SAP-C02 Question 520
Select 2A company runs a high-performance computing (HPC) workload on Amazon EC2 that processes large datasets for scientific research. The workload is non-critical and can tolerate interruptions but requires significant compute capacity to complete. The company is looking to minimize costs while maintaining sufficient capacity. Which combination of architectural choices should you recommend?
- A
Use Spot Instances for the majority of the compute capacity and implement a Spot Fleet with a mix of instance types.
- B
Use Reserved Instances to ensure the required compute capacity is always available.
- C
Implement Auto Scaling groups with a combination of On-Demand and Spot Instances, using a priority-based allocation strategy.
- D
Enable EC2 Instance Scheduler to start and stop instances based on a predefined schedule.
- E
Use AWS Compute Savings Plans to reduce costs for the workload.
Show answer and explanation
Correct answers: A, C
Explanation
For non-critical HPC workloads that tolerate interruptions, Spot Instances are the most cost-effective choice. Using a Spot Fleet with a mix of instance types ensures capacity flexibility. Additionally, combining Spot and On-Demand Instances in an Auto Scaling group with a priority-based allocation strategy balances cost and availability. Reserved Instances and Compute Savings Plans are not suitable for the dynamic nature of this workload, and EC2 Instance Scheduler does not apply to unpredictable, high-performance tasks.
- A. Correct.
Using Spot Instances is an excellent cost-saving choice for non-critical workloads that can tolerate interruptions. Spot Fleets allow you to request a mix of instance types, increasing availability while minimizing costs.
- B. Incorrect.
Reserved Instances are more suited for steady-state, long-term workloads with predictable usage. They are not ideal for non-critical, interruptible HPC workloads.
- C. Correct.
Auto Scaling with a combination of On-Demand and Spot Instances ensures flexibility and cost efficiency. A priority-based allocation strategy allows the use of cheaper Spot Instances first, with On-Demand Instances as a fallback.
- D. Incorrect.
EC2 Instance Scheduler is used for predictable workloads that need to be turned on/off at specific times. It is not applicable to dynamic, high-performance computing workloads.
- E. Incorrect.
AWS Compute Savings Plans are cost-effective for long-term, steady-state workloads. However, they are not ideal for workloads that rely on Spot Instances for cost optimization.