We would need to validate that:

• System critical pods aren't swapping.
• Pods of BE/Guaranteed class aren't swapping.

The main point is: swap adds latency at times, on the low-level on memory allocations, but mostly to the system if there is too much swapping, and we have seen pod LivenessChecks to start to fail due to swap traffic.
Because of this we try to find a mechanism how critical cluster and node level pods can be protected from swapping. In order to prevent essentail services to stay healthy.

If we protect system services, then in a KNI world we also have to protect "infrastructure" worklaods such as operators, controllers, CSI pods, CNI pods, etc.

It's less about a user to say I want to disable swap for this pod.