How does the EVA occupancy levels work?

[Thomas Kemp_1]

The EVA has two "pools" of storage: user space for creating disks and a second "pool" for the disk failure protection, right?

If you have a disk group with "double" disk failure protection, and if you set an occupancy level of 80% on a disk group, and it trips that limit, does it lower your disk failure protection then, or does the failure protection drop to "single" and to "none", once the user space has been completely exhausted?

If you start losing disks, is it true that the disk failure protection will drop from "double", "single", and "none", and if it gets really bad with disk failures, the pool of storage for creating disk space will start being used for protection?

[Tom O'Toole]

Actually just the opposite. Free space in the pool of storage for creating vdisks is used to reconstruct in the event of failures, before the area reserved for failure protection is dipped into.

The idea is, you will not be able to use the amount you have allocated for failure protection. If you could use it, you would continue to operate at reduced redundancy in the event of failures, rather than reconstruct to full redundancy. Once you've allocated that failure protection level, it should not drop as a result of creating vdisks (but it could as a result of disk failure).

The terminology is a bit confusing if you ask me, since failure protection is really a function of the VRAID level of your vdisks, not the 'failure protection level' of the disk group, which has to do with how much disk capacity has been reserved for the purposes of reconstructing bvdisk raid levels.