1. Where the bill comes from #
The three major items.
Cost drivers #
- Compute: instance type and duration, Lambda (requests and duration), containers.
- Storage: volume, class, requests, forgotten snapshots.
- Data transfer: egress to the Internet, between AZs and between regions, NAT Gateway (processing per GB).
Common mistake: NAT Gateways, unattached EBS volumes and IP addresses, and old snapshots are frequent sources of wasted spending.
2. Measuring and alerting #
Cost Explorer, Budgets, tags.
Tools #
| Tool | Usage |
|---|---|
| Cost Explorer | Analysis by service, account, tag; forecasts |
| AWS Budgets | Alerts on actual and forecast cost or usage |
| Cost allocation tags | Attribute costs to teams and projects (projet, environnement) |
| Cost Anomaly Detection | Automatic detection of drifts |
| Compute Optimizer, Trusted Advisor | Right-sizing and savings recommendations |
aws ce get-cost-and-usage --time-period Start=2026-10-01,End=2026-10-28 --granularity MONTHLY \
--metrics UnblendedCost --group-by Type=DIMENSION,Key=SERVICE
aws budgets create-budget --account-id 123456789012 --budget file://budget.json --notifications-with-subscribers file://alertes.json3. Optimizing #
The effective levers.
Levers #
- Delete what is not used (orphaned resources, forgotten environments).
- Right-size based on measured usage.
- Stop staging at night and on weekends.
- Commit the steady baseline (Savings Plans) and put interruption-tolerant processing on Spot.
- Choose Graviton and suitable storage classes.
Good practice: Set up a monthly cost review per team: visibility changes behavior more than any tool.
