JBOD and data integrity

[Decius]

I have a 250GB and a 500GB SATA drive in one of my systems, and would like them in one big chunk of storage space. Naturally this would lead me to JBOD. One thing I'm curious about with that spec is data integrity.

If one of the drives fail does it affect data across the entire JBOD array, or just the data on the dead drive? Would an overly large file sitting at the end of one disk be spanned to the second, and if so would that file be lost?

Thanks :D

[LstOfTheBrunnenG]

Depends on what you use.

If you use a RAID controller, fake or real, it depends on the controller and possibly the filesystem.

If you use Windows software RAID, I think it pretty much says flat out the data is gone.

Linux LVM, I just had a friend lose a drive. The data on the good drive was unaffected.

[Decius]

onboard nvidia raid controller, using NTFS

[cornfield]

I guess you could also set it up and then unplug a drive to simulate drive failure and see if the other drive is affected. Just use different systems. Anyone use windows home server i think it has some sort of JBOD like pool system, so does Solaris10 with zfs I believe. I am also curious about this as I'm about to start work on a 14 drive rackmountable system.

[Slip Digby]

why would you want to span drives? you're asking for trouble.

if you're worried about data integrity go raid 5

[thebeephaha]

I love my raid5 aside from the fact its so slow, but that wouldn't be an issue if I had a hardware raid controller.

[Sparkyy]

Yeah, if you go spanned volumes, you lose one drive under windows and your whole system goes down in flames. Either live with two drives or else just look in to raid 5.

[xenotype]

Id like to add from experience here, I had two 250 gig, and two 500 gig (all ata) hd's in a JBOD array, one of them started to click, and the only thing i noticed was that any time the array would reference data from that drive, i would get slowdowns, and the painful click of death. I backed up the array onto multiple hd's and re-arranged it, RMA'd the bad drive, and am back at full capacity.

Sometimes, but not always, you can save a non-fault tolerant array from despair, when one of the drives begins to fail.