How we saved over $3 million in idle compute costs with Datadog Kubernetes Autoscaling

Datadog developed Datadog Kubernetes Autoscaling (DKA) to automate multidimensional workload scaling and provide intelligent resource recommendations, addressing the trade-off between wasteful overprovisioning and risky underprovisioning. By replacing complex, manual configurations with a single unified resource, Datadog’s Rapid team significantly improved service reliability and reduced costs by more than 50% in initial rollouts. Ultimately, the company-wide adoption of DKA has eliminated over $3 million in annualized idle compute costs across tens of thousands of deployments.