starlake-ai

starlake-ai / quack-on-demand

Public

Production-grade Arrow FlightSQL gateway in front of DuckDB Quack + DuckLake. Multi-tenant pools, pluggable auth (DB/JWT/OIDC), table-level ACLs, role-aware routing, and a live admin console

starlake.aiquack-on-demand
11
1
89% credibility
Found May 30, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Scala
AI Summary

Quack on Demand is a production-ready gateway that sits in front of DuckDB and DuckLake, letting teams share a single database over a standard SQL connection. It handles authentication, access control at the table level, and automatically manages multiple query nodes so one person running a heavy query doesn't slow everyone else down. The project includes a React admin panel for managing tenants, pools, and permissions, plus a live dashboard showing cluster health and query performance. It supports connecting existing identity systems (Google, Azure, Keycloak, JWT) so teams can use their normal logins, and stores permission rules in a simple relational table that updates in real time. The whole system runs as a single self-contained service that can be deployed via Docker or as a standalone JAR.

How It Works

1
🔍 Discover a smarter database gateway

You hear about Quack on Demand as a way to share one DuckDB database with your whole team without everyone needing their own copy installed.

2
🚀 Launch with one click

You spin up the ready-made stack using Docker, and within seconds your admin dashboard is live at localhost with your database connected.

3
🔐 Set up who can see what

You open the admin panel, create tenants, and write simple rules like 'analysts can read the sales data' without touching any code.

4
📊 Connect any analytics tool

You open DBeaver or your Python notebook, point it at the gateway, and run queries just like you would against any database — no special drivers needed.

5
Choose your path
🔓
Quick start mode

Leave access open for a sandbox environment where anyone on your network can explore the data

🔒
Secure enterprise mode

Connect your company's identity provider so logins go through your existing team accounts and permissions

6
📈 Watch your cluster in real time

The live dashboard shows you which queries are running, how fast each node responds, and alerts you if something needs attention.

🎉 Your team queries live data together

Everyone connects through the same gateway, sees the data they should see, and gets fast results — all from one shared database that stays in sync.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is quack-on-demand?

Quack on Demand is a production-grade gateway that sits in front of DuckDB's new Quack client-server protocol, adding the infrastructure that serious deployments need. Built in Scala using Apache Arrow Flight SQL, it exposes a standard JDBC/ODBC/ADBC interface while managing multi-tenant pools of DuckDB workers, enforcing table-level ACLs, and handling authentication through Postgres, JWT, or OIDC providers. A React admin console gives you live metrics, statement history, and an ACL editor. You deploy it as a single uber-jar that shares Postgres with DuckLake for catalog metadata.

Why is it gaining traction?

DuckDB's Quack extension is a game-changer for turning DuckDB into a shared server, but it ships with minimal auth and no multi-tenancy. This project fills that gap with role-aware routing (READONLY/WRITEONLY/DUAL nodes), per-statement ACL enforcement via SQL parsing, and support for enterprise identity providers like Keycloak and Azure AD. The self-healing architecture (dead Quack nodes respawn automatically) and comprehensive Docker setup make it viable for actual production use rather than just demos.

Who should use this?

Data teams running DuckDB as a shared query engine who need multi-tenant isolation, fine-grained access control, and a management UI. Backend developers building BI tools or data APIs that want Arrow Flight SQL compatibility with DuckDB under the hood. Teams already using Postgres who want to layer DuckLake on top without abandoning their existing metastore.

Verdict

The feature set is impressive and the architecture is sound, but the credibility score of 0.8999999761581421% and low star count (11) signal a young project. Documentation is thorough and the test suite claims 574+ tests, but real-world production hardening is unproven. Worth evaluating for non-critical workloads or as a reference implementation; hold off on mission-critical systems until the community grows.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.