Currently the model-monitoring prometheus doesnt have access to system metrics (cpu,memory,disk) for model-serving pods. While we can get those metrics from the openshift monitoring prometheus, we should ideally have all the metrics we need accessible in one place.