Base
Hardened build
minimal ports, security patches applied at build time
pg_duckdb DuckDB Analytics Engine for PostgreSQL 17 on Ubuntu 24.04 LTS
Azure
Databases
DuckDB's columnar analytics engine running inside PostgreSQL, so analytical queries and data lake files are fast without moving the data anywhere.
Overview
pg_duckdb is the official PostgreSQL extension for DuckDB. It embeds DuckDB's vectorised, columnar execution engine directly into the PostgreSQL backend, so aggregations, window functions and large joins over ordinary PostgreSQL tables are executed by an analytics engine rather than the row at a time executor, with no change to the SQL and no export step. The same extension makes Parquet, CSV and JSON files queryable as tables, whether they sit on local disk or in object storage, so transactional data and data lake files can be joined in a single query.
Why the cloudimg image
cloudimg compiles pg_duckdb and its engine from the pinned upstream release onto a hardened, fully patched appliance, preloaded and proven working before the image is ever captured: the build gate asserts that a real query plan is handed to DuckDB rather than merely returning rows. Sample Parquet and CSV files ship with the image, so the first data lake query works on a brand new instance with no cloud account and nothing to configure. The database superuser and application role passwords are generated on first boot with no default login, the database listens only on loopback until those secrets exist, every remote connection is encrypted with a per instance TLS certificate, and 24/7 support is included.
Common uses
- Analytics on live transactional data
- Querying Parquet and CSV data lakes
- Mixed transactional and analytical workloads