Introduction
Cost optimization is one of the six pillars of the AWS Well-Architected Framework and a critical domain on the SAP-C02 exam. AWS Solutions Architect Professional candidates are expected to design architectures that not only meet performance and reliability requirements but also minimize unnecessary spending across compute, storage, networking, and data transfer.
The SAP-C02 exam frequently tests your ability to recommend the most cost-effective solution for a given scenario. This means understanding when to use Reserved Instances versus Savings Plans, how to architect for spot capacity, and how to reduce data transfer costs across regions and accounts.
This guide covers the cost optimization strategies you need to master for the SAP-C02, including compute savings models, storage tiering, data transfer patterns, right-sizing techniques, and the AWS tools that support cost visibility and governance.
The Cost Optimization Pillar
AWS defines cost optimization around five design principles:
- Implement cloud financial management - Invest in understanding your cloud costs and building processes to manage them.
- Adopt a consumption model - Pay only for what you use, and scale resources with demand.
- Measure overall efficiency - Track workload output relative to cost.
- Stop spending money on undifferentiated heavy lifting - Use managed services to eliminate operational overhead.
- Analyze and attribute expenditure - Use tagging and cost allocation to understand where money goes.
For the SAP-C02, you need to apply these principles across multi-account, multi-region architectures and understand how organizational structures (AWS Organizations, consolidated billing) affect cost management.
Compute Savings: Reserved Instances vs Savings Plans vs Spot
Compute costs are typically the largest portion of an AWS bill. The SAP-C02 expects you to know exactly when to recommend each pricing model.
Pricing Model Comparison
| Feature | On-Demand | Reserved Instances (RI) | Savings Plans | Spot Instances |
|---|---|---|---|---|
| Commitment | None | 1 or 3 years | 1 or 3 years | None |
| Discount | 0% | Up to 72% | Up to 72% | Up to 90% |
| Flexibility | Full | Limited to instance type/region | Compute SP: any instance, region, OS | Workload must tolerate interruption |
| Payment Options | Per-hour | All Upfront, Partial, No Upfront | All Upfront, Partial, No Upfront | Market-rate per hour |
| Scope | N/A | Regional or Zonal | Account or Organization | N/A |
| Best For | Unpredictable workloads | Steady-state, known instance types | Flexible compute commitment | Fault-tolerant batch processing |
Reserved Instances in Depth
Reserved Instances come in three classes:
- Standard RIs - Up to 72% discount. Cannot change instance family. Can change Availability Zone, networking type, and instance size within the same family (for Linux).
- Convertible RIs - Up to 66% discount. Can change instance family, OS, tenancy, and payment option. More flexible but less discount.
- Scheduled RIs - (Deprecated for new purchases) Reserved capacity for predictable recurring schedules.
Key SAP-C02 concepts for RIs:
- Capacity Reservation - Zonal RIs provide capacity reservation; regional RIs do not but offer automatic size flexibility.
- RI Marketplace - Standard RIs can be sold on the RI Marketplace if no longer needed. Convertible RIs cannot.
- Consolidated Billing - RIs purchased in one account can apply to linked accounts within the same AWS Organization when sharing is enabled.
- Normalization Factor - Regional RIs use normalization factors to apply across instance sizes within the same family (e.g., one m5.xlarge RI covers two m5.large instances).
Savings Plans in Depth
Savings Plans offer a simpler commitment model:
- Compute Savings Plans - Apply to any EC2 instance, Lambda function, or Fargate task regardless of region, instance family, OS, or tenancy. Maximum flexibility.
- EC2 Instance Savings Plans - Apply to a specific instance family in a specific region. Higher discount than Compute SP but less flexibility.
- SageMaker Savings Plans - Apply to SageMaker usage.
When to recommend Savings Plans over RIs on the SAP-C02:
- When workloads might change instance families or regions over the commitment period
- When the organization uses a mix of EC2, Lambda, and Fargate
- When simplicity of management is a priority
- When migrating workloads across instance types during modernization
When to recommend RIs over Savings Plans:
- When capacity reservation is required (only zonal Standard RIs guarantee capacity)
- When the organization needs to sell unused commitments on the marketplace
Spot Instances and Spot Fleet
Spot Instances offer the deepest discounts but can be interrupted with 2 minutes of notice. For the SAP-C02, know these patterns:
- Spot Fleet - Requests a collection of Spot and optionally On-Demand instances to meet target capacity. Supports diversification across instance types and AZs.
- EC2 Fleet - Similar to Spot Fleet but also manages On-Demand and Reserved Instance capacity in a single API call.
- Spot placement score - Helps identify the best Region or AZ for your Spot workload.
- Capacity-optimized allocation strategy - Launches instances from the pool with the most available capacity, reducing interruption risk.
Common SAP-C02 Spot use cases: batch processing, CI/CD build environments, big data analytics, containerized microservices with graceful shutdown handling, and stateless web application tiers behind load balancers.
Storage Cost Optimization
S3 Storage Class Comparison
| Storage Class | Use Case | Availability | Min Duration | Retrieval Fee |
|---|---|---|---|---|
| S3 Standard | Frequently accessed data | 99.99% | None | None |
| S3 Intelligent-Tiering | Unknown or changing access patterns | 99.9% | None | None (monitoring fee) |
| S3 Standard-IA | Infrequent access, rapid retrieval | 99.9% | 30 days | Per-GB retrieval |
| S3 One Zone-IA | Infrequent, non-critical, reproducible | 99.5% | 30 days | Per-GB retrieval |
| S3 Glacier Instant Retrieval | Archive with millisecond access | 99.9% | 90 days | Per-GB retrieval |
| S3 Glacier Flexible Retrieval | Archive, minutes to hours retrieval | 99.99% | 90 days | Per-GB + per-request |
| S3 Glacier Deep Archive | Long-term archive, 12+ hour retrieval | 99.99% | 180 days | Per-GB + per-request |
S3 Lifecycle Policies
For the SAP-C02, you should be able to design lifecycle policies that transition objects through storage classes automatically. A typical pattern:
- Days 0-30: S3 Standard
- Days 31-90: S3 Standard-IA
- Days 91-365: S3 Glacier Instant Retrieval
- Days 366+: S3 Glacier Deep Archive
- After 7 years: Delete (for compliance data with retention requirements)
EBS Volume Optimization
- gp3 vs gp2 - gp3 volumes are up to 20% cheaper than gp2 and allow independent configuration of IOPS and throughput. The SAP-C02 frequently tests migration from gp2 to gp3 as a cost optimization.
- EBS Snapshots - Use Amazon Data Lifecycle Manager to automate snapshot creation and deletion. Archive infrequently accessed snapshots to the EBS Snapshots Archive tier (up to 75% cheaper).
- Delete unattached volumes - Use AWS Config rules or custom automation to identify and remove orphaned EBS volumes.
Data Transfer Cost Optimization
Data transfer is one of the most overlooked cost categories. SAP-C02 scenarios frequently involve multi-region and hybrid architectures where transfer costs can be significant.
Data Transfer Cost Rules
| Transfer Type | Cost |
|---|---|
| Inbound to AWS | Free |
| Within the same AZ (private IP) | Free |
| Between AZs in same Region | ~$0.01/GB each direction |
| Between Regions | $0.02/GB (varies by region pair) |
| To the internet | $0.09/GB (first 10 TB, tiered pricing) |
| Via NAT Gateway | $0.045/GB processing + standard transfer |
| Via VPC endpoints | $0.01/GB (Gateway endpoints free for S3/DynamoDB) |
Architecture Patterns to Reduce Transfer Costs
- VPC Gateway Endpoints - Free for S3 and DynamoDB traffic. Always use these instead of routing through a NAT Gateway.
- VPC Interface Endpoints (PrivateLink) - Small per-hour and per-GB cost but eliminates NAT Gateway processing fees and internet transfer costs for AWS service calls.
- CloudFront - Data transfer from CloudFront to the internet is cheaper than direct transfer from S3 or EC2. CloudFront to origin (S3, ALB) is free.
- S3 Transfer Acceleration - Uses CloudFront edge locations for faster uploads. Costs more but improves performance for global users.
- Regional data locality - Keep compute and storage in the same region. Process data where it lives rather than transferring it.
- Compression - Compress data before transfer between regions or to the internet.
Right-Sizing and Resource Optimization
AWS Compute Optimizer
AWS Compute Optimizer analyzes utilization metrics and provides recommendations for:
- EC2 instances - Suggests instance types based on CPU, memory, network, and disk utilization
- Auto Scaling groups - Identifies over-provisioned or under-provisioned groups
- EBS volumes - Recommends volume type and size changes
- Lambda functions - Suggests memory configuration changes
- ECS services on Fargate - Recommends CPU and memory configurations
For the SAP-C02, understand that Compute Optimizer uses machine learning to analyze 14 days of CloudWatch metrics (or up to 3 months with enhanced infrastructure metrics) and can be enabled at the organization level.
Right-Sizing Process
- Identify - Use Compute Optimizer, Cost Explorer right-sizing recommendations, or CloudWatch metrics to find over-provisioned resources.
- Analyze - Review utilization patterns. Check peak vs average usage. Consider whether burst capacity (T-series instances) would be sufficient.
- Modify - Resize instances, change instance families, or switch to Graviton-based instances (which offer up to 40% better price-performance).
- Monitor - Verify performance after changes. Set CloudWatch alarms for resource utilization thresholds.
Cost Visibility and Governance
AWS Cost Explorer
Cost Explorer provides visualization and analysis of your AWS costs and usage. Key features for SAP-C02:
- Forecasting - Predicts future costs based on historical usage
- Savings Plans recommendations - Analyzes usage to recommend optimal commitment
- RI utilization and coverage reports - Tracks how well you are using purchased RIs
- Granularity - Daily or monthly views, filterable by service, account, tag, and more
- Resource-level analysis - Drill down to individual resource costs (requires hourly granularity to be enabled)
Cost Allocation Tags
Tags are essential for cost attribution in multi-account architectures:
- AWS-generated tags - Automatically created by AWS (e.g.,
aws:createdBy) - User-defined tags - Custom tags you apply (e.g.,
Environment:Production,Team:Platform) - Activation - Tags must be activated in the Billing console to appear in Cost Explorer and Cost and Usage Reports
- Tag policies - Use AWS Organizations tag policies to enforce consistent tagging across accounts
AWS Budgets
AWS Budgets allows you to set custom spending thresholds and receive alerts:
- Cost budgets - Track actual and forecasted spend
- Usage budgets - Track specific usage metrics (e.g., EC2 hours)
- RI/SP utilization budgets - Alert when utilization drops below a threshold
- Budget actions - Automatically apply IAM policies or SCPs when budgets are exceeded (e.g., deny new resource creation)
AWS Cost Anomaly Detection
Uses machine learning to detect unusual spending patterns. Can be configured per service, per account, per cost allocation tag, or per cost category. Sends alerts via SNS or email when anomalies are detected.
Cost-Optimized Architecture Patterns
Pattern 1: Mixed Compute Strategy
For a workload with predictable baseline and variable peaks:
- Baseline (70% of peak): Covered by Savings Plans or Reserved Instances
- Predictable scaling (next 20%): On-Demand Auto Scaling
- Burst capacity (top 10%): Spot Instances with On-Demand fallback
Pattern 2: Serverless for Variable Workloads
When traffic is highly variable or bursty, serverless architectures eliminate idle capacity costs:
- API Gateway + Lambda instead of ALB + EC2 fleet
- DynamoDB on-demand mode instead of provisioned capacity for unpredictable workloads
- Step Functions for orchestration instead of always-on worker instances
For a deeper dive into serverless patterns, see our guide on serverless architecture patterns for SAP-C02.
Pattern 3: Tiered Storage with Lifecycle Automation
- Use S3 Intelligent-Tiering for data with unknown access patterns
- Implement lifecycle policies for data with known access decay
- Use EFS Infrequent Access for file data accessed less than once per day
- Archive database backups to S3 Glacier Deep Archive
Pattern 4: Multi-Region Cost Awareness
When designing multi-region architectures:
- Use active-passive instead of active-active when RPO/RTO requirements allow, to save on standby region compute costs
- Route data through CloudFront to reduce cross-region transfer costs
- Centralize logging and analytics in a single region rather than replicating across all regions
- Use S3 Same-Region Replication only when needed; cross-region replication adds transfer costs
Pattern 5: Container Cost Optimization
- Use Fargate Spot for fault-tolerant containerized workloads (up to 70% discount)
- Right-size ECS task definitions using Compute Optimizer recommendations
- Use Karpenter (for EKS) or capacity providers (for ECS) to optimize instance selection
- Implement cluster autoscaling with mixed instance types for EKS node groups
SAP-C02 Exam Tips for Cost Optimization
- Read the question carefully for constraints. If a question mentions “minimum operational overhead,” serverless or managed services may be preferred even if slightly more expensive per unit.
- Understand commitment trade-offs. Questions often compare RIs, Savings Plans, and Spot. Know which one fits each scenario.
- Data transfer is a common trap. Many architectures look correct but have hidden data transfer costs. Watch for NAT Gateway processing fees and cross-region transfers.
- Tags and governance scale. In multi-account scenarios, cost allocation tags and AWS Budgets with actions are the answer for cost visibility and control.
- Graviton is often the right answer. When a question asks to improve price-performance without changing the application, Graviton-based instances are a strong option.
Frequently Asked Questions
What percentage of the SAP-C02 exam covers cost optimization?
Cost optimization is woven throughout the SAP-C02 exam, particularly in Domain 2 (Design for New Solutions) and Domain 4 (Accelerate Workload Migration and Modernization). While not a standalone domain, expect cost-related considerations in 15-25% of questions.
Should I choose Savings Plans or Reserved Instances for the SAP-C02?
For exam questions, recommend Savings Plans when flexibility is needed (changing instance types, regions, or using Lambda/Fargate) and Reserved Instances when capacity reservation is required or when the customer wants to sell unused commitments on the RI Marketplace.
How does consolidated billing affect cost optimization?
Consolidated billing through AWS Organizations aggregates usage across all member accounts, which helps achieve volume discounts on services like S3 and data transfer. Reserved Instances and Savings Plans purchased in any account can apply to usage in other accounts when sharing is enabled.
What is the difference between Cost Explorer and the Cost and Usage Report?
Cost Explorer provides a visual interface for analyzing costs with built-in filtering and forecasting. The Cost and Usage Report (CUR) is a detailed CSV/Parquet file delivered to S3 that contains the most granular billing data available. Use CUR for custom analytics with Athena or QuickSight.
How do VPC endpoints reduce costs?
VPC Gateway Endpoints (for S3 and DynamoDB) route traffic through the AWS network instead of the internet, eliminating NAT Gateway processing fees ($0.045/GB) and internet data transfer charges. VPC Interface Endpoints have a small per-hour and per-GB fee but still avoid NAT Gateway costs.
What role does AWS Compute Optimizer play in cost optimization?
Compute Optimizer uses machine learning to analyze historical utilization and recommend optimal instance types, Auto Scaling group configurations, EBS volumes, Lambda memory settings, and Fargate task sizes. It can be enabled organization-wide and provides estimated monthly savings.
How do cost allocation tags work across AWS Organizations?
User-defined cost allocation tags must be activated in the management (payer) account. Once activated, they appear in Cost Explorer and the Cost and Usage Report for all member accounts. Use tag policies in AWS Organizations to enforce consistent tagging standards across the organization.
Can I use Spot Instances for production workloads?
Yes, but only for workloads designed to handle interruption. Use strategies like Spot Fleet diversification across instance types and AZs, capacity-optimized allocation, and graceful shutdown handling. Combine Spot with On-Demand fallback in Auto Scaling groups for production reliability.
Conclusion
Cost optimization on the SAP-C02 is about more than choosing the cheapest option. It requires balancing cost with performance, reliability, and operational requirements across complex multi-account architectures. By mastering compute pricing models, storage tiering, data transfer patterns, and AWS cost management tools, you will be well prepared to tackle cost optimization questions on the exam.
The best way to test your understanding of these strategies in exam-like conditions is through realistic practice questions. Try Sailor.sh’s SAP-C02 mock exams to assess your readiness and identify any remaining gaps before exam day.