HPE EVA Storage

EVA 6400/8400 Disk Group Allocation Level

 
JGREEN1
Occasional Visitor

EVA 6400/8400 Disk Group Allocation Level

A little clarification please.  Based upon an earlier support session I had with HP, I was lead to beleive that the Disk Group Allocation Level should not be above a certain percentage based upon the performance level of the disks in the group.  So much so that, when using Near-line 1TB 7200 RPM FATA drives, the allocation level shouldn't go much higher than 50%-60% allocated (DG is single drive protection level, 84 drives).  This seemed a bit excessive, however when we scaled back the allocation levels on our DGs (two other DGs with 15K or 10K drives) to levels less than 80% we did notice a marked improvement in performance.

I realize that fewer presentations means less I/O, but I'm looking for a best practice guide that spells this practice for me.  In doing so, I'm finding many Community postings with people stating that they are at 95% Allocated and that the Allocation Level is considered only when things such as disk failure and leveling are taken into consideration.  I should also point out that the hosts attached are either VMware ESXi, MSSQL clustered servers, or Oracle ASM RAC 11.2 or 12.

My question is this: Given that I don't have the benefit of Performance Advisor to check LUN and host port utilization, etc., how should I go about calculating required Allocation Levels in order to maintain low latencies?  I know that without the tools this will be an in-exact science; however, any guidance is appreciated!!!

Thanks,

Jason