MSA 2062 Extremely Slow

[NimblePS]

Hello,

I seem to be running into a strange performance issue with our MSA 2062 anytime I try using the FC (10K) drives regardless of the RAID type and configuration. I have tried using RAID 1, 5, 6, 10 and also the MSADP+ and they all behave the same way regardless of the disk configuration.

Hardware Details and Configuration:

The array came with 2 built-in SSD drives and 14, 10K, 2.4TB Seagate drives. We are using 4, 10G SFP connections to 2 dedicated 10G Cisco switches using 2 dedicated iSCSI subnets as recommended by the HPE MSA 1060/2060/2062 STORAGE ARRAYS in page 4.

We are connecting 3 servers to the MSA 2062 array using dedicated and routed iSCSI VLANs. We have enabled jumbo frames for each 10g connections and all servers are able to reach all iSCSI data IP’s.

MSA

               iSCSI_A1 10.10.10.21

               iSCSI_A2 10.10.11.21

               iSCSI_B1 10.10.10.22

               iSCSI_B2 10.10.11.22

SERVER01

               iSCSI_A 10.10.10.11

                         A1 10.10.10.21

                          B1 10.10.10.22

               iSCSI_B 10.10.11.11

                          A2 10.10.11.21

                          B2 10.10.11.22

SERVER02

               iSCSI_A 10.10.10.12

                         A1 10.10.10.21

                         B1 10.10.10.22

               iSCSI_B 10.10.11.12

                          A2 10.10.11.21

                          B2 10.10.11.22

SERVER03

               iSCSI_A 10.10.10.13

                         A1 10.10.10.21

                         B1 10.10.10.22

               iSCSI_B 10.10.11.13

                         A2 10.10.11.21

                         B2 10.10.11.22

I have followed the MSA installation guide, Storage Management Guide and the HPE MSA Storage Configuration and Best Practices for VMware vSphere guides and have applied all required settings to the MSA 2062 with no success. 

If I use a RAID 1 group using the 2 SSD’s, I have no issues at all when moving a VM to this VMFS store. However, if I use any of the 10K disks, regardless of the RAID configuration I choose, the performance is terribly slow. This is not a connection issue as I have ruled that out.

The array, hard disks and all servers are up to date with their firmware levels. The MSA health check tool does not report any issues and confirmed the array is in a healthy state. I find it hard to believe that 10k disks are underperforming.

Any help is greatly appreciated.

[ArunKKR]

Hi,

I hope that the controller firmware is the latest version IN110P001,
It has fixes for IO stall which leads to latency  till controller is restarted.

https://support.hpe.com/hpesc/public/docDisplay?docId=a00119276en_us&docLocale=en_US

Ensure that path selection policy is set to Round robin and IOPS limit is set to 1

https://kb.vmware.com/s/article/2069356

Since its MSA 2062 it would be having performance tier license. 
You may consider creating a single Pool using SSD disks and SAS disks to improve overall performance.

Execute the below commands few times when ever you experience latency:

show disk-statistics

Abort any scrub jobs before executing the above command.

If the disk IOPS is above 150 consistently it indicates an overloaded array.

You could try mapping an MSA test volume directly to a windows vm and perform an IO meter test to check the performance.