Thanks for your interest! TL;DR: pg_duckdb is an extension that runs on the same server as Postgres, while MyDuck operates as a standalone replica that can run

Thanks for your interest! TL;DR: pg_duckdb is an extension that runs on the same server as Postgres, while MyDuck operates as a standalone replica that can run on a separate server. Details: pg_duckdb uses DuckDB as an execution engine while keeping the data stored in Postgres in its usual row-oriented format. It identifies queries that can be optimized using DuckDB and executes them there. This is achieved by registering a replacement scan in DuckDB, allowing it to directly scan Postgres data, converting it into an analytics-optimized columnar format on the fly. MyDuck, on the other hand, functions as a standby server replicating changes from a primary Postgres server. The replicated data is stored in DuckDB’s native columnar format. This approach leverages both DuckDB’s optimized columnar storage and its execution engine, generally resulting in faster query performance compared to scanning Postgres data directly. Additionally, pg_duckdb is currently limited to full table scans — it doesn’t yet support skipping unnecessary data even with restrictive WHERE conditions (though there are plans to add index-based scans: https://github.com/duckdb/pg_duckdb/issues/243 https://github.com/duckdb/pg_duckdb/issues/243). In contrast, DuckDB’s native storage, used by MyDuck, includes extensive support for data skipping (a.k.a. filter pushdown). Hope this clarifies the differences! Feel free to reach out with any more questions.