MLS-C01 Question 318
Single answerYou are deploying a machine learning workload on Amazon SageMaker, which involves training multiple models simultaneously using multiple GPU instances. During testing, you receive an error indicating that your account has exceeded a service quota for a specific resource. What is the MOST appropriate action to resolve this issue?
- A
Submit a request to AWS Support to increase the service quota for the required resource.
- B
Switch the workload to CPU instances to bypass the GPU instance quota.
- C
Use the AWS Management Console to directly increase the service quota for the required resource.
- D
Reduce the number of models being trained simultaneously to stay within the service quota.
Show answer and explanation
Correct answer: A
Explanation
AWS enforces service quotas (formerly known as limits) to ensure fair usage and resource availability across accounts. When a quota is exceeded, the correct approach is to determine if the quota can be increased and submit a request through AWS Support or the Service Quotas service. This approach ensures you can scale your workload without compromising performance or functionality.
- A. Correct.
Correct: AWS allows you to request a service quota increase for certain resources through the Service Quotas or AWS Support interface. This is the most appropriate action to resolve the issue while maintaining the workload requirements.
- B. Incorrect.
Incorrect: Switching to CPU instances may avoid the GPU instance quota, but it could significantly degrade the performance of your workload and is not a direct solution to the problem.
- C. Incorrect.
Incorrect: Service quotas cannot be increased directly from the AWS Management Console. You must submit a request through the Service Quotas service or AWS Support.
- D. Incorrect.
Incorrect: Reducing the number of models being trained simultaneously might temporarily avoid exceeding the quota but does not address the long-term scalability of your workload.