching drives).

If the data is not valuale at all (movies,mp3,pr0n et.at.) make a big JBOD array! (But you said you want to have redundancy...so no doable way at all!)

And NO.....ZFS would not be able to do this!

share|improve this answer
    
There is more than one way to get redundancy. One is mirroring - in which case you need the same size drives. (or do crazy things). Another is mirroring at the extent level using LVM mirroring. –  MikeyB Dec 16 '13 at 16:28
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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