MLA-C01 Question 254
Single answerYou are tasked with deploying a machine learning model for real-time inference using Amazon SageMaker. The model requires low-latency predictions, needs to handle intermittent bursts of traffic, and should scale automatically to manage varying workloads. Which deployment option in SageMaker best meets these requirements?
- A
SageMaker Batch Transform
- B
SageMaker Endpoint with Auto Scaling
- C
SageMaker Processing Job
- D
SageMaker Asynchronous Inference
Show answer and explanation
Correct answer: B
Explanation
For real-time inference with low latency and the ability to handle varying workloads, SageMaker Endpoint with Auto Scaling is the most appropriate choice. It allows the model to scale up or down automatically based on the incoming traffic while ensuring minimal latency for predictions. Other options either target batch processing, asynchronous predictions, or data processing tasks.
- A. Incorrect.
SageMaker Batch Transform is used for offline, batch predictions, not real-time inference. It does not meet the low-latency requirement.
- B. Correct.
SageMaker Endpoint with Auto Scaling is designed for real-time inference with low latency and can handle variable workloads by automatically scaling the deployed model instances.
- C. Incorrect.
SageMaker Processing Job is used for preprocessing, postprocessing, and data transformation tasks, not for deploying models for inference.
- D. Incorrect.
SageMaker Asynchronous Inference is suitable for long-running inference requests but does not provide the low-latency capability required for real-time predictions.