Kubernetes Node Calculator (Nodes for Your Pods)
Enter the pods, their resource requests and the node size.
Your numbers
Your results
Kubernetes Node Calculator (Nodes for Your Pods) by TheCalcTools
How the Kubernetes Node Calculator (Nodes for Your Pods) works
The scheduler places pods by their requests, so nodes needed = total requests ÷ allocatable capacity per node, for CPU and memory separately, and also the pod-count limit (110 by default). The largest of the three wins.
Allocatable capacity is lower than the machine size because the kubelet, OS and DaemonSets reserve some. Add spare nodes so a node can fail or drain without pods going pending.
The formula
Worked example
Using the default values shown in the calculator:
| Pods (replicas in total) | 120 |
|---|---|
| CPU request per pod (millicores) | 250 |
| Memory request per pod (MiB) | 512 |
| vCPUs per node | 8 |
| Memory per node (GiB) | 32 |
| Reserved for system and daemons | 10 % |
| Maximum pods per node | 110 |
| Spare nodes for failures or upgrades | 1 |
- 120 pods × 250m CPU and 512 MiB on 8 vCPU / 32 GiB nodes with 10% reserved: 6 nodes (CPU-limited, with 1 spare).
Result:
| Nodes needed (including spares) | 6 |
|---|---|
| Limiting resource | CPU |
| CPU utilisation of requests | 83.3% |
| Memory utilisation of requests | 41.7% |
Frequently asked questions
Why are my nodes full when CPU usage is low?
Scheduling uses requests, not actual usage. Right-size requests from observed usage to pack more pods.
Sources and further reading
- Kubernetes documentation: Considerations for large clusters (110 pods per node)
- Kubernetes documentation: Resource management for pods and containers
Related calculators
More in Cloud & Networking · All calculators
This calculator gives estimates for information only. Check important results independently before you rely on them. See our disclaimer.