The extension currently relies on the parquet_fdw extension to query the parquet files.
If DuckDB implements it's own Foreign Data Wrapper for Postgres then querying would be greatly sped up compared to parquet_fdw since DuckDB optimiser is known for better pipelining / vectorisation of queries.
I was wondering if this is something we could collaborate on and package the two extensions as a complete analytics package for Postgres instances. I would be happy to contribute to the creation of this FDW if you can share some pointers.
Thanks for reaching out! What you are describing sounds instead quite similar to pg_duckdb, which we have been working on in collaboration with a few other parties. Feel free to send a PR there if you are interested in contributing!
I'm the author of the pg_analytica extension that enables periodic parquet exports for Postgres tables. See: https://github.com/sushrut141/pg_analytica
The extension currently relies on the parquet_fdw extension to query the parquet files. If DuckDB implements it's own Foreign Data Wrapper for Postgres then querying would be greatly sped up compared to parquet_fdw since DuckDB optimiser is known for better pipelining / vectorisation of queries.
I was wondering if this is something we could collaborate on and package the two extensions as a complete analytics package for Postgres instances. I would be happy to contribute to the creation of this FDW if you can share some pointers.
Thanks