4.2 EC2 Pricing Models: Savings Plans, Spot Fleets & Graviton

🎯 Key AWS SAA-C03 Architectural Takeaways

• Compute Savings Plans provide up to 66% discounts over On-Demand rates in exchange for a 1- or 3-year hourly spend commitment, applying automatically across families, OS, and regions.
• EC2 Instance Savings Plans offer higher discounts (up to 72%) but are locked to a specific instance family within an individual AWS Region.
• Spot Instances provide up to 90% savings over On-Demand by utilizing spare EC2 capacity, ideal for stateless, fault-tolerant, and interruptible batch workloads.
• AWS Graviton (ARM64) instances provide up to 40% better price-performance over comparable x86 instances for containerized and open-source workloads.

Compute costs typically represent the largest portion of an AWS infrastructure bill. To optimize spending, architects align compute requirements with AWS pricing models: On-Demand, Savings Plans, and Spot Instances. On-Demand instances are reserved for short-term, spiky workloads or newly launched applications where usage patterns are not yet established.

For predictable steady-state usage, AWS Savings Plans deliver substantial discounts in exchange for committing to a consistent dollar spend per hour for a 1- or 3-year term. Compute Savings Plans provide the greatest architectural flexibility, automatically discounting EC2, AWS Fargate, and AWS Lambda regardless of instance family, size, operating system, or AWS Region.

For stateless, fault-tolerant, and asynchronous workloads (such as machine learning training, rendering, CI/CD runners, and batch processing), EC2 Spot Instances offer up to 90% savings by utilizing spare cloud capacity. To protect Spot jobs from interruption, architects deploy EC2 Spot Fleets configured with the 'capacity-optimized' allocation strategy, which launches instances across diverse instance types and Availability Zones with the deepest capacity.

At the hardware architecture level, migrating workloads to AWS Graviton processors delivers immediate price-performance gains. Custom-built by AWS using 64-bit ARM Neoverse cores, Graviton3 and Graviton4 instances (e.g., m7g, c7g, r7g) cost roughly 20% less per hour and provide 20% higher performance than comparable x86 instances, delivering up to 40% net price-performance improvements.

⚠️ Common Pearson VUE / AWS Exam Traps

• Using Spot Instances for stateful databases or critical real-time production workloads that cannot tolerate a 2-minute termination notice.
• Selecting lowest-price Spot allocation strategy for long-running batch jobs — capacity-optimized allocation selects deeper capacity pools to minimize interruptions.
• Running 24/7 steady-state production workloads on On-Demand pricing without evaluating Compute Savings Plans.