MLS-C01 Question 250
Select 2You are tasked with deploying a machine learning model to a production environment on AWS. The solution must handle high traffic, maintain availability during failures, and ensure scalability for unpredictable workloads. Which combination of AWS services and configurations should you use to meet these requirements?
- A
Deploy the model using Amazon SageMaker Endpoint with auto-scaling enabled.
- B
Use an Amazon EC2 instance with the model deployed and manually scale instances as traffic increases.
- C
Store model artifacts in Amazon S3 and serve predictions directly from S3.
- D
Configure Amazon SageMaker Endpoint to use multiple Availability Zones and enable automatic scaling.
- E
Integrate Amazon SageMaker Endpoint with AWS Application Load Balancer for traffic distribution.
Show answer and explanation
Correct answers: A, D
Explanation
To meet the requirements of performance, availability, scalability, resiliency, and fault tolerance for deploying a machine learning model in production, Amazon SageMaker Endpoint with auto-scaling and multi-AZ configuration is the best choice. Auto-scaling ensures that the deployment can handle unpredictable workloads, while multi-AZ configuration provides high availability and fault tolerance in case of failures in specific zones. Other options either lack critical features for scalability and fault tolerance or are not suitable for serving real-time predictions.
- A. Correct.
This is correct. Amazon SageMaker Endpoint with auto-scaling ensures that the deployment can handle increased traffic and scale up or down as needed, meeting scalability and performance requirements.
- B. Incorrect.
This is incorrect. Manually scaling EC2 instances does not provide the required scalability and fault tolerance, as it requires human intervention and cannot guarantee quick responses to traffic surges.
- C. Incorrect.
This is incorrect. Amazon S3 is not designed to serve predictions directly; it is used for storing data or model artifacts, not for hosting real-time inference endpoints.
- D. Correct.
This is correct. Configuring Amazon SageMaker Endpoint with multiple Availability Zones and auto-scaling ensures high availability, fault tolerance, and the ability to handle failures in specific zones.
- E. Incorrect.
This is incorrect. AWS Application Load Balancer is not typically integrated directly with SageMaker Endpoints for traffic distribution; SageMaker already handles this internally with its endpoint configurations.