I'm reminded of people who use S3 as an eventually-consistent database. Encode your rows in CSV and use them as names of empty objects. Query using paginated LI

I'm reminded of people who use S3 as an eventually-consistent database. Encode your rows in CSV and use them as names of empty objects. Query using paginated LIST requests for a prefix of the columns.