AWS Certified Machine Learning Specialty · Free Practice Question Medium
Question 9
A machine learning development team is in the process of deploying a custom machine learning model using Amazon SageMaker. The model has been containerized to facilitate deployment. To ensure successful deployment and operation of the custom container on Amazon SageMaker, which of the following requirements must be met by the container? (CHOOSE TWO)
-
A
Must respond to /invocations and /ping requests on port 80.
-
B
Must respond to /invocations and /ping requests on port 8080.
-
C
Must respond to GET requests on the /ping endpoint within 5 seconds.
-
D
Must accept all socket connection requests within 250 ms.
-
E
Must be compressed in ZIP format for deployment.
Reveal correct answers
Correct answers: B, D
Explanation
Focus on the specific requirements for communication protocol, response time, and compression format for model artifacts.
Correct Choice: Must respond to /invocations and /ping requests on port 8080.
The question specifies that the container must communicate via port 8080, aligning with this choice. This port is often used for web applications, including custom inference containers in SageMaker.
Correct Choice: Must accept all socket connection requests within 250 ms.
While fast response times are important, the question specifically mentions a 2-second response time for ping requests, not a 250 ms requirement for all socket connections. This choice introduces a more general condition not specified.
Incorrect Choice: Must respond to GET requests on the /ping endpoint within 5 seconds.
The stipulated requirement is under 2 seconds for ping responses. This choice relaxes the requirement unnecessarily, misaligning with the specific constraint provided.
Incorrect Choice: Must respond to /invocations and /ping requests on port 80.
The container's specified response port is 8080, not the standard HTTP port 80. This choice contradicts the explicit requirement.
Incorrect Choice: Must be compressed in ZIP format for deployment.
The strategy correction is needed based on the new information: Model artifacts need to be compressed in tar format, not zip. This choice would be incorrect based on the revised requirement for model artifact compression, highlighting the importance of understanding AWS and container-specific packaging formats.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
