Cost Optimization

AWS cost management — strategies for right-sizing, pricing models, reserved capacity, and tooling to reduce cloud spend without sacrificing performance.

Overview

Cost optimization is the practice of reducing AWS spend while maintaining required performance and availability — it covers pricing model selection, right-sizing resources, storage tiering, and using AWS cost management tools to gain visibility and control.

Pricing Models

Model	Commitment	Discount vs On-Demand	Best For
On-Demand	None	0%	Unpredictable workloads, short-term, dev/test
Savings Plans (Compute)	1 or 3 yr	Up to 66%	Flexible across EC2, Fargate, Lambda
Savings Plans (EC2 Instance)	1 or 3 yr	Up to 72%	Locked to instance family + region
Reserved Instances	1 or 3 yr	Up to 72%	Steady-state, predictable workloads
Spot Instances	None	Up to 90%	Fault-tolerant, flexible, interruptible
Dedicated Hosts	On-Demand or Reserved	Varies	Licensing (BYOL), compliance requirements

Right-Sizing

Right-sizing means matching instance types and sizes to actual workload requirements — it is the most impactful cost optimization lever for compute.

Process

1. Collect metrics (CloudWatch: CPU, Memory, Network, Disk)
2. Identify underutilised instances (< 40% average CPU)
3. Recommend smaller instance type or family change
4. Implement change (resize, migrate, or terminate)
5. Monitor and repeat continuously

Tools for Right-Sizing

Tool	Description
AWS Compute Optimizer	ML-based recommendations for EC2, EBS, Lambda, ECS on Fargate
AWS Cost Explorer (right-sizing)	RI and instance right-sizing recommendations
Trusted Advisor	Checks for idle and underutilised resources
CloudWatch Agent	Memory and disk metrics (not available by default on EC2)

Storage Cost Optimization

Strategy	Implementation
S3 Lifecycle policies	Transition objects: Standard → IA → Glacier → Deep Archive
S3 Intelligent-Tiering	Automatic tier movement based on access patterns (no retrieval fee)
EBS volume type selection	gp3 over gp2 (cheaper baseline IOPS); delete unattached volumes
EBS Snapshots	Delete old snapshots; use EBS Snapshot Archive for rarely accessed
S3 Storage Lens	Visibility into bucket-level usage patterns and anomalies

Database Cost Optimization

Strategy	Implementation
Aurora Serverless v2	Scales to zero ACU; pay only during active queries
RDS Reserved Instances	1-yr or 3-yr commitment for steady-state databases
DynamoDB On-Demand mode	Pay-per-request for unpredictable traffic; switch to Provisioned for steady
ElastiCache Reserved Nodes	Commit for predictable caching workloads
Read Replicas	Offload reads to cheaper read replicas instead of scaling primary

Network Cost Optimization

Strategy	Implementation
Same-AZ deployment	Avoid cross-AZ data transfer charges where HA is not critical
VPC Endpoints (Gateway)	Free S3/DynamoDB access without NAT Gateway data charges
CloudFront	Cache at edge to reduce origin data transfer
NAT Gateway alternatives	NAT Instance for low-traffic; VPC endpoints where possible
S3 Transfer Acceleration	Only enable when cross-region speed improvement justifies cost

AWS Cost Management Tools

Tool	Purpose
AWS Cost Explorer	Visualise, filter, and forecast spend; RI/SP recommendations
AWS Budgets	Set cost, usage, and reservation budgets with alerts
Cost Allocation Tags	Tag resources to attribute costs to teams, projects, or environments
AWS Cost and Usage Report	Detailed hourly/daily CSV report delivered to S3 for analysis
Savings Plans / RI recommendations	Automated analysis of usage patterns for commitment purchases
AWS Compute Optimizer	Right-sizing recommendations using ML on utilization data
Trusted Advisor	Cost optimization checks (idle RDS, underused EC2, etc.)

Quick Reference

Optimization Area	Quick Win	Tool / Service
Compute	Right-size or terminate idle instances	Compute Optimizer
Compute	Purchase Savings Plans for steady-state workloads	Cost Explorer
Compute	Use Spot for fault-tolerant batch processing	EC2 Spot / Spot Fleet
Storage	Enable S3 Lifecycle policies to tier down cold data	S3 Management
Storage	Delete unattached EBS volumes and old snapshots	Trusted Advisor
Database	Use Reserved Instances for production RDS/ElastiCache	Cost Explorer
Database	Switch DynamoDB to On-Demand for sporadic workloads	DynamoDB Console
Networking	Use VPC Gateway Endpoints for S3/DynamoDB	VPC Configuration
Visibility	Set up AWS Budgets with SNS alerts	AWS Budgets

AWS Implementation Options

Cost Strategy	AWS Service / Feature
Commitment-based discounts	Savings Plans, Reserved Instances
Interruptible compute discounts	EC2 Spot Instances, Spot Fleet
Auto-scale to demand	Auto Scaling Groups, Aurora Serverless, DynamoDB On-Demand
Storage tiering	S3 Lifecycle, S3 Intelligent-Tiering, EBS Snapshot Archive
Spending alerts	AWS Budgets, CloudWatch billing alarms
Usage analysis	Cost Explorer, Cost and Usage Report, Cost Allocation Tags
Right-sizing	Compute Optimizer, Trusted Advisor

SAA/SAP Exam Tips

SAA Tip: "Reduce cost for steady-state workloads" → Savings Plans or Reserved Instances. "Reduce cost for fault-tolerant, flexible workloads" → Spot Instances.

SAP Tip: Compute Savings Plans are more flexible than EC2 Instance Savings Plans — they apply across EC2, Fargate, and Lambda. EC2 Instance Savings Plans offer a deeper discount but are locked to a specific instance family and region.

SAA Tip: "Reduce data transfer costs to S3/DynamoDB" → Use VPC Gateway Endpoints (free). NAT Gateway charges per GB of data processed, making it expensive for high-volume S3 traffic.

Amazon EC2 — purchasing options (On-Demand, RI, Spot, Savings Plans)
Amazon S3 — storage classes and lifecycle policies
Auto Scaling — scale compute to match demand dynamically
Well-Architected Framework — Cost Optimization pillar

Cost Optimization

On this page