Fully-managed DuckDB* on AWS Lambda for limitless scalability.
Serverless DuckDB* is as easy asBoilingData.execQuery({ sql: `SELECT COUNT(*) FROM parquet_scan('s3://your-bucket/datafile.parquet');` })
.
Your data is rapidly loaded into AWS Lambda memory from your S3 buckets at query time.
Once data is warmed, we route your queries to the Lambda instances where data is already in memory, so your hit rate is always 100% once the dataset is loaded. You get in-memory computing speed and single tenanted resources to each and every of your queries. No shared clusters, but serverless power.
After your data has been loaded into a Lambda's memory, we keep it warmed for up to 2 hours so that subsequent queries have instant response times.
We run Lambdas in the same region as your data to ensure fast response times and zero egress costs.
Every query gets a dedicated, single tenanted cluster of Lambdas which accesses your data using an IAM role, meaning you are in complete control of what data can be accessed.
We never store or cache your data outside of Lambda memory and your queries will never share compute resources with other users.
Using BoilingData you can run DuckDB on Lambda without any of the headaches.
Your AWS IAM Role - your control
Hosted DuckDB* - high performance embedded OLAP SQL engine
Complete transparency of the Lambda lifecycle
SQL over WebSocket - simple integration and interaction