What is the DuckDB Quack protocol, and how does it change DuckDB's deployment model?

Quack is a client-server protocol, introduced as part of a broader 2026 effort to make DuckDB deployable as a real, standalone network-accessible service, complete with a production-grade OAuth/OIDC authentication layer, rather than remaining exclusively an in-process, embedded library.

This represents a genuine expansion of DuckDB's deployment model, historically defined almost entirely by its embedded, single-process design: with Quack, a DuckDB instance can run as a persistent, shared service that multiple separate client applications or users connect to over a network, authenticate against, and issue queries to, much closer to how a traditional client-server database like PostgreSQL is deployed and consumed.

This doesn't replace DuckDB's embedded mode, which remains the default and most common way DuckDB is used, but it opens up new deployment patterns that weren't previously well supported: a centrally-managed, shared analytical DuckDB service that a team or organization can connect to from multiple applications, without each one needing to embed and manage its own separate DuckDB instance and data. Given how recently this protocol emerged, it's worth treating it as an actively evolving part of the ecosystem rather than a long-settled feature.