How Datadog migrated its Kubernetes fleet on AWS to Arm at scale

In 2023, Datadog undertook a massive project to migrate nearly its entire Kubernetes fleet on AWS to Graviton-powered EC2 instances due to Arm's growing adoption, widening software support, and impressive cost and energy efficiency. The migration process involved benchmarking performance and costs for each service, identifying the most expensive workloads, creating a multi-architecture build of each service, ensuring optimal performance in staging, verifying instance availability, deploying to production, tracking the migration with four KPIs, and using Datadog Cloud Cost Management to measure cost savings. The scope of the migration encompassed roughly 70 percent of Datadog's AWS workloads, resulting in a 10 percent reduction in its AWS bill and improved durability and flexibility of systems.