article --summary
The summary
Older database protocols were designed before big data and tend to send results row by row, repeating column names and types with every row, which becomes costly when moving large result sets. DuckDB's answer is Quack, an HTTP-based client/server protocol that was previewed in May and reached its first production release in October as part of DuckDB 2.0.0 LTS.
Quack sends data in columnar chunks, serializes values efficiently because DuckDB is a vectorized engine, and streams results as the engine emits them, so clients can start working before a query finishes. Developers get features such as the CONNECT statement for querying remote servers, -serve and -connect command-line options, and concurrent multi-client writes under MVCC isolation.
According to DuckLabs' Guillermo Sánchez Dionis, the main benefit is a very small memory footprint and minimal server overhead, with more stability improvements expected in upcoming 2.0 patch releases.
Editorial note: This is a summary of an article first published by The New Stack. The full piece is on their site.