Basic Information
| Field | Details |
|---|---|
| English name | Cost Optimization |
| Full name | Cost Optimization |
| Chinese description | 成本优化 |
| Japanese description | コスト最適化 |
| Exam frequency | ⭐⭐⭐⭐⭐ |
| Often confused with | 性能优化 |
In one sentence
Cost optimization is not a one-time budget cut; it is continuous measurement, selection, monitoring, and improvement.
Four core actions
- Right Sizing: choose compute, storage, and database size from real metrics.
- Match Supply to Demand: use Auto Scaling, serverless, and schedules to reduce idle capacity.
- Choose Pricing Model: match commitment and interruption tolerance to pricing.
- Measure and Govern: use tags, budgets, alerts, and cost allocation continuously.
EC2 pricing choices
- On-Demand: short-term, unpredictable, or experimental workloads.
- Savings Plans / Reserved Instances: stable, long-term usage that can be committed.
- Spot: interruptible, fault-tolerant, batch, and retryable workloads.
- Dedicated Host: licensing, compliance, or physical-host control.
Exam focus
- Evaluate interruption tolerance, workload stability, term, and licensing before choosing pricing.
- Auto Scaling, S3 Lifecycle, serverless, and data-transfer paths all affect total cost.
- Savings must not break RTO, RPO, security, or performance requirements.
- The cheapest individual service does not always create the lowest total architecture cost.
Common misconceptions
- Serverless is not always cheapest, and Spot is unsuitable for non-interruptible work.
- Buying a commitment discount does not complete cost optimization.
- Compute-only analysis misses storage, logs, NAT Gateway, and data transfer.
Key takeaway
Start from business goals, then match capacity, size, and pricing model—and keep measuring.
Related services
Cost Explorer, AWS Budgets, Compute Optimizer, Trusted Advisor, Auto Scaling, Savings Plans, S3 Lifecycle.
Pricing principles, tools, and service-level optimization
- Pay as you go, save when you commit, and pay less by using more describe the three pricing principles.
- Track compute, storage, data transfer out, requests, logs, NAT Gateway, cross-AZ or cross-Region traffic, and snapshots.
- Billing shows charges, Budgets sets thresholds, Cost Explorer analyzes history and forecasts, Pricing Calculator estimates planned changes, and Compute Optimizer recommends sizes.
- Right-size EC2; evaluate RDS size, storage, replicas, backups, and cache together; use appropriate S3 storage classes and lifecycle controls.
- Compare the original network path with endpoint pricing: S3 and DynamoDB gateway endpoints have no hourly or data-processing charge, while interface endpoints do.