Scaling for Kubernetess
Kubernetes scaling means adjusting capacity to demand and deciding where Pods run. The Horizontal Pod Autoscaler changes replica counts based on metrics, the Vertical Pod Autoscaler adjusts requests, and the scheduler uses node selectors, affinity, taints and tolerations to place Pods on suitable nodes.
Stage: Operations · Part of the Kubernetes roadmap · Last reviewed 19 November 2026
What to learn in Scaling
- Configure a Horizontal Pod Autoscaler on CPU utilisation and read its status
- Explain why accurate resource requests are required for autoscaling to work
- Constrain placement with nodeSelector, affinity and anti-affinity rules
- Use taints and tolerations to reserve nodes for specific workloads
- Differentiate Pod autoscaling from cluster node autoscaling
Scaling interview questions
How does the Horizontal Pod Autoscaler decide how many replicas to run?
It compares the current metric value, such as average CPU utilisation relative to requests, with the target, then scales the replica count proportionally within min and max bounds. It depends on the metrics server and on Pods having resource requests set.
What is the difference between a taint and a node affinity?
A taint is set on a node and repels Pods unless they carry a matching toleration. Node affinity is set on the Pod and attracts it to nodes with certain labels. Taints keep workloads away, affinity pulls them toward chosen nodes.
Where Scaling fits in the Kubernetes roadmap
Scaling belongs to the Operations stage (advanced level), which takes about 22 hours. Scale and release.
Other units in this stage
Back to the full Kubernetes roadmap