Scale Smarter Not Harder: How Extending Cluster Autoscaler Saves Millions

This conference talk explores how Datadog optimizes cloud infrastructure costs by extending Kubernetes Cluster Autoscaler to make smarter scaling decisions. Learn how to handle demanding resource requests efficiently when scaling thousands of instances daily across multiple cloud providers. Discover the methodology for scoring instance types and implementing these recommendations through the Cluster Autoscaler's gRPC expander. Understand practical strategies that can be applied to both small and large Kubernetes environments to make informed instance type selections that significantly reduce cloud spending. The speakers share real-world experiences of how this approach has saved millions in infrastructure costs at Datadog through intelligent machine specification selection.