MSA Storage

Performance degradation with HPE MSA 2040 using Hyper-V VM's

Occasional Collector

Performance degradation with HPE MSA 2040 using Hyper-V VM's

We have experienced inconsistent results when executing performance tests while using an MSA 2040 cluster storage with Hyper-V enviornments. The performance inconsistencies have been observed in multiple in house and external applications. 

Hammer DB has been used to perform a simple test SQL server performance test on VM running in Hyper V hosts and stored in the MSA 2040 cluster storage. Results from these tests showed a performance degradation trend over time.

Rebooting the cluster storages and Hyper-V hosts  temporarily alleviated performance degradation at the beginning of the observations. However, over an extended period of time, rebooting the storage is only capable or recovering a portion of the initial performance levels, and the results trend shows that performance continues to degrade over time.


Re: Performance degradation with HPE MSA 2040 using Hyper-V VM's

What is Controller firmware version running on MSA2040?

Are you using MSA2040 as Linear array or Virtual array?

Is MSA2040 all Components like Controller, Drive enclosure IO module and drives are up to date with firmware or not ?

Is Hyper-V host up to date with latest SPP or firmware/driver/BIOS updated for all components ?

If all of the compoenents of MSA and Host up to date then I would suggest to create two same size Vdisk(if possible created with same number and same type of drives) owning by two different controllers of MSA2040 in terms of Linear array. Then create one volume from respective vdisks. Then present them to the Hyper-V host and create some VMs. Then install same application. Then perform performance test seperately. Check if same setup both Controllers giving similar result or not. You need to capture all statistical outputs from MSA as well. You will come to know if CPU of any of the controller performance bad then you can replace that controller.

In case of Virtual array, instead of Virtual Disk you need to consider two Pools for two controllers and then create Virtual Disk group(VDG) and Volumes accordingly. Then do the similar test as explained above.

Any device will show improve in performance after restart so we can't conclude anything from this.

As you know, an extensive performance analysis/recommendation is available as a chargeable deliverable from the HPE Consulting and Integration team. So I would recommend getting in touch with the Tech services team who carry out an end to end performance analysis to do a detailed study on the environment & provide a report on what best to do to improve the performance on your environment.


Hope this helps!

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


I work for HPE
Accept or Kudo