AWS Certified Data Engineer Associate DEA-C01 Practice Question

An AWS Glue streaming ETL job can read from multiple sources in the same Spark application. By using a JDBC connection it can pull records from the RDS table, simultaneously consume events from the Kinesis stream, perform the join and currency conversion in memory, and write partitioned Parquet files directly to S3. Glue is serverless, so no cluster provisioning or patching is required, and the streaming job naturally meets the 5-minute latency goal.

Amazon Redshift can query data directly from a Kinesis stream and an RDS database (using federated queries), but it is primarily an analytical data warehouse, not an ETL service. Using it for continuous transformations every 5 minutes would be operationally more complex and less efficient than a dedicated Glue job. AWS Database Migration Service supports ongoing replication from RDS but cannot ingest from Kinesis and offers only limited transformation capabilities. Amazon Athena federated queries would require the click-stream data to first be staged to S3 and would also need separate orchestration to achieve near-real-time joins, creating a higher operational burden than a single Glue job.