At some point in every Kubernetes journey, teams realize their clusters keep scaling even when utilization is low. Costs rise, workloads become unreliable, and performance issues persist despite more nodes and replicas. Horizontal Pod Autoscalers (HPA) and Cluster Autoscaler help, but not always will solve everything.

This session dives into the critical role of autoscaling configuration through real-world examples and performance benchmarks. We will look at advantages of Vertical Pod Autoscaling as well, and draw on experience from thousands of production environments. And we will demonstrate how small tuning changes can dramatically improve both efficiency, reliability, and cost.

Through comparisons of different scaling strategies across workload types and language runtime configurations, attendees will see how thoughtful A/B Performance Testing and tuning can achieve business goals without over-provisioning or guesswork.