AWS EC2 Pricing Models: SAA-C03 Study Guide
Mastering EC2 pricing is essential for the SAA-C03 exam, as “Design Cost-Optimized Architectures” represents a significant portion of the scoring. Selecting the wrong pricing model can lead to unnecessary costs or application downtime.
The Real-World Analogy
Think of EC2 pricing like housing:
- On-Demand: A hotel room. You pay a high nightly rate but can leave whenever you want.
- Reserved Instances/Savings Plans: Signing a 1-year or 3-year apartment lease. You get a much lower monthly rate because you committed to staying.
- Spot Instances: A “standby” flight ticket. It’s incredibly cheap, but the airline can kick you off if a full-paying passenger shows up.
- Dedicated Hosts: Buying the entire building. You have physical control over the hardware, often for legal or licensing reasons.
Core Pricing Models
1. On-Demand Instances
Pay for compute capacity by the second (minimum 60 seconds) with no long-term commitment. Best for short-term, irregular workloads that cannot be interrupted.
2. Reserved Instances (RI)
Provide a significant discount (up to 72%) compared to On-Demand. You commit to a specific instance configuration (Type, Region, Tenancy, OS) for a 1 or 3-year term.
- Standard RI: Highest discount, but you cannot change instance families.
- Convertible RI: Lower discount, but allows you to change instance families, OS, and tenancies.
- Scheduled RI: For workloads that run on a recurring schedule (e.g., every Thursday night).
3. Savings Plans
The modern alternative to RIs. You commit to a consistent amount of usage (measured in $/hour). Compute Savings Plans are the most flexible, applying to EC2, Fargate, and Lambda regardless of region or family.
4. Spot Instances
Leverage unused EC2 capacity for up to a 90% discount. The Catch: AWS can reclaim the instance with a 2-minute warning. Ideal for stateless, fault-tolerant, or batch jobs.
5. Dedicated Hosts vs. Dedicated Instances
- Dedicated Host: Physical server fully dedicated to your use. Essential for Bring Your Own License (BYOL) and strict regulatory requirements.
- Dedicated Instance: Instances running on hardware dedicated to a single customer, but you don’t have control over the physical host (no socket/core visibility).
Comparison Table
| Model | Discount | Commitment | Best For |
|---|---|---|---|
| On-Demand | 0% (Base) | None | New workloads, Unpredictable spikes. |
| Reserved (Standard) | Up to 72% | 1 or 3 Years | Steady-state, predictable usage. |
| Savings Plans | Up to 72% | 1 or 3 Years | Flexibility across EC2/Fargate/Lambda. |
| Spot Instances | Up to 90% | None | Stateless, Batch, Big Data, Testing. |
| Dedicated Host | Varies | None or RI | BYOL, Compliance, Licensing. |
Decision Matrix: If–Then Guide
- If the workload is stateless and can handle interruptions ➔ Choose Spot Instances.
- If you have “Bring Your Own License” (BYOL) requirements ➔ Choose Dedicated Hosts.
- If you have a steady-state database running 24/7 ➔ Choose Reserved Instances or Savings Plans.
- If you are testing a new app and don’t know the traffic ➔ Choose On-Demand.
Exam Tips and Gotchas
- Spot Instance Termination: If AWS terminates a Spot instance, you are not charged for the partial hour. If you terminate it, you pay for the full hour.
- Spot Fleets: A collection of Spot and On-Demand instances. Can be configured to “capacity optimized” to reduce interruptions.
- Convertible RI vs. Compute Savings Plan: Savings Plans are generally preferred now because they automatically apply to Lambda and Fargate.
- T-Series Credit: Remember that T2/T3 instances use “CPU Credits.” If you run out of credits, performance is throttled to a baseline.
Topics covered:
Summary of key subtopics covered in this guide:
- On-Demand flexibility and use cases.
- Reserved Instance types (Standard vs. Convertible).
- Savings Plans (Compute vs. EC2 Instance).
- Spot Instance mechanics and interruption handling.
- Dedicated Hosts vs. Dedicated Instances (Tenancy).
- Cost-optimization strategies for SAA-C03.
Infographic: EC2 Pricing Strategy
The Spectrum of Flexibility vs. Cost Savings
AWS Organizations: RIs and Savings Plans can be shared across accounts if “Discount Sharing” is enabled.
IAM: Control who can purchase RIs or bid on Spot instances.
Cost Explorer: Use this to get recommendations for RI and Savings Plan purchases.
Scenario: A company has a 24/7 web server and a batch processing job that runs for 4 hours a day but can be paused.
Solution: Use Savings Plans for the web server baseline and Spot Instances for the batch processing to minimize costs.
Auto Scaling Groups (ASG): You can mix Spot and On-Demand instances in a single ASG to balance cost and availability.
Capacity Reservations: Unlike RIs, On-Demand Capacity Reservations ensure you have the capacity when you need it, but you pay On-Demand rates unless combined with an RI.