database-cloud-optimization-cost-optimize

Start by analyzing costs by service and resource over a time window, then identify waste such as idle EC2 instances, unattached storage, and old snapshots. Apply rightsizing recommendations based on CPU and memory utilization, and validate changes in staging with rollback snapshots before production rollout.

Use boto3 with CloudWatch to pull CPUUtilization metrics for running instances over the past 7 days. Instances averaging under 5% CPU are flagged as unused, with estimated savings calculated from their instance type cost.

Reserved Instances suit stable workloads with low usage variation, offering the deepest discounts on 3-year all-upfront terms. Savings Plans fit variable workloads by committing to a baseline hourly spend while retaining flexibility across instance families.

Yes, the Skill covers AWS, Azure, and GCP, though the provided implementation playbook contains concrete code samples primarily for AWS services like EC2, S3, EBS, and Cost Explorer. The same analysis patterns apply conceptually to other providers.

Avoid optimization during active incident response or when you lack access to billing and resource data. Always validate changes in staging first and ensure backups and rollback paths exist before resizing or deleting resources.