MLS-C01 Question 281
Select 4A company is deploying a machine learning model as a web service on AWS. The model requires GPU instances for inference and needs to handle variable traffic loads efficiently during peak and non-peak hours. The company also wants to ensure minimal manual intervention in scaling the infrastructure. Which of the following configurations will achieve this requirement?
- A
Use an Auto Scaling group with a launch template configured to use GPU-based EC2 instances.
- B
Set up an Auto Scaling policy based on CPU utilization metrics to scale in and out automatically.
- C
Manually add or remove EC2 instances to the Auto Scaling group based on traffic patterns.
- D
Use an Elastic Load Balancer (ELB) to distribute traffic across instances in the Auto Scaling group.
- E
Schedule scaling actions based on predictable traffic patterns to reduce costs.
Show answer and explanation
Correct answers: A, B, D, E
Explanation
To handle variable traffic loads efficiently with minimal manual intervention, the combination of an Auto Scaling group, resource-based scaling policies (e.g., CPU utilization), Elastic Load Balancer, and scheduled scaling actions for predictable traffic patterns is ideal. This configuration automates scaling, ensures optimal GPU resource utilization, and improves cost-efficiency. Manual scaling is not suitable as it contradicts the requirement for minimal manual intervention.
- A. Correct.
Correct. Using an Auto Scaling group with a launch template configured to use GPU-based EC2 instances ensures that the right instance type is provisioned automatically during scaling.
- B. Correct.
Correct. An Auto Scaling policy based on CPU utilization metrics enables automatic scaling based on real-time resource needs, ensuring efficient handling of traffic fluctuations.
- C. Incorrect.
Incorrect. Manually adding or removing instances introduces operational overhead and does not align with the requirement for minimal manual intervention.
- D. Correct.
Correct. An Elastic Load Balancer (ELB) ensures traffic is evenly distributed among instances, improving availability and performance under variable loads.
- E. Correct.
Correct. Scheduled scaling actions are cost-effective for predictable traffic patterns, reducing unnecessary resource usage during non-peak hours.