An introduction to Amazon Athena.

• Serverless query service analysing data stored in S3 using SQL.
• Supports CSV, JSON, ORC, Avro, Parquet.
• $5 per TB of data scanned (10MB minimum per query).
• Commonly used with QuickSight for reporting / dashboards.
• Using columnar data can help save on costs.
  • Use Apache Parquet or ORC.
  • Improves performance and lowers cost.
  • Glue to convert data to Parquet or ORC.
• Compressing data using bzip2, gzip, or lz4 can reduce the amount of data that needs to be retrieved.
• Partitioning data (ex. by year=2026/month=08/) lets Athena skip irrelevant S3 objects – biggest cost/performance lever of all.
• Uses AWS Glue Data Catalog to store table/schema definitions – Glue Crawlers can auto-discover schema.
• Federated Query lets Athena query data outside S3 (ex. DynamoDB, RDS, CloudWatch Logs) using Lambda-based connectors.
• Permissions live at the S3/Glue Catalog layer, not inside Athena itself — restricting athena:* alone doesn’t block direct S3 access.
• Workgroups isolate teams, enforce result encryption, and cap data scanned per query/day.
• Commonly paired with CloudTrail/VPC Flow Logs for SQL-based log investigation.

More reading:

• https://docs.aws.amazon.com/athena/latest/ug/what-is.html