EC2, RDS, storage, egress, and Reserved Instances for mid-market cost optimization.
For most mid-market AWS bills, a few levers drive the majority of waste. This post focuses on what actually moves the bill when you do AWS rightsizing: compute, databases, storage, egress, and reservations. For a full audit framework, see our FinOps audit (2026) and cloud cost optimization checklist.
Over-provisioned EC2 instances are the single biggest leak for many teams. Use CloudWatch (or Cost Explorer utilization) to find instances with consistently low CPU and memory over 14+ days. Downsize to a smaller instance family or move variable workloads to Spot. Lambda: check concurrency and duration; over-provisioned concurrency can add up.
Right-size instance and storage. Look for underused replicas, oversized instance types, and storage that could move to gp3 or lower tiers. Multi-AZ and read replicas are necessary for many workloads but should be justified by usage.
EBS: delete unattached volumes and old snapshots; align volume types and size to actual IOPS needs. S3: move cold data to IA/Glacier; enable lifecycle policies; clean up incomplete multipart uploads and abandoned buckets.
Data transfer out is often the second-largest line item. Optimize with CloudFront for static assets, regional colocation where possible, and review cross-region and internet egress. We go deeper in reduce egress costs (CDN/WAF).
Align RI/SP coverage to steady-state usage. Avoid over-buying for variable or short-lived workloads. Use Savings Plans for flexibility if your mix of EC2, Lambda, Fargate changes.
Get a tailored audit and managed-infrastructure recommendations: run our free FinOps audit.