: Use the Automatic Workload Repository (AWR) to match VM shapes to actual usage.

: Use Terraform-based stacks to prevent "zombie" resources and ensure clean teardowns.

: Automatically adjust instance counts based on CPU or memory.