Spanning hard drives - data loss?

Soldato
Joined
25 Dec 2008
Posts
5,976
Location
Sheffield/Norwich
I like to keep my media together for organisational reasons, but I have too much in each category (e.g. my ripped films/tv shows) to fit on a single HDD, so having recently discovered the facility to span different HDDs, I've created myself a couple of 1.8tb partitions across three drives (2x(1.5tb + half of 1tb) each). It's not really a problem as I have a backup of each, but I'm interested to know if losing (for example) the 1tb drive to failure would cause both partitions to fail completely or if I would be able to readily recover the data from each of the 1.5tb drives.
As far as I know the data isn't striped, but I really have very little idea about how drive spanning works - I would assume it fills up the first HDD and moves onto the second, probably fragmenting a file that 'goes over the edge' of the first HDD, as Windows views it as a single partition. But can anyone verify this? And so does spanning drives increase the chance of data loss, in the way RAID 0 does?
 
I think it depends what method has been used. I doubt its as easy as that, and im just guessing, but I would think that it would kill all the files across the bit thats spanned. Purely because the OS never sequentially places the files.
 
I think what you've done is stick the drives together, the phrase used being concatenate. Similar idea to raid but with none of the advantages. What program did you use to do this?

I'm pretty certain if one drive goes down, they're all down. This is fine, but as it's as unstable as raid 0 and slower, raid 0 is probably a better call.

I'm pretty sure you don't want to do this if it's purely for convenience, as the inconvenience of having the array go down will be considerable. I'd suggest raid, as this will leave you with a single volume which is either more robust, faster, or both than the current solution.
 
I think it depends what method has been used. I doubt its as easy as that, and im just guessing, but I would think that it would kill all the files across the bit thats spanned. Purely because the OS never sequentially places the files.

Surely the OS would write the files to the (fresh, blank) disk sequentially in the initial write? Since this is basically just storage, only infrequently being updated, they would all stay in the same place, just being added to as I buy more films. So would the files really be fragmented across the hard drives, in that scenario, do you think?

I think what you've done is stick the drives together, the phrase used being concatenate. Similar idea to raid but with none of the advantages. What program did you use to do this?

I'm pretty certain if one drive goes down, they're all down. This is fine, but as it's as unstable as raid 0 and slower, raid 0 is probably a better call.

I'm pretty sure you don't want to do this if it's purely for convenience, as the inconvenience of having the array go down will be considerable. I'd suggest raid, as this will leave you with a single volume which is either more robust, faster, or both than the current solution.

I just used the Windows Disk Management Console (7, but I expect it's the same on Vista). It creates a dynamic disk, which can then span two or more drives. The problem I have with RAID 0 is the two hard drives have to be the same size otherwise it ignores any extra space on the larger hard drive, and I don't have enough space to afford that. (Plus, remember this is all backed up, so if I were to designate the concatenation RAID 9, what I essentially have is RAID 9+1).
I accept that the MBR will be corrupted (or whatever happens) if one of the drives fails, so I won't be able to access it through My Computer. However, if the data is not striped, surely the actual files on the unaffected HDD will be intact, although unreadable. If this were the case, I would be able to recover those files using my data recovery software, would I not? I suppose what I am asking is would the files themselves, on the unaffected disks, be corrupted?
 
I just used the Windows Disk Management Console (7, but I expect it's the same on Vista). It creates a dynamic disk, which can then span two or more drives. The problem I have with RAID 0 is the two hard drives have to be the same size otherwise it ignores any extra space on the larger hard drive, and I don't have enough space to afford that. (Plus, remember this is all backed up, so if I were to designate the concatenation RAID 9, what I essentially have is RAID 9+1).
I accept that the MBR will be corrupted (or whatever happens) if one of the drives fails, so I won't be able to access it through My Computer. However, if the data is not striped, surely the actual files on the unaffected HDD will be intact, although unreadable. If this were the case, I would be able to recover those files using my data recovery software, would I not? I suppose what I am asking is would the files themselves, on the unaffected disks, be corrupted?

Firstly RAID9? RAID10 is usually short hand for RAID 1+0. RAID x - the number has nothing todo with the number of disks but is an identifier for the architecture of the layout.

When dealing with large volumes, loosing your MBR means often recovering is like looking for a needle in a haystack. Better to have two or more manageable sub-volumes.

RAID is done at block level. The newer JBOD MS server based domestic NAS boxes use file based replication rather than block level instead.
 
Firstly RAID9? RAID10 is usually short hand for RAID 1+0. RAID x - the number has nothing todo with the number of disks but is an identifier for the architecture of the layout.

When dealing with large volumes, loosing your MBR means often recovering is like looking for a needle in a haystack. Better to have two or more manageable sub-volumes.

RAID is done at block level. The newer JBOD MS server based domestic NAS boxes use file based replication rather than block level instead.

I'm not actually RAIDing. I just hypothetically described what I am doing (using the Windows Management Console to create a partition that spans across two drives) as RAID 9 (which obviously doesn't exist) so I could use RAID terminology to explain that I have a backup of the spanned partition.

I see what you're saying though. What I might do is pull one of the hard drives and see what I can recover from the hard drives left before deciding whether to scrap the idea and go back to separating the files on different hard drives/partitions.
 
OK, I tried unplugging a drive to see what happened, and (as expected) I was no longer able to access the spanned partition. Nor could I recover any data from the remaining hard drive - apparently, my recovery software can't cope without a drive letter!
Backups for the win. :) (why is that tla starred out??)
Thanks for the input.
 
I had this happen for real to me this week.

I had 2 1TB in a dynamic volume spanned, I added a 250GB that I had lying around, all worked fine. Then we had a powercut. The volume apeared offline in windows, and couldn't be brought back by any of the common fixes.

I tried the main linux based free recovery software without any joy they seemed to fail because they were lookikng for a 2GB drive which didn't exist.

Unfortunatly the drive with my backup on had just been RMA'd so after trying the trial I stumped for a copy of GetDataBack which worked perfectly, and was able to recover the files and folders.

I've learnt my lesson and I'm going to move to a system with redundancy like WHS.
 
Back
Top Bottom