AWS Athena
Source (Direct Connection) — Vendo queries your data through AWS Athena. Data stays in your S3 buckets — nothing is copied.
What Vendo Accesses
Vendo connects to AWS Athena and queries your S3-backed tables for use in attribution, segments, and data models.
Supported Data
| Data | Description |
|---|---|
| Glue Catalog Tables | Any table registered in AWS Glue Data Catalog |
| Partitioned Tables | Supports Hive-style partitioning on S3 |
| Multiple Formats | Parquet, ORC, CSV, JSON, and Avro |
| Federated Queries | Cross-account and cross-region queries via Athena federation |
Connection Behavior
| Behavior | Description |
|---|---|
| Direct Query | No data import — queries run through Athena |
| S3 Results | Query results staged in your S3 output bucket |
| IAM Auth | Authenticates via IAM access key or role |
| Schema Discovery | Auto-discovers Glue Catalog databases and tables |
Platform Details
| Setting | Value |
|---|---|
| Auth Method | IAM access key (access key ID + secret) |
| Connection Type | Direct connection (no sync) |
| Protocol | Athena JDBC / REST API |
| Billing | Queries billed per TB scanned in your AWS account |
Setup
- Navigate to App Connections > Add App Connection
- Select AWS Athena from the data warehouse category
- Enter your AWS region and Athena workgroup
- Provide IAM credentials (access key ID + secret access key)
- Specify the S3 output location for query results
- Click Save — Vendo will discover your Glue Catalog tables
Permissions Required
| Permission | Purpose |
|---|---|
athena:StartQueryExecution | Run queries |
athena:GetQueryResults | Read query results |
s3:GetObject on data buckets | Read source data |
s3:PutObject on output bucket | Write query results |
glue:GetTable, glue:GetDatabase | Discover table schemas |
Troubleshooting
- Access denied — Check IAM policy includes all required permissions for Athena, S3, and Glue
- Tables not appearing — Verify the Glue Catalog database exists in the specified region
- High costs — Athena charges per TB scanned. Use partitioned tables and columnar formats (Parquet) to reduce costs
Last updated on