The data team must be paged immediately when a nightly Glue ETL job fails. What's the simplest serverless pattern?
A.Poll job status hourly from a cron
Polling adds delay and a scheduler to maintain.
B.EventBridge rule on Glue job state-change events (FAILED) → SNS to the on-call group
Correct. Native events → instant notification.
C.Check CloudWatch logs each morning
Manual checking isn't alerting.
D.A Lambda polling the Glue API every minute
Lambda polling works but reinvents EventBridge.
Why this answer
Glue emits state-change events — an EventBridge rule on FAILED → SNS/Lambda gives push-based, instant alerting with zero polling infrastructure. This event-driven ops pattern recurs across AWS services.
Correct answer: B · Last reviewed 2026-11-03
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Data Operations practice session.
Practice Data Operations