Transparency of Kepler information especially the power models that it is using

[sunya-ch]

What would you like to be added?

In addition to the validation on the CI test, I think we should add transparency for users to know which power source or models they are using in their Kepler exporter.
I think this could be reported in the Kepler status by Kepler-operator or exported by some API call (such as /info) in the kepler service.

Here is initial list that I think we should report:

• power source for each granularity and components (platform, package,..)
  • for power model case,
    • link to model validation report
    • maximum and minimum power number predicting with minimum and maximum resource utilization (for CPU time we can compute from the number of cores and threads)
• resource utilization metric availability
• resource utilization metric validation result (number > 0)
• current idle power (even if it is not exported to the prometheus metrics for the default VM)

Why is this needed?

Currently, when there is an issue on power number, we have to see the head part of the log to confirm that which is not user friendly. It is important to make the source of power of Kepler transparent not only for validating and debugging in the development process but also for transparency in power number reporting.