This sounds very similar to Trino’s (and by extension, Athena’s) architecture. SQLite -> parquet (for columnar instead of row storage) Lambda -> Worker Tasks F

This sounds very similar to Trino’s (and by extension, Athena’s) architecture. SQLite -> parquet (for columnar instead of row storage) Lambda -> Worker Tasks FDW -> Connector Postgres Aggregation -> Worker Stage We run it in Kubernetes (EKS) with auto-scaling, so that works sort of like lambda.