cAdvisor

cAdvisor is a container and node resource analyzer that continuously collects CPU, memory, filesystem, network, and process metrics for every container on a host and exposes them through a web interface and a Prometheus endpoint. It attaches to running containers to read their cgroup statistics, keeps recent usage in memory, and reports lifecycle events as containers start and stop.

Platform engineers use it to see which container is eating memory or CPU long before the host becomes unstable.

Deployed on your own infrastructure it keeps performance telemetry inside your cluster, adds no licence cost per host or container, and is reconfigured with the same tooling as everything else.