VM total disk latency above a defined threshold (65.0 Milliseconds threshold 50.0 Milliseconds)

[rschmid]

Hello,

on one vSphere 5.5 host we have several alarms regarding VM total disk latency like the following:

warning 30.08.2016 09:06:08 Datastore Highest Latency" (63.0 Milliseconds) is above a defined threshold (50.0 Milliseconds)

Warning 24.08.2016 09:12:54 Datastore Highest Latency" (65.0 Milliseconds) is above a defined threshold (50.0 Milliseconds)

After a short time Alarm metrics come back to normal.

Should we consider relocation these VMs to another datastore very soon?

Kind regards,

Roland

[vfk]

It is a good idea to investigate what could be causing the latency, as VM latency is aggregate of device latency (HBA) and ESXi Kernel.

You will find these two articles helpful in understanding storage latency in vsphere;

Troubleshooting Storage Performance in vSphere – Part 1 - The Basics - VMware vSphere Blog

Troubleshooting Storage Performance in vSphere – Part 2 - VMware vSphere Blog

[rschmid]

Hi,

thank you for your answer. I have a further question regarding defined thresholds.

Viewing the mentioned Latency for e.g. SATA storage of 12.7 MS our latency of 65.0 MS is far beyond the treshold.

Do I understand the thesholds value correct?

(our storage is SATA)

Kind regards,

Roland

[vfk · Accepted solution]

The slide shows the general IOPs and Read & Write throughput you can expect per spindle depending on the RAID configuration and/or drive type you have in your array.

These are not hard numbers, just an indication of what you should expect, but storage configuration vary and it all depends on how your storage is configured and how many disks (spindles) in your storage, and the type of raid configured.

[shane1973]

How many spindles do you have and what brand/model of 7200 RPM drives are they?  Onboard cache can make a slight difference as well as the model of drive. Not all 7200 RPM SATA drives are created equal.