How does this tool convert millicores and memory units?

Millicores (e.g. 500m) are converted to 0.5 CPU cores. Memory suffixes (Ki, Mi, Gi, Ti) use binary power-of-two multiples (1 Gi = 1,024 Mi), standard in Kubernetes.

Why is node sizing based on requests instead of limits?

The Kubernetes scheduler places pods exclusively based on resource requests, not limits. Sizing nodes for requests ensures all pods can be scheduled without hitting OutOfMemory (OOM) eviction.

Kubernetes Resource Estimator & Node Sizer

Estimate cumulative CPU cores and Memory capacity required by multi-document Kubernetes YAML manifests. Automatically scales by replica counts and recommends cluster node specifications offline.

KUBERNETES MANIFEST (YAML) 0 lines

WORKLOAD BREAKDOWN & NODE RECOMMENDATION 0 workloads

Cluster Node Recommendation

Parse manifests to see recommended node count and VM instance sizes.

Workload	Kind	Replicas	CPU Req	RAM Req	RAM Limit
Paste manifests or load sample.

Frequently Asked Questions

How are replicas calculated?

If spec.replicas is defined (e.g. replicas: 4), the container CPU and memory amounts are multiplied by 4. If omitted (as in DaemonSets or standalone Pods), a multiplier of 1 is assumed.

What overhead is factored into the node sizing recommendation?

A 15% allocatable buffer is factored in for standard Kubernetes daemon overhead (including kubelet, containerd, OS processes, and CoreDNS pods).