AWS Certified Data Engineer Associate DEA-C01 Practice Question

Hourly sales data lands as multiple CSV files under s3://sales/raw/YYYY/MM/DD/HH/. A _SUCCESS flag finishes the batch. The team must launch an AWS Glue job only after the flag arrives, retry transient errors with exponential backoff, and send one SNS alert if all retries fail, using as little custom code as possible. Which architecture satisfies these needs?

Deploy Amazon Managed Workflows for Apache Airflow and create a DAG that polls the S3 prefix every minute, runs the Glue job when files are complete, and includes Python-based retries and SNS alerts.
Create an Amazon EventBridge rule that matches the _SUCCESS object, then invoke an AWS Step Functions state machine that uses a Glue StartJobRun task with Retry and a Catch path that publishes to SNS.
Configure an S3 PUT event to invoke a Lambda function for every file. The function tracks files in DynamoDB and starts the Glue job when all files are present, implementing retries and SNS publishing in code.
Schedule the Glue job with an hourly Glue trigger, enable job-level retries, and use a CloudWatch alarm on job failures to notify SNS.

Report Issue

Answer Description

An Amazon EventBridge rule can watch for an S3 ObjectCreated event whose name matches the _SUCCESS flag. When the rule fires, it starts an AWS Step Functions state machine. Step Functions offers a native Glue StartJobRun integration, so the job can be invoked without custom code. The state machine definition can include a Retry clause with exponential backoff and a Catch path that calls the built-in SNS Publish integration to send a single alert if the job ultimately fails. All orchestration, retries, and notifications are handled by managed services, so no Lambda functions or custom DAG code are required. Glue time-based triggers could start too early, Lambda plus DynamoDB adds undifferentiated code, and MWAA requires additional infrastructure and Python logic.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.