A banking company uses an application to collect large volumes of transactional data.

[Reference: Building Idempotent Applications with Kinesis, Alternatives Considered:, B (Checkpoint configuration): While updating the checkpoint configuration can help with some aspects of duplicate processing, it is not a full solution for exactly-once delivery., C (Design data source): Ensuring events are not ingested multiple times is ideal, but network outages can make this difficult, and it doesn’t guarantee exactly-once delivery., D (Using EMR): While using EMR with Flink or Spark could work, it introduces unnecessary complexity compared to handling idempotency at the application level., References:, Amazon Kinesis Best Practices for Exactly-Once Processing, Achieving Idempotency with Amazon Kinesis, , ]