EKS Pod Autoscaling Now Reacts Up to 40x Faster

Hey, Shii-chan here, hope you're doing well!

AWS What's New aws.amazon.com

Today I found a nice update for everyone running EKS.

What was announced?

According to AWS What's New, pod autoscaling just got faster across all Amazon EKS Provisioned Control Plane clusters. AWS did this by increasing the Horizontal Pod Autoscaler (HPA) sync concurrency to up to 40 times the default Kubernetes value.

The story so far

Kubernetes' HPA continuously watches your workload metrics and adjusts pod counts to match demand. But in clusters running hundreds or thousands of HPA objects, how fast the control plane can evaluate all of them in parallel determines how quickly your workloads actually scale. When that "HPA sync concurrency" setting stays low, you tend to see a lag between rising load and pods actually scaling out.

What changes

EKS Provisioned Control Plane now raises HPA sync concurrency up to 40 times the default, so the control plane can evaluate far more HPA objects at once, shrinking the time between detecting increased load and scaling out pods. Large clusters running lots of HPA objects benefit the most.

Dive Deep

This improvement applies automatically to every customer using EKS Provisioned Control Plane, and it requires no configuration changes at all. In other words, your cluster is just faster starting today, with nothing to do on your end. For more details, check the "EKS Provisioned Control Plane" page in the EKS User Guide.

Wrap-up

• HPA sync concurrency is now raised up to 40x the Kubernetes default
• The time between detecting load and scaling out pods is shorter
• Applies to every cluster using EKS Provisioned Control Plane
• No configuration changes needed — it's applied automatically

If you're running a large EKS cluster with lots of HPA objects, this is a welcome win that requires zero effort on your part!