A company is migrating its data processing application to the AWS Cloud. The application processes several short-lived batch jobs that cannot be disrupted. Data is generated after each batch job is…

Amazon EC2 Spot Instances can be interrupted, which violates the requirement that the batch jobs 'cannot be disrupted.'

While Amazon EC2 On-Demand Instances are suitable for non-disruptible jobs and S3 Glacier Deep Archive is cost-effective for long-term retention, manually transitioning data to S3 Glacier Deep Archive via lifecycle policies has more operational overhead than S3 Intelligent-Tiering, and Deep Archive has longer retrieval times if data unexpectedly needs to be accessed after 30 days.

Amazon EC2 Spot Instances can be interrupted, which violates the requirement that the batch jobs 'cannot be disrupted.'

Deploying Amazon EC2 On-Demand Instances ensures that the short-lived batch jobs run without interruption, fulfilling the core requirement for reliability. Storing the generated data in Amazon S3 Intelligent-Tiering automatically optimizes storage costs by moving objects between frequent access, infrequent access, and archive access tiers (including Deep Archive Access for objects not accessed for 180+ days) based on actual access patterns. This aligns perfectly with the requirement for data accessed for 30 days and retained for 2 years at the lowest possible cost with minimal operational overhead for managing lifecycle policies.