MSA2040 - How to increase performance

[Harlandt]

Hi guys

We are using a HPE MSA 2040 in our 2 node failover cluster.  Against advice I chose to populate the enclosure with 12 x 1.2TB 10K SAS drives and opted for RAID5.  We have 16 virtual machines in use.  None of the VMs are particularely stressed.  Day to day running is fine, however when I log in to remote desktop on 1 or more VMs it is painfully slow.

For reference, here's the spec of the hosts:

HPE DL380 Gen9, Xeon E5-2630 v4 @2.20GHz, 128GB DDR4, HPE 10G 2 port 546SFP+ (to core switch), HP H241 SAS Controller - 12Gb/s SAS (to MSA)

Is there anything I can do to improve performance?  I don't mind adding disks, but I don't fancy major surgery.

Thanks

Tom

[SUBHAJIT KHANBARMAN_1]

There are information missing like you haven't mentioned you are using this MSA2040 as linear array or Virtual array. If linear array then how many vdisks you have created with those drives? if Virtual array then how many VDG created with those drives and for which Pool ? How many volumes created from this MSA and presented as Datastore to Hosts where you are hosting VMs.

If I assume you have created single vdisk or VDG and created single volume out of it. Then presented this to Host as Datastore where 16 VMs running. Then next question is this Windows or VMWare?

As long as this is single Vdisk or VDG there shouldn't be any issue with MSA.

If you are logging in as remote desktop to each VM then you are actually using RAM for that Host which is shared resource among the VMs. The more VMs you do remote desktop your RAM usage getting heavier which is creating performance issue here. So you need to check Host level resource utilization rather than Storage because Storage primarily responsible for space.

Hope this helps!
Regards
Subhajit

I am an HPE employee

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

************************************************************************

[Harlandt]

Hi Subhajit

Thank you for taking the time to reply.

The MSA2040 is set up as 1 single virtual array with 11 of the disks mentioned earlier (plus a spare disk).  There are 2 volumes, one is the cluster volume and the other is the Quorum.  We are just using Pool A.

We are using Windows Hyper-V.

Although the each physical host has 128GB DDR4 do you believe that is the bottleneck?  Am I better off adding a second CPU to each physical host?  The CPU usage on each host is usually around 25% so not particularly stressed.

Thanks

Tom

[SUBHAJIT KHANBARMAN_1]

I can't say if more CPU or more memory required at the Host level or not. You need to check resource utilization at the Host level means at the time of login to remote desktop of any VM you need to monitor how much CPU and memory getting used. That may give you some idea. Then login to second VM through remote desktop and again monitor the resource utilization at the Host level.

You need to keep in mind that login to VM through RDP is happening at the Host filesystem level but Storage works in block level.

Hope this helps!
Regards
Subhajit

I am an HPE employee

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

**********************************************************************