The data were collected in a Prometheus-like data format: each entry has a timestamp, a value and key-value labels containing additional information. Metrics were gathered from different exporters both open-source and custom. Entries regarding the power consumption and resource consumption per process (i.e., per container) and per server are exposed. The traffic was generated by iperf3 client and server (deployed as Kubernetes pods) running for 24 hours. The data have a time granularity of 10 seconds. Three scenarios are evaluated: - Server and client running on the same K8s node. - Server and client running on different K8s nodes. - Server running inside the K8s cluster and client outside. - Server running inside the K8s cluster and client outside, with client activating/deactivating every hour.