A research institute keeps petabytes of instrument logs in S3. Analysts need serverless SQL over that data, table definitions maintained centrally, and column-level permissions enforced in one place. Which service fits BEST?
Show answer and explanation
Answer: A
Athena runs SQL directly against data in S3 without servers to manage, and it scales to very large datasets. The Glue Data Catalog holds the table definitions, and Lake Formation applies fine-grained permissions such as column-level grants across those tables from one place. A self-managed Presto cluster adds sizing, patching and scaling work, and its local permission files bypass central governance. A relational database is not designed to hold and scan petabytes of logs economically. Glue ETL transforms data but does not offer an interactive SQL endpoint.
Why the other options are wrong
- B. Self-managed Presto requires cluster sizing, patching and scaling, which the requirement rules out. Local permission files also sit outside Lake Formation, so governance would be fragmented.
- C. A relational database would require importing petabytes of logs, which is impractical at that volume. Importing into one table also discards the S3 data lake layout the institute already uses.
- D. Glue ETL jobs transform and catalog data but do not provide an interactive SQL query endpoint for analysts. A query engine would still be missing from this design.