Question

A data engineering team stores several years of clickstream data as Parquet files in a single S3 prefix (s3://analytics/clickstream/). Analysts query this data with Amazon Athena, almost always filtering by a specific event date and country. Queries are slow and expensive because Athena scans the entire dataset each time. The team wants to redesign the S3 layout to minimize the amount of data scanned per query, with the least ongoing maintenance. What should they do?

Accepted Answer

Reorganize the data into a Hive-style partitioned layout such as /year=YYYY/month=MM/day=DD/country=XX/ and update the Glue Data Catalog so Athena prunes partitions during queries.

Answer

Enable S3 Intelligent-Tiering on the clickstream prefix so infrequently accessed objects move to cheaper storage and reduce query cost.

Answer

Convert the Parquet files to CSV and enable Athena result caching so repeated queries do not re-scan the data.

Answer

Increase the Parquet file size to 1 GB per object and rely on Athena's automatic full-table scan optimization.