There’s 2 kind of users concerned with energy:

1. Datacenters and edge. They want their application to run with as little energy as possible (eg because your single-node OCP is powered by a solar panel)
2. Compliance and marketing users, who want to achieve their ESG goals and need to know how they are performing in regards to CO2.

The combination of Resource Optimization for OpenShift and productization of KEPLER should allow us to provide information like this to users: “your container was requesting 500 MB of RAM and 0.4 CPU cores, which costed you $57 USD last week and consumed 302 W. According to what we saw in the past 7 days, you should have requested 385 MB of RAM and 0.2 CPU cores, which would have costed you $44 USD and consumed 198 W”. That should lead customers to either consolidate more workloads in the same cluster, or downsize their nodes, therefore saving energy.

Hint: Kepler (Kubernetes-based Efficient Power Level Exporter) is a Red Hat project which uses eBPF to probe energy related system stats and exports as Prometheus metrics. Let's look at Kepler, and maybe other alternatives, and help our customers to be greener.

https://github.com/sustainable-computing-io/kepler

https://docs.google.com/presentation/d/1IOM6Vg7NhCajNxI5RtdN7dEgxzG48IQCJdLm-KC1NLQ/edit?usp=sharing