MLS-C01 Question 286
Single answerYou are deploying an image classification model as a REST API endpoint using Amazon SageMaker. The model is expected to handle high traffic during peak hours. To ensure availability and fault tolerance, you want to implement load balancing while minimizing operational overhead. Which approach would be the most appropriate?
- A
Deploy the model to a single SageMaker endpoint instance and use Auto Scaling to handle traffic spikes.
- B
Deploy multiple SageMaker endpoint instances under an Application Load Balancer (ALB) for traffic distribution.
- C
Deploy the model to a SageMaker endpoint with Multi-Model Endpoint support and configure Auto Scaling.
- D
Deploy the model to multiple SageMaker endpoints in different regions and use Route 53 for traffic management.
Show answer and explanation
Correct answer: C
Explanation
Amazon SageMaker provides built-in scaling and traffic management capabilities, which eliminate the need for external load balancers like ALBs. Using a Multi-Model Endpoint allows multiple models to share the same endpoint and effectively handle varying traffic patterns through automatic scaling. This approach reduces operational complexity while ensuring high availability and fault tolerance.
- A. Incorrect.
This option is incorrect because using a single SageMaker endpoint instance, even with Auto Scaling, may not distribute traffic effectively, especially during sudden spikes or failures. Load balancing is not explicitly addressed here.
- B. Incorrect.
This option is incorrect because SageMaker endpoints do not directly integrate with Application Load Balancers. SageMaker endpoints already have built-in mechanisms for scaling and traffic management.
- C. Correct.
This option is correct because SageMaker endpoints with Multi-Model Endpoint support can host multiple models and scale automatically based on traffic. This reduces operational overhead and ensures traffic is distributed efficiently.
- D. Incorrect.
This option is incorrect because setting up multiple endpoints in different regions with Route 53 requires more manual configuration and operational effort compared to SageMaker's built-in scaling and traffic management features.