Latency on 10Gbit network

[Wouters_2]

I have a discussion with a customer about network latency.
They say the ping times are not stable enough and this causes failures in their mail business application. As far as I can see there is no package loss on the network so it’s pure latency where we are talking about.
Now I want to know if the latency we see on the network is “normal” or is there something wrong ?

Let me first give a description of their network environment:

They running a Windows 2019 Hyper-V Cluster on 4 HPE Proliant DL360 nodes.
All have Intel gold 6244 CPU’s , 384 GB memory, and HPE 562SFP+ 10Gbe network adaptors.
This is connected to an IRF of 2 HPE 5510 Switches with DAC cables.
Latest drivers / SSP are installed.

I know that ICMP is a low priority protocol so it’s not the best way to test performance, but the reseller of there core application says that they don’t see this kind of latency spikes by other customers. They blame the errors in there application on the latency spikes.

When I drain a node and start a ping from an other physical with almost same specs as the nodes I get a ping from around 0.2 ms (using fPing).
When is move back some Vm’s and the servers get some load the baseline of the ping stays almost the same (let’s say 0.3 ms).
The other thing what happens is that there are more spikes starts to show up.
Every few second a small raise to 1 or 2 ms.
Also there are more spike, multiple per minute from 4 to 10 ms (and sometimes higher).

When I test the latency with TCP there are also spikes from around 10 ms.

Can someone tell me it this is normal for this configuration with on a 10Gbit network?
10 ms is not very much, however this is 30 times higher than the baseline.

[MV3]

Hello 

Please raise a support case with HPE for further analysis. 

cheers