A team needs Spark-based ETL jobs a few times a day with NO cluster management, paying only per job-second. Which service fits best?
A.Amazon EMR on EC2
EMR means cluster ops and paying for idle cluster time.
B.AWS Glue jobs — serverless Spark billed per DPU-second
Correct. Serverless, per-job billing, catalog-native.
C.Self-managed Spark on EC2
Maximum ops burden for a routine need.
D.Amazon Athena
Athena queries data — it doesn't run Spark ETL.
Why this answer
Glue ETL is serverless Spark — no clusters, per-second billing, integrates natively with the Data Catalog. EMR is the answer when you need specific framework versions, fine cluster control, or long-running analytics clusters.
Correct answer: B · Last reviewed 2026-11-03
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Ingestion & ETL practice session.
Practice Ingestion & ETL