MLS-C01 Question 359
Single answerYou are tasked with deploying a machine learning model for a real-time fraud detection system. The model has been trained and resides in an Amazon S3 bucket. The solution must be highly available, scalable, and capable of handling unpredictable traffic patterns during peak hours. Additionally, you want to minimize the operational overhead of managing servers. Which approach should you use to deploy the model?
- A
Deploy the model using Amazon SageMaker Hosting Services with an endpoint configured for auto-scaling.
- B
Deploy the model on an Amazon EC2 instance and manually configure Auto Scaling Groups.
- C
Use AWS Lambda to host the model and call it for inference.
- D
Deploy the model with Amazon ECS using a Docker container.
Show answer and explanation
Correct answer: A
Explanation
Amazon SageMaker Hosting Services is the best choice for deploying and operationalizing machine learning models in this scenario. It offers managed endpoints with built-in support for auto-scaling, making it highly suitable for unpredictable traffic patterns with minimal operational complexity. Other options either involve higher management overhead or are not optimized for real-time model inference at scale.
- A. Correct.
Correct: Amazon SageMaker Hosting Services provides a fully managed environment for deploying machine learning models. It supports auto-scaling and eliminates the need for managing underlying infrastructure, making it a great choice for real-time, scalable, and highly available solutions.
- B. Incorrect.
Incorrect: While deploying on Amazon EC2 could work, it requires significant manual effort for provisioning, maintaining, and scaling the infrastructure, which contradicts the requirement to minimize operational overhead.
- C. Incorrect.
Incorrect: AWS Lambda is designed for short-lived, stateless tasks, and may not be suitable for hosting machine learning models that require significant memory or processing time. Using Lambda could result in latency issues, especially during high traffic.
- D. Incorrect.
Incorrect: Amazon ECS can be used to host models in Docker containers, but it requires infrastructure management and configuring auto-scaling manually, which increases operational overhead.