Control Plane Resource Calculator
Size control plane CPU and memory for the API server, scheduler and controller manager.
About Control Plane Resource Calculator
Control plane sizing scales with cluster size (nodes, pods, objects) and API request rate. API server needs ~1 core per 250 QPS, etcd needs ~1 core per 500 QPS plus memory proportional to stored objects. These are heuristics based on Kubernetes community guidance for production clusters.
Formula
API Server CPU
apiCPU = max(2, ceil(QPS / 250))
- QPS
- API queries per second
Worked example
With the default values below, Control Plane Resource Calculator returns Total Control Plane CPU of 7. Change any input above and every figure here is recalculated in your browser.
Inputs used
- Cluster Nodesaccepts 1–100,000 nodes
- 50 nodes
- Total Podsaccepts 1–1,000,000 pods
- 2,000 pods
- API Server QPSaccepts 10–100,000 req/s
- 500 req/s
- etcd Object Countaccepts 100–10,000,000 objects
- 10,000 objects
Results
- Total Control Plane CPU
- 7 cores
- Total Control Plane Memory
- 16 GB
- API Server CPU
- 2 cores
- etcd Memory
- 4 GB
Step by step
- 1API server CPU= 2 cores
- 2API server memory= 8GB
- 3etcd CPU= 2 cores
- 4etcd memory= 4GB
- 5Total control plane= 7 CPU, 16GB RAM
Assumptions and limits
- The result is derived only from the 4 inputs listed above. Anything not among them is not part of the calculation.
- In particular, this calculator does not take inflation, tax, fees or charges or compounding frequency into account.
- Figures are rounded for display to whole numbers, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.
How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.
Frequently Asked Questions
Should I run control plane on dedicated nodes?
Yes — for production clusters, dedicated control plane nodes prevent workload interference and ensure the API server, etcd, and scheduler have guaranteed resources.
How does etcd performance affect the cluster?
etcd is the bottleneck for all state changes — slow etcd (high latency, disk saturation) directly causes slow API responses, failed leader elections, and cluster instability.