MLS-C01 Question 263
Select 4You are managing a machine learning model deployed on Amazon SageMaker. The model processes real-time predictions for thousands of requests daily. Recently, you noticed an increase in errors for some prediction requests, and you want to build a robust error monitoring solution to detect and address such issues proactively. Which of the following steps should you take to monitor and analyze these errors effectively?
- A
Enable Amazon CloudWatch Logs for your SageMaker endpoint to capture and analyze invocation errors.
- B
Set up SageMaker Model Monitor to track data quality and generate alerts for drift or anomalies in input data.
- C
Create an Amazon SNS topic and subscribe to it for real-time notifications of invocation errors logged by SageMaker.
- D
Use AWS Lambda to parse CloudWatch Logs for error patterns and store aggregated metrics in Amazon S3 for further analysis.
- E
Deploy a SageMaker endpoint configuration with automatic retries to mitigate temporary invocation errors without logging them.
Show answer and explanation
Correct answers: A, B, C, D
Explanation
To build a robust error monitoring solution for a SageMaker endpoint, you should use a combination of tools like CloudWatch Logs for capturing and analyzing invocation errors, SageMaker Model Monitor for detecting data quality issues, and Amazon SNS for real-time notifications. AWS Lambda can further help parse logs and store insights for analysis. Automatic retries, while useful for transient errors, do not address the need for proactive error monitoring and analysis.
- A. Correct.
Correct. Amazon CloudWatch Logs can be used to capture invocation errors and analyze them. This is an essential step in error monitoring.
- B. Correct.
Correct. SageMaker Model Monitor helps in detecting data quality issues, drift, and anomalies, which could lead to errors in predictions.
- C. Correct.
Correct. Amazon SNS can be used to receive real-time notifications when errors are logged, enabling faster response to issues.
- D. Correct.
Correct. AWS Lambda can process CloudWatch Logs to identify specific error patterns, and storing metrics in S3 allows for deeper analysis.
- E. Incorrect.
Incorrect. While automatic retries can mitigate some errors temporarily, it does not contribute to monitoring or logging the errors for long-term analysis.