Run DuckDB analytics on your Amazon DynamoDB data with zero-ETL

This article demonstrates how to run DuckDB analytics on Amazon DynamoDB data using a zero-ETL integration that replicates data to Apache Iceberg tables on Amazon S3 Tables.

• Zero-ETL integration replicates DynamoDB data to S3 Tables on a 15-minute refresh interval using point-in-time recovery backups
• DuckDB runs as an embedded analytical database in AWS Lambda, executing SQL queries against Iceberg data without server management
• Lambda function URL with IAM authorization provides secure HTTPS access to run ad hoc SQL queries and return JSON results
• Solution uses AWS CDK construct to simplify setup of required resources, permissions, and integrations
• Data remains queryable by Athena, Redshift, EMR, and other Iceberg-compatible engines from the same S3 Tables catalog
• Queries return in ~400ms on warm invocations; costs include PITR backup storage, initial export, CDC replication units, and Lambda invocations

This approach separates analytical workloads from operational DynamoDB traffic while maintaining data freshness through automated replication without custom ETL pipelines.