Unit economics for Kubernetes and traditional workloads