AWS Certified Solutions Architect – Associate (SAA-C03)Design Cost-Optimized ArchitecturesEasy
A startup is building a new machine learning inference service that needs to process requests in real-time. The service experiences highly variable traffic patterns, with peak usage during business hours and very low usage overnight. The architects want to minimize operational overhead and pay only for the compute resources consumed during actual inference requests. Which EC2 purchasing option should they choose?
- ASpot Instances
- BSavings Plans
- COn-Demand Instances
- DReserved Instances
Show answer & explanationAnswer & explanation
Correct answer: C. On-Demand Instances
On-Demand Instances are the most suitable choice for highly variable, real-time workloads where you only want to pay for the compute capacity you use, without committing to a long-term contract or risking interruptions.
Why the other options are wrong
- A. Spot Instances are for fault-tolerant, flexible workloads that can tolerate interruptions, which is not suitable for a 'real-time machine learning inference service' that needs guaranteed availability.
- B. Savings Plans offer discounted prices in exchange for a commitment to a consistent amount of compute usage (measured in $/hour) over a 1-year or 3-year term, which is not optimal for highly variable usage with low overnight demand.
- D. Reserved Instances require a 1-year or 3-year commitment and are best for steady-state workloads, which does not fit the 'highly variable traffic patterns' requirement.
EC2 On-Demand Instances
EC2 On-Demand Instances allow you to pay for compute capacity by the hour or second with no long-term commitments.
- No upfront payment or long-term commitment.
- Pay for compute capacity by the hour or second.
- Ideal for unpredictable, short-term, or spiky workloads.
- Highest cost per hour compared to other options but offers maximum flexibility.
Memory trick: On-Demand: Only pay for what you need, when you need it.