Fraud detection must flag suspicious card transactions within SECONDS of occurrence. Which processing model fits?
A.Batch processing on a nightly schedule
Nightly batch finds fraud hours too late.
B.Stream processing
Continuous low-latency processing = streaming. Correct.
C.Weekly ETL
Weekly is even slower than nightly.
D.Manual report generation
Manual reporting isn't a processing model for this.
Why this answer
Stream processing handles data continuously as it arrives — sub-second to seconds latency. Batch processes accumulated data on a schedule (minutes to hours latency) — wrong fit for real-time fraud windows.
Correct answer: B · Last reviewed 2026-11-05
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Core Data Concepts practice session.
Practice Core Data Concepts