A nightly analytics job runs thousands of parallel, fault-tolerant tasks for 4 hours. Tasks can be interrupted and retried without harm. Which approach provides scalable, efficient capacity?
A.One very large On-Demand EC2 instance
One big instance cannot parallelize thousands of tasks efficiently.
B.EC2 Spot Instances managed by AWS Batch or an Auto Scaling group
Correct. Spot capacity plus a batch scheduler matches fault-tolerant parallel work.
C.Reserved Instances sized for peak
Reserving peak capacity 24/7 wastes money for a 4-hour nightly job.
D.Dedicated Hosts in a single AZ
Dedicated Hosts are for licensing/compliance isolation, not elastic batch throughput.
Why this answer
Interruptible, retryable batch work is the canonical Spot Instance use case — high parallel capacity at large discounts. AWS Batch can orchestrate queues and compute environments over Spot capacity.
Correct answer: B · Last reviewed 2026-10-25
Official AWS references
Practice this interactively
Answer without seeing the solution first in the Compute Selection practice session.
Practice Compute Selection