There are two primary components to autoscaling on EKS: pod and node autoscaling. Pods are scaled up as their request limits are reached, and in turn nodes must be scaled up once the total number of pods exceed the current nodes' capacity.

Pelotech has expertise in both of these aspects of autoscaling. With the help of Kubernetes' native HorizontalPodAutoscaler, we can help you fine tune each of your pods so they scale approriately to your needs. For nodes, the Cluster Autoscaler project paired with managed node groups of both spot and on-demand fallback instances will give you the best of both high availability and cost optimization.

If necessary, we can also help you make any changes to your architecture to allow for horizontal autoscaling. For example, we can help you migrate to stateless services.

Pelotech will perform load testing on your cluster to confirm the autoscaling is as fine tuned as possible. Contact the Pelotech team today for help with your autoscaling needs.