Overview
This section explores AWS cost optimization techniques, financial modeling strategies,
and governance frameworks used to control and forecast cloud spending.
It focuses on practical, real-world FinOps implementation.
Understanding AWS Cost Components
- Compute (EC2, Lambda, ECS, EKS)
- Storage (EBS, S3, EFS)
- Database (RDS, DynamoDB)
- Data Transfer (Inter-AZ, Inter-Region, Internet)
- Support Plans
Compute Optimization Strategy
Right-Sizing
Analyze CPU, memory, and network utilization via CloudWatch.
Downsize instances where utilization remains consistently below 40%.
Auto Scaling
Implement dynamic scaling to avoid paying for idle compute capacity.
Savings Plans vs Reserved Instances
| Feature |
Savings Plans |
Reserved Instances |
| Flexibility |
High |
Moderate |
| Applies To |
Compute usage |
Specific instance family |
| Discount Range |
Up to 72% |
Up to 75% |
Storage Cost Optimization
- S3 Lifecycle Policies (Standard → IA → Glacier)
- EBS volume type optimization (gp3 vs gp2)
- Delete unattached volumes
- Enable Intelligent Tiering
Data Transfer Optimization
- Avoid unnecessary cross-AZ traffic
- Use CloudFront for content delivery
- Architect for same-AZ communication where possible
Tagging & Cost Allocation Strategy
- Environment (Prod, Dev, QA)
- Application Name
- Cost Center
- Owner
- Project ID
Forecasting Model Example
Example: Current monthly spend = $50,000
Expected growth rate = 5% monthly
Projected annual spend:
Yearly Spend ≈ 50,000 × (1.05¹²) × 12
≈ $838,000
FinOps Governance Model
- Centralized cost visibility dashboard
- Monthly cost review meetings
- Engineering accountability model
- Budget alerts & anomaly detection
- Quarterly Savings Plan evaluation
Common Cost Leakage Areas
- Idle EC2 instances
- Unattached EBS volumes
- Old snapshots
- Unused Elastic IPs
- Overprovisioned RDS instances
Key Lessons
- Optimization is continuous, not one-time
- Tagging must be enforced at resource creation
- Architecture decisions directly impact cost
- Engineering teams must own their spend