MLA-C01 Question 258
Single answerYou are tasked with deploying a machine learning model for real-time inference with low latency requirements. The model is trained using Amazon SageMaker and needs to handle unpredictable traffic patterns. Which deployment strategy should you use to meet these requirements?
- A
Deploy the model to a SageMaker hosting endpoint with auto-scaling enabled.
- B
Use Amazon SageMaker Batch Transform to process requests in bulk.
- C
Host the model on an Amazon EC2 instance with a load balancer.
- D
Use SageMaker Processing Jobs to handle real-time inference.
Show answer and explanation
Correct answer: A
Explanation
SageMaker hosting endpoints are designed for real-time inference and can be configured with auto-scaling to handle variations in traffic and meet low-latency requirements. This makes them the ideal choice for such scenarios, compared to options like Batch Transform or Processing Jobs, which are not optimized for real-time use cases.
- A. Correct.
Correct: Deploying the model to a SageMaker hosting endpoint with auto-scaling enabled is the best option for real-time inference with low latency and unpredictable traffic, as it automatically adjusts resources to meet the demand.
- B. Incorrect.
Incorrect: Batch Transform is designed for offline, batch inference jobs and is not suitable for real-time, low-latency scenarios.
- C. Incorrect.
Incorrect: Hosting the model on an EC2 instance with a load balancer may work but would require significant manual setup and management, making it less efficient and scalable than SageMaker hosting endpoints.
- D. Incorrect.
Incorrect: SageMaker Processing Jobs are used for data preprocessing, postprocessing, or batch inference tasks, not for real-time inference.