perf: Investigate CPU pinning for in nodes

[mmatczuk]

• https://builders.intel.com/docs/networkbuilders/cpu-pin-and-isolation-in-kubernetes-app-note.pdf
• https://www.alibabacloud.com/blog/kubernetes-cpu-management-policies_595205
• https://www.percona.com/blog/2019/05/09/measuring-mysql-performance-kubernetes/

[gnumoreno]

https://kubernetes.io/blog/2018/07/24/feature-highlight-cpu-manager/ https://kubernetes.io/docs/tasks/administer-cluster/cpu-management-policies/#cpu-management-policies https://kubernetes.io/docs/tasks/configure-pod-container/quality-service-pod/#create-a-pod-that-gets-assigned-a-qos-class-of-guaranteed https://docs.openshift.com/container-platform/3.9/scaling_performance/using_cpu_manager.html

[mykaul]

Can we update the status of this issue?

[tnozicka]

I think this is done. @zimnx do you agree or do you recall something missing?

[zimnx]

yes, feature-wise it's done. There're couple of followups to it though: #1151, #1147. We could also invest some time veryfing whether tuning for bigger multi numa instances is correct.

[tnozicka]

Given we track those separately, I'll close this one.