You can now deploy the Kubernetes Cluster Autoscaler in clusters running Kubernetes version 1.17 (and above) to automatically add and remove nodes in specified node pools.

The Cluster Autoscaler adds nodes when pods cannot be scheduled due to insufficient resources. It removes nodes that have been under-utilized for an extended period of time, and when their pods can be placed on other existing nodes. The Cluster Autoscaler scales node pools based on resource requests from pods, rather than resource utilization of a node pool.

You specify which node pools the Cluster Autoscaler is to manage, a minimum and maximum size for each node pool, and how to perform the autoscaling.

As well as the Cluster Autoscaler, you can also:

• dynamically scale the number of pods supporting an application using the Horizontal Pod Autoscaler
• dynamically adjust resource requests and limits for containers running in pods using the Vertical Pod Autoscaler

For more information, see Autoscaling Kubernetes Clusters.