New Member


On the MSA 2050 HPE Best Practices recommends 1 volume per controller in a Vcenter clustered environment. I have 2xRaid10 and 1xRaid5 disk groups, all 1.2TB SAS 10K standard hdd's, per controller.  I need to have 6 different datastores housing specific VM workloads. Is it correct to make 1 large volume per controller or have a 1:1 ratio between datastore and diskgroup, thus having multiple volumes per controller?


Re: MSA2050

In order to get full potential of the array, it's suggested to configure it symmetrically i.e. the same disk or disk group configuration in both Pools, and that workloads are distributed between two volumes of which one is in each Pool. Furthermore, the disk configurations should abide to the Power of 2 rule which results in clean Page writes across parity protected Virtual Disk Groups. Datastores should have a 1:1 relationship with a Volume, and minimising the number of Datastores/Volumes by implementing small numbers of large Datastores will result in the best performance.

The best practice of not having too many volumes is less of a concern in Virtual Storage as it was in Linear. Virtual Storage introduces the concept of wide-striping, and therefore I/O for a single volume is distributed across all drives within the Tier in which the Page which is being read or written does or will reside. Of course, if you only have a single Virtual Disk group within the Pool, then this benefit is not realised, however once there are two or more it will quickly demonstrate a tangible improvement.


Hope this helps!

I am an HPE employee

If you feel this was helpful please click the KUDOS! thumb below!


I work for HPE
Accept or Kudo