MLA-C01 Question 286
Select 2You are tasked with deploying a machine learning solution on AWS. The architecture requires a managed Jupyter notebook environment for data preparation and model training, a scalable compute cluster for training jobs, and a centralized location to store training and evaluation data. Which combination of AWS services should you script to meet these requirements?
- A
Amazon SageMaker for the Jupyter notebook environment, training jobs, and data storage
- B
Amazon SageMaker for the Jupyter notebook environment and training jobs, and Amazon S3 for data storage
- C
Amazon EC2 for the Jupyter notebook environment, Amazon EMR for training jobs, and Amazon RDS for data storage
- D
Amazon SageMaker for the Jupyter notebook environment, Amazon EC2 Auto Scaling for training jobs, and Amazon S3 for data storage
- E
Amazon SageMaker Studio for the Jupyter notebook environment, Amazon SageMaker Training Jobs for training, and Amazon S3 for data storage
Show answer and explanation
Correct answers: B, E
Explanation
The correct answers are a combination of options that leverage managed solutions to simplify deployment and align with best practices for machine learning on AWS. Amazon SageMaker's notebook instances or SageMaker Studio can provide the managed environment for Jupyter notebooks. SageMaker Training Jobs offer scalable compute resources for training, and Amazon S3 serves as the centralized data storage solution. This combination ensures the infrastructure is optimized and adheres to AWS's machine learning service design principles.
- A. Incorrect.
This option is incorrect because while Amazon SageMaker can handle the Jupyter notebook environment and training jobs, it is not designed to serve as a centralized data storage solution. Data storage in machine learning workflows is typically done using Amazon S3.
- B. Correct.
This option is correct because Amazon SageMaker provides managed Jupyter notebooks (notebook instances) and training job capabilities, while Amazon S3 is an optimized and scalable service for storing training and evaluation data.
- C. Incorrect.
This option is incorrect because using Amazon EC2 for Jupyter notebooks and Amazon EMR for training jobs creates unnecessary complexity compared to the fully managed capabilities of Amazon SageMaker. Additionally, Amazon RDS is not ideal for storing large unstructured datasets like training data.
- D. Incorrect.
This option is incorrect because while Amazon SageMaker can manage the Jupyter notebook environment and Amazon S3 is suitable for data storage, EC2 Auto Scaling for training jobs is less efficient and more complex than using SageMaker's built-in training job functionality.
- E. Correct.
This option is correct because Amazon SageMaker Studio provides a managed Jupyter notebook environment, and SageMaker Training Jobs are designed specifically for scalable machine learning model training. Amazon S3 is the ideal service for centralized data storage in this architecture.