A finance office drops nightly CSV budget extracts into Amazon S3. Analysts complain Athena queries scan too much data and run slowly. Which approach best improves analytic performance for those ad-hoc SQL workloads?
Select an answer to reveal the explanation.
Short Explanation
CSV is like dumping loose receipts on a desk; Parquet is like filing them in labeled drawers Athena can skim fast. Glue ETL can reshape those nightly drops into columnar Parquet so queries read less junk. Cranking concurrency, shell scripts on EFS, or zipping into Glacier won’t fix scan-heavy analytics.
Full Explanation
Converting row-oriented CSV to columnar formats such as Parquet with AWS Glue ETL typically reduces bytes scanned and improves Athena query performance, especially when combined with partitioning and compression. Merely increasing Athena concurrency does not shrink per-query data scanned. Querying CSV on EFS with ad-hoc scripts forgoes the serverless lake pattern. Archiving ZIPs in Glacier Instant Retrieval optimizes cold storage cost and retrieval class, not interactive analytic scan efficiency on hot lake data.