In testing MicroShift 4.16 performance, I have found increased CPU consumption relative to 4.15. Specifically we see it when a workload is applied to the kubelet (such as pod creation), as well as after the cluster has been actively used and the CPU idle usage also increases (and stays at this new higher level).

4.16.4

Consistently

1. Use kube-burner to execute a test like node-density
2. Take a look at metrics captured during the test run (or afterwards)
3. Compare this data to the same test run against MicroShift 4.15
4. Notice the increased CPU consumption relative to 4.16

MicroShift 4.16 uses more CPU

MicroShift 4.16 uses a similar amount of CPU compared to 4.15