We help IT Professionals succeed at work.

VMWare 6.5 ESXi Latency to DataStore

At my work place - I regularly see messages like this in Solar Winds warning of latency between ESXi hosts and their data store. I have no idea why 20ms is chosen as the warning level. And I'm not sure exactly what type of latency is being referred to in this context. With ICMP I'm familiar with RTT time. But I imaging this is a different measurement. Does anyone know how vSphere measures "latency" from an ESX to its datastore? From a VM to its datastore? Thank you.

"NetPerfMon Event Log : The total latency between a san-server-01.acmeloans.com and a datastore is above 20 ms." action was executed successfully
Watch Question

Andrew Hancock (VMware vExpert / EE Fellow)VMware and Virtualization Consultant
Expert of the Year 2017
latency could be datastore/storage it's often referred to as well as network.

and it's normal if the datastore is under pressure.

Are you suffering from datastore or VM performance issues ?

20ms is often the quotated magic number...

see here


But you need to take it into context is it 24/7 you see this value ?
amigan_99Network Engineer


Well it's a matter of trying to exculpate the network which I'm more involved with. UCS Blade-Fabric Interconnect-Nexus 6k-Tegile Storage array. 80 Gbps from the FIs to the 6k and the port channels for that barely hit 10% at peak. So when the VMW admins say they're seeing latency in their measures I'm not sure how to gauge that against what I'm seeing on the network side. And I'm not sure what's reasonable for the VM/Win admins to expect from the latency they're seeing.

Thanks much for the article. Next need to see how I can get logging from the array.
VMware and Virtualization Consultant
Expert of the Year 2017
Is this happening all the time ?

using esxtop, you can check if the VM latency too storage is too high.