Exam AWS Certified Machine Learning Engineer - Associate MLA-C01 topic 1 question 45 discussion

I pick C. The company needs to implement a solution to record and monitor all the API call events for the SageMaker endpoint. Its needs to RECORD all events.

C. CloudTrail + CloudWatch dashboard + CloudWatch alarm ✅ Correcto. AWS CloudTrail captura todos los eventos de API, incluidas las invocaciones a endpoints de SageMaker (InvokeEndpoint). CloudWatch puede procesar estos logs para generar métricas personalizadas. Puedes usar un alarm de CloudWatch para notificar cuando el número de eventos supera un umbral. Esto cubre tanto registro como notificación

✅ Breakdown of the correct answer: CloudTrail: Logs all SageMaker API calls, including InvokeEndpoint. Provides auditing and visibility into who called the endpoint, when, and with what parameters. Essential for tracking all invocation events (as required). CloudWatch Dashboard + Alarm: Use a CloudWatch dashboard to monitor invocation patterns and metrics. Set up an alarm based on metrics like the number of invocations (from CloudTrail logs or the Invocations metric) to trigger notifications when thresholds are breached.

Cloud trail is not an option for real time solution

CloudWatch Invocations metric: This metric tracks the number of API calls made to a SageMaker endpoint. By adding this metric to a CloudWatch dashboard and setting up an alarm, you can monitor the number of API calls and receive notifications when the threshold is breached.

SageMaker endpoint API calls → Record with CloudTrail → Monitor metrics in CloudWatch → CloudWatch alarm for notifications