Data store high read and Write latency

[adityachauhan41]

Hi,

we are observing very frequent spikes read and write latency in one of the data stores, what is the cause this and how to investigate this?

[adityachauhan41]

[daphnissov]

How should we know the cause in your environment? Start by looking at the VMs first.

[MikeStoica]

Also, I'd recommend to use esxtop for troubleshooting.

[pragg12]

Hi,

Welcome to VMTN.

I can see a pattern in your IO graph. Everyday, around same time, IOs peak on your datastore. Now, you have to deep dive on the VMs running from this datastore. Check what jobs/tasks are running on those VMs and their timings, any job running on those particular VMs in vCenter, like snapshot. Note down all timelines and match it with the IO peak time on your datastore. As updated by other person, ESXTOP is your friend.

Also, check what's the disk speed tier (like 10K, SSD etc) of the datastore. If the speed is less but the IO load is more, you may see this graph. Find the offending VM and either fine tune the job(s) or else move the VM to another datastore with more speed.

Edit: formatting

[SK0390]

if you generate ESXTOP command, provide us OUTPUT