You are working on a data ingestion pipeline that requires replayability. This means that the pipeline should be able to replay the ingested data in case of failures or for reprocessing. Which AWS service would you use to achieve this, and why?