Calculate total serving capacity of a Kubernetes Deployment across its replicas.
Cluster-wide deployment capacity is bounded by node count multiplied by the effective pods-per-node ceiling (whichever of CPU, memory or kubelet max-pods binds tightest). Comparing current total replicas against that ceiling shows how much room remains before the cluster autoscaler needs to add nodes, or before pods start going Pending due to insufficient schedulable capacity.
maxPods = nodes × podsPerNode
Most teams target 70-80% steady-state utilization, keeping 20-30% headroom for traffic spikes, rolling deployments and unscheduled pod restarts without waiting on cluster-autoscaler node provisioning time.
No — subtract DaemonSet pods (CNI, log shippers, monitoring agents) per node from pods-per-node first, since they consume capacity on every node regardless of workload replicas.
This calculator shows a static snapshot; pair it with the cluster autoscaler calculator to model how quickly new nodes come online once utilization crosses your scale-out threshold.