ScaleIO host disk failure

Hey Matt,

Thanks for replying unfortunately the disks we currently use are 800Gb and the new ones from the Vendor are 960.

Considering the amount of nodes in the Array, I think the performance hit will be the least of my worries. If the array will handle it and rebuild accordingly, I'll be happy! that will work!

I just wondered if ScaleIO will throw up an error if the disk is bigger than the rest in the array?

Ideally you want to keep the same size and rotational speed of devices within the same Storage Pool. This ensures that everything will be evenly loaded and that one (or two) devices don't become the bottleneck to everything else.

For example, if you have 300GB 10K SAS drives today, but will be adding 600GB 10K SAS drives to the same pool -- those 600GB drives will incur overall more I/O per GB than the other drives and will ultimately limit the performance of the entire pool to what those drives can deliver in terms of maximum performance.

In order to optimise this in your situation, my idea would be to first partition the drives to at least match the current drive size you have, and then only use that partition when adding it back into the existing pool. (i.e, short-stroking, and you will waste the rest of the space on those drives).

If that's the delta in size then I wouldn't worry too much about it, you should be good to go as is.

The only other concern is if this is a licensed cluster, you will need to have the sufficient licensed capacity to add these few GB's, otherwise you might get an error (or as mentioned before, you can also just partition them first).

Also if this is a licensed cluster, don't be shy to raise an SR to get any assistance. Good luck!