A cost-saving strategy uses AWS EventBridge Scheduler to automatically Start/Stop EC2 and ECS services on a daily schedule, avoiding Lambda functions or code changes.

Key considerations include proper IAM role configuration, UTC-based timezone handling, the difference between stopping and terminating resources, and accounting for 3–5 minutes of application warm-up time.

Practical guidance and a checklist for non-production cost optimization extend this approach to other non-production resources like RDS, DocumentDB, ASGs, and ElastiCache, with potential higher savings from weekend-only shutdowns.

Lessons learned emphasize validating actual usage, keeping schedulers aligned per resource type, and providing manual override options for emergencies or hotfixes.

Architecture details include cron-like schedules (9:00 PM and 12:00 AM IST) and AWS actions such as ec2:StartInstances, ec2:StopInstances, and ecs:UpdateService with desiredCount set to 1 or 0.

The solution rests on four EventBridge schedulers—two for EC2 start/stop and two for ECS scale up/scale down—operating under a dedicated least-privilege IAM role.

Staging environments running 24/7 on EC2 and ECS Fargate but used only about 3 hours daily illustrate an 87.5% waste in compute costs.

Overall takeaway: automating non-production resource shutdowns can generate substantial, compounding savings across larger environments, with typical reductions of 40–70% when applied broadly.