Large member disks in an array?

[vr.]

If I understand things right, putting twelve 500 GB drives all in one RAID-0 array is going to equate to about 6 TB.

Many operating systems can't see above 2 TB for booting so how would the data be dispersed on the drives? Do things truncate in unpredictable ways? Or does each drive really only get a hundred MB or so written to it?

 

[PigLover]

The answer depends upon what Raid controller and what OS you are using. Most (better) raid controllers have an option that will carve the array so it shows up to the OS as if it were multiple 2Gb drives. Linux, OS-X, Vista & Win7 all support GPT disks and would have no problem at all with a single 6Gb logical drive.

A better question is this: are you nuts? A 12 drive Raid-0 is a recipe for disaster. The probability of a single drive failure in this config is very high - and losing a single drive in a raid-0 irrecoverably loses the whole array. If you do this, you better have a rock-solid (and frequent) backup approach or at least be working with data you can recover.

 

[vr.]

A better question is this: are you nuts?

Probably!

But the question was more for illustration than for an actual implementation approach.

The "more to the story" which may or may not be relevant is that I have an Areca 1231 controller (circa 2007) and have two sets of 500 GB drives attached to it. The system is an ESXi machine which has its own VMFS limitations in play... But I haven't seen any options to specify the size of the volumes I create like I have in Intel ICH setups. On one Areca RAID set I basically picked 6 drives and chose a mode from the drop down and it created the RAID set with a raw capacity. This started me wondering as to what happens to the unused space with big drives and many, many spindles.

 

[drescherjm]

Normally there is no unused space. Well unless you mix and match large and small drives. As for the 2TB limit, although I have had 2TB raid arrays for 7 years now at work, I have not used them on an operating system with any such limit.

On linux software raid you normally raid partitions (not whole disks) so if you mix and match drive sizes whatever is left on a large drive in an array of smaller drives can be used by partitioning it.

 

[PigLover]

Probably!

But the question was more for illustration than for an actual implementation approach.

The "more to the story" which may or may not be relevant is that I have an Areca 1231 controller (circa 2007) and have two sets of 500 GB drives attached to it. The system is an ESXi machine which has its own VMFS limitations in play... But I haven't seen any options to specify the size of the volumes I create like I have in Intel ICH setups. On one Areca RAID set I basically picked 6 drives and chose a mode from the drop down and it created the RAID set with a raw capacity. This started me wondering as to what happens to the unused space with big drives and many, many spindles.

If your OS can't support more than 2TB, don't create a volume with more than 2TB. Your 6TB Raid-0 Raidset can hold 3 2TB volumesets. Simple.