Athena Pricing Calculator - SQL Queries, Provisioned Capacity & Spark

Reduce Athena costs by: (1) Partitioning tables to scan only relevant data, (2) Using columnar formats like Parquet or ORC with compression, (3) Limiting columns in SELECT statements instead of SELECT *, (4) Using provisioned capacity for predictable workloads, and (5) Implementing query result caching to avoid re-scanning data.

Use Athena provisioned capacity when you have predictable, high-volume query workloads that scan more than 100 TB per month. Provisioned capacity offers consistent performance with DPU-hour pricing, which can be more cost-effective than on-demand ($5/TB) for heavy users. For sporadic or unpredictable workloads, on-demand pricing is typically better.

Athena for Apache Spark charges based on DPU-hours consumed by your notebook sessions. A DPU (Data Processing Unit) provides 4 vCPUs and 16 GB of memory. Pricing varies by region, typically around $0.30-0.40 per DPU-hour. Spark applications automatically scale compute resources based on workload demands, and you only pay for actual usage time.

Athena charges based on the amount of data read from S3 to execute your query, measured in TB. Partitioning, columnar formats (Parquet/ORC), and compression significantly reduce data scanned. For example, a query on a 1 TB Parquet table might only scan 100 GB if properly partitioned and selecting specific columns, costing $0.50 instead of $5.