MLS-C01 Question 271
Select 3You are a Machine Learning Engineer managing a distributed training job on Amazon EC2 instances. To ensure consistency and reduce setup time across multiple instances, you decide to create a golden AMI with all necessary dependencies pre-installed, including Python, TensorFlow, and your custom training scripts. Which steps should you take to create and manage the golden AMI effectively for this use case?
- A
Launch an EC2 instance, install all required software and dependencies, and create an AMI from the instance.
- B
Use AWS Systems Manager Automation to automate the creation and patching of the AMI.
- C
Run a distributed training job on each instance to verify the AMI before creating it.
- D
Include only the necessary dependencies in the AMI to minimize the AMI size.
- E
Store the golden AMI in an S3 bucket for version control and easy access.
Show answer and explanation
Correct answers: A, B, D
Explanation
Creating and managing golden AMIs involves launching a base EC2 instance, installing necessary dependencies, and creating an AMI from it. AWS Systems Manager Automation can help automate this process, making it more efficient and scalable. Additionally, it is a best practice to include only essential dependencies in the AMI to optimize size and performance. However, AMIs are managed within Amazon EC2 and not stored in S3.
- A. Correct.
Correct. This is the standard process for creating a golden AMI: Launch an EC2 instance, configure it with the required software, and then create an AMI.
- B. Correct.
Correct. AWS Systems Manager Automation can streamline and standardize the AMI creation process, making it easier to patch and update the AMI.
- C. Incorrect.
Incorrect. Running a distributed training job to verify the AMI is unnecessary. Instead, you should test the functionality of the software and dependencies on one instance before creating the AMI.
- D. Correct.
Correct. Minimizing the AMI size by including only the necessary dependencies improves performance and reduces storage costs.
- E. Incorrect.
Incorrect. AMIs are stored in Amazon EC2 and cannot be stored in S3. Version control and access are managed within the EC2 AMI management system.