How we saved over $3 million in idle compute costs with Datadog Kubernetes Autoscaling
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Datadog's platform team Rapid adopted Datadog Kubernetes Autoscaling (DKA) to replace fragmented manual autoscaling across 1,800+ services. DKA's multidimensional scaling mode handles both horizontal replica scaling and vertical resource rightsizing through a single declarative resource, resolving the WPA/VPA incompatibility that previously blocked automated vertical scaling. In an initial data center rollout, DKA cut costs by over 50% by surfacing overprovisioned workloads and automatically rightsizing them. It also identified underprovisioned pods running at 100% CPU and corrected their allocations. Rapid configured 3,000 deployments in a single day, and the approach has since spread to ~30,000 deployments across Datadog, eliminating more than $3 million in annualized idle compute costs.
