Based on the number of Pods & Nodes in a cluster, the resource usage differs.
Below some values we identified by testing with the clusterloader tool.
Storage
As each user cluster contains an etcd StatefulSet with 3 pods, it requires 3 5GB volumes.
5GB might be for some scenarios too big, therefore the size can be configured cluster-wide via the values.yaml.
CPU & Memory
Medium HA Cluster
Specs
- 3 API Servers
- 100 Nodes
- 500 Pods
Resource usage:
NAME CPU(cores) MEMORY(bytes)
apiserver-88967d5c9-9chfn 513m 783Mi
apiserver-88967d5c9-cqq6z 257m 662Mi
apiserver-88967d5c9-xj2h2 446m 639Mi
controller-manager-7f86977967-cmdj9 75m 143Mi
dns-resolver-b88b48ddc-lsbxp 1m 7Mi
dns-resolver-b88b48ddc-zjgvl 1m 7Mi
etcd-0 63m 285Mi
etcd-1 86m 331Mi
etcd-2 72m 351Mi
kube-state-metrics-6bc577c974-9927q 18m 47Mi
machine-controller-7678c8668b-n5fzz 28m 25Mi
openvpn-server-b967fd79-26vqv 1m 3Mi
prometheus-0 17m 238Mi
scheduler-85c7699468-mmqrq 14m 71Mi
Total
- Memory:
~3590Mi
- CPU:
~1592m
Small Cluster
Specs
- 1 API Servers
- 10 Nodes
- 100 Pods
Resource usage:
NAME CPU(cores) MEMORY(bytes)
apiserver-57b447b75f-gvz5s 150m 518Mi
controller-manager-5cd975b49-qsqvs 29m 85Mi
dns-resolver-58cb9fdcb5-4tzw6 1m 7Mi
dns-resolver-58cb9fdcb5-zm4wq 1m 7Mi
etcd-0 20m 48Mi
etcd-1 23m 51Mi
etcd-2 41m 76Mi
kube-state-metrics-7cc464c7d8-7tbrl 2m 19Mi
machine-controller-d89b4f87f-mxbx2 5m 17Mi
openvpn-server-5979c74b64-msrzx 1m 2Mi
prometheus-0 6m 118Mi
scheduler-76b5f555fc-qwswn 8m 37Mi
Total