Amazon Athena pricing guide details
Pricing Overview
Amazon Athena pricing is based on data processed or compute used. SQL queries can be billed based on data scanned, while Capacity Reservations use compute-based pricing; the published example uses $0.30 per DPU hour. Apache Spark applications are charged at an hourly rate for executing code.
Athena queries data directly from Amazon S3, with no additional storage charge for querying data through Athena. Standard Amazon S3 charges apply for storage, requests, and data transfer, including query results stored in an S3 bucket. Federated SQL queries are billed per TB scanned, subject to a 10 MB minimum per query, and can also incur standard AWS Lambda charges.
Plan Comparison
Athena pricing is based on data processed or compute used. SQL queries can be billed by data scanned, while SQL queries with Capacity Reservations use DPU-hours.
Scroll sideways to see all columns. With a keyboard, focus this table and use the arrow keys.
| Pricing approach | Cost basis |
|---|---|
| SQL queries | Data scanned |
| Capacity Reservations | $0.30 per DPU-hour |
Athena queries data directly from Amazon S3. Athena does not charge additional storage fees for querying data, but standard S3 charges apply for storage, requests, and data transfer. Query results stored in an S3 bucket are also billed at standard S3 rates.
Hidden Costs and Considerations
Several costs sit outside Athena's query pricing that teams often overlook. S3 storage fees apply for all data Athena reads, and S3 GET request charges accumulate with every query. Data transfer costs apply when moving results across AWS regions. Cancelled queries still incur charges for data scanned before cancellation, which catches teams off guard during development. Using uncompressed CSV files instead of Parquet can inflate costs 5-10x for the same dataset. Partitioning strategies also matter — poorly partitioned tables force full scans.
Cost Estimates by Team Size
Amazon Athena pricing is based on data processed or compute used, rather than a fixed charge by team size. SQL query charges can be based on data scanned, while Apache Spark code is charged at an hourly rate.
For Capacity Reservations, AWS’s pricing example calculates cost at $0.30 per DPU hour. In that example, 160 DPUs for 15 minutes cost $12.00, and 16 DPUs for 45 minutes cost $3.60, for a total of $15.60 for one hour.
Actual costs also depend on related AWS services. Data stored in or read from Amazon S3 is billed at standard S3 rates, and AWS Glue Data Catalog use is charged at standard Data Catalog rates.
How Amazon Athena Pricing Compares
Amazon Athena is a serverless, interactive analytics service that lets you analyze data where it lives using SQL or Apache Spark, with no infrastructure to set up or manage. Pricing is based on data processed or compute used.
For SQL workloads using Capacity Reservations, Athena charges $0.30 per DPU-hour. Workgroups can separate users, teams, applications, or workloads, set limits on query or workgroup data processing, and track costs.
Athena queries data directly from Amazon S3. Athena does not add storage charges for querying that data, but standard S3 charges apply for storage, requests, and data transfer, including storage of query and Spark calculation results. AWS Glue Data Catalog and federated-query Lambda usage can also incur their respective standard charges.