Disadvantages of Disk Failure Protection set to None on EVA3000

[B.LeBlanc]

I have one DiskGroup with 24 146GB disks and roughly 9 VDISKS all configured for RAID5.

Currently Disk Failure Protection is set to none.

Besides having the RAID5 VDISKS running in degraded mode if a disk fails is there any other disadvantage to haveing Disk Failure Protection set to none?

Thanks in advance.

[Ivan Ferreira]

There are several factors that makes hard to predict how muchs disks can fail and still have your raid running, as free space, RSS, RAID level.

Basically, I always use protection single to feel/be more protected. With protection at least single, you could loss more disks.

Also, as disks fails often, you should have more protection. In your current configuration you may not be able to loose more than one disk before replacement of the failed disk.

Then, as won't reconstruct the data until you replace the disk you will be reading from parity that have a performance impact. After the replacement of the disk leveling will start (log process specially with low free space), that also may affect the performance.

[AlexZhang]

Protection level set to "single" and set up the Call-home notification function will be the best choice for your data safety.

[Matt Tourney]

Technically, if you have enough unused capacity in your disk group, your VRAID5 vdisks should reconstruct and your disk group will relevel after the failure of a drive. However, you as the storage administrator will bear the full burden of risk to ensure that you have enough disk space for reconstruction of that data. That is not a burden that I would take on, if I were you. That could be a career-ending decision for a storage administrator if you have a multiple drive failure at some point in the near future...

If you set it to single protection, yes, you will lose roughly 292GB of available capacity in the disk group, but you are reserving that space to ensure that you absolutely will survive a single drive failure in the event that you have allocated all of the "available" disk group capacity to vdisks.

You can change your disk protection level to Single on the fly without impacting your data, provided you have enough available capacity in your disk group. If this is indeed a production server disk array, I highly recommend that you do so.

[Nguyen Anh Tien]

Disadvantages of Disk Failure Protection set to None on EVA3000 are:
1,Performance will be reduce when one disk fails!
2,If more than 2 disks fail, data will be loss.
3, You need spend time and some how risky for re-build RAID when you re-place faulty disk by new disk
3, Hot plug disk is not available on your system.

Advantages of Disk Failure Protection set to None on EVA3000:
1, Increase usable space:
if you set it is single (single protection) will lost twice time of protection disk.
HTH
tienna

[Uwe Zessin]

I am a bit confused by some of these claims.

- Disadvantages:

1: well, whenever a disk fails, you get a performance reduction - it has nothing to do with the protection level which is only a space reservation. On the other hand: with a non-zero protection level or enough free space in the disk group, the recovery will start automatically and take some performance away, too.

2: I think that claim is too broad - it depends on which disks fail and at which interval - you can have protection level:0, have 2 disks fail in 2 different RSS at the same time and still no user data lost (provided that you don't use any VRAID-0 virtual disks). There are also situations, where the failure of two specific disks can result in data loss, even if you have set the protection level to 2 - think about the loss of both mirror members.

3a: what is that supposed to mean? Any hardware change is riscy and in the end you would replace the bad disk even if you have protection level greater than 0

3b: Do you claim you have to shut down the disk array for a replacement when the protection level is 0?


- Advantages

there is no dedicated "protection disk" - the EVA reserves space by using the size of the largest disk in the disk group in its calculation

[B.LeBlanc]

Thanks everyone for your responses

The disk array is currently in an acceptance environment.

When it moves into production I am going to use Single level protection.

I'm new to the EVA's and the default configuration was Disk Failure Protection set to None.

By the time I noticed this we had already used most of the space so there wasn't any room left to lose the 300GB.

Worst case when going to production we will order 2 more 146GB disks, for the cost of the drives it's worth it, just for that fact that we're better protected.

Thanks again everyone.