RAID 5 - replacing failed disks with larger disks

I have a RAID 5 array with 12 disks, each 12 TB. It is controlled by a hardware RAID controller (LSI 9380-4i4e) on Windows Server 2012.

As they eventually fail one by one, if I replace the failed 12 TB disks with 14 TB disks, as I understand it, the RAID array will not get bigger - the smallest disk determines the volume size.

My question: what happens when the last 12 TB disk fails, and I replace it with a 14 TB disk?

What will happen to the volume size? Will it suddenly get bigger, or will a unused partition appear, that I can expand the volume into? Or will the whole array have to be reconstituted, and the data recopied, to take advantage of the larger disks?

Solution 1:

With traditional RAID¹, data is spread linearly – a given sector on the RAID volume will be written to roughly the same place on all physical drives. Adding more space at the end of the RAID volume means adding more space at the end of all physical drives.

So in your case, I expect that the controller will simply ignore the final 2 TB of the disk until all disks have been replaced with something larger. Only at that point, it should offer to expand the volume to the available physical space, which you'll need to do through the RAID controller's management interface. (Usually this can be done live, on a running system.)

When the volume is finally resized, Windows will suddenly see the virtual RAID disk device becoming larger – with some new "unallocated" space at the end, and you will be able to use diskmgmt.msc to expand the last partition to fill that space.

(I haven't used LSI hardware RAID specifically, only HP so far, but there's no reason to expect significant differences in the same "RAID 5".)

¹ As opposed to systems like Btrfs or Storage Spaces, where it's more of a "X copies somewhere, as long as their redundancy is guaranteed" layout that does allow for the array to make use of differently sized disks.