MSA 2060 Performance Graph

[metrobg]

I'm hoping someone can assist me or shed some light on my concern. We have just deployed an MSA 2062 configured with 6, 1.92 GB  SSD drives using raid 10 as the performance tier, and an MSA DP+ disk group consisting of 8, 1.2 TB 10k drives and 4 1.8 TB drives as the capacity tier. 

The array is being used in a vmware environment currently only hosting about 30 rather small Linux vm's and a couple of Windows servers. What's bothering me is that the latency graph hovers continually hovers between 30 ms and 25 ms, occasionally dropping as low as 15 ms when there is a spike in IO activity. Is that normal, or am I missing something?  Could disk scrubbing be affecting the latency?  Any feedback is greatly appreciated.

[Cali]

Hi,

you can disable scrubbing for some days to check: set advanced-settings background-scrub off

If so, you can schedule scrubbing during low usage time or enable or disable it by schedule.

You will find the commands in: HPE MSA 1060/2060/2062 CLI Reference Guide

Cali

[venkat_y]

Hello@metrobg,

Please check the link below for HPE MSA 2060 storage CLI. 
https://support.hpe.com/hpesc/public/docDisplay?docId=a00105313en_us&docLocale=en_US 
You may schedule the disk group scrubbing by using SMU.
Settings > System > Properties > Scrub properties > Enable/disable Disk group scrub > Click on "Create a schedule."

Hope this helps!
Regards,
Venkat

[venkat_y]

Hello@metrobg,

Let us know if you were able to resolve the issue. If you are satisfied with the answers, please click the "Accept As Solution" button for the most helpful response, as this will benefit all community members.

Regards,
Venkat

[JonPaul]

@metrobg 
If you have those 3 disk-groups in 1 pool then your problem is configuration.
6x 1.92TB  SSDs --- Very Fast
8x  1.2TB  10k HDDs  --- Fast
4x  1.8TB 10K HDDs --- Kinda slow
The MSA will use the SSD tier as a much as it can but when it needs more capacity or data is on the HDD disk-groups it will have decently fast access to data on the 1.2TB disk-group and relatively slow access to the 1.8TB disk-group. These 2 disk-groups will be used evenly by all volumes on the Pool.

No disk-scrubbing will not affect latency with the latest firmware (IN210P002).  The scrubbing process was changed to lower the scrubbing workload.  The scrub on the HDD disk-groups will take days, my 12x 1.2TB MSA-DP+ disk group with minimal I/O takes 5 days, to complete and comprises a single read of the entire disk-group.  The scrub will also halt itself if I/O load is sufficient.  
It is not recommended to disable or schedule scrub as either of these could lead to hidden areas of drive degradation which would be found when you needed to reconstruct your data.

Some places to look for details:
CLI>  show controller-statistics    ==> are the controller CPUs being pushed >90%,  are fowarded commands being sent
CLI>  show host-port-statistics    ==> are the approprate ports being used (i.e. more than 1),  what are the queue depths
CLI>  show disk-statistics   ==> are the HDDs being pushed hard, or a specific disks.  Please note that the SSDs are capable of way more I/O load than HDDs so don't compare SSD to HDD.