An honest account of where this sits, what it does better than the alternatives, and what the alternatives still do better. If you only need one of the things below, one of those other tools is probably the right answer.
Self-hosted with no seat count, no viewer tier and no license server to call home to. The tools that charge for viewers get expensive precisely when your work succeeds and more people start looking at it.
One published comparison puts a 10-editor, 10-viewer
setup on a notebook-BI platform at $27,000 a year.
SQLDesk: the cost of the box you run it on.
SQLDesk forked Redash, which has been in maintenance mode for years — dependency bumps and security patches rather than new work. That makes the fork additive rather than divergent: there is no fast-moving upstream to fight, and the features people have been asking for are simply unbuilt.
Redash upstream: roughly one release a year since 2022,
community forum read-only, ~7 volunteer maintainers.
Not a policy — a removal. The upstream shipped hard-coded endpoints that POSTed usage statistics, and separately your admin's name and email, to servers you don't control. Both are gone. Version checking exists but is off until you point it at an endpoint of your own.
Removed: version.redash.io/api/report (usage data)
Removed: version.redash.io/subscribe (admin name + email)
Everything in the query path gets measured before and after, on real data volumes, and the figures are published. No "blazing fast" — just what it was, what it is, and how it was measured.
Cached 27MB result: 547ms → 99ms
Cache lookup on a year of history: 66.7ms → 0.064ms
See the full workings →
A permissive licence with an explicit patent grant, which BSD 2-Clause does not provide. The inherited code keeps its original copyright and licence text, reproduced in full — because the alternative is a licence violation dressed up as a rebrand.
LICENSE → Apache 2.0 · LICENSE.redash → BSD 2-Clause, verbatim
NOTICE → attribution, and an explicit "not affiliated" statement
SQL, dashboards and alerts today; notebooks and Kafka streams next — under one permission model, one login and one deployment. The value isn't any single surface, it's that a note can read a saved query, join it against a stream, and publish the result to a dashboard without leaving the product or re-declaring who may see what.
The open-source, self-hosted corner that covers all three surfaces is genuinely thin. That's the gap SQLDesk is built for.
| Tool | Dashboards | Notebooks | Streaming | Self-hosted | Shape |
|---|---|---|---|---|---|
| SQLDesk | ✓ | planned | planned | ✓ | SQL-first, open, unlimited viewers |
| Metabase | ✓ | — | — | ✓ | Easiest for non-technical users |
| Apache Superset | ✓ | — | — | ✓ | Most powerful for SQL teams |
| Redash | ✓ | — | — | ✓ | Maintenance mode |
| Briefer | ✓ | ✓ | — | ✓ | Notebooks + dashboards, multiplayer |
| Hex / Deepnote | ✓ | ✓ | — | — | SaaS, per-seat including viewers |
| Timeplus | ✓ | — | ✓ | ✓ | Streaming-first, ClickHouse storage |
| Tinybird | — | — | ✓ | — | Managed real-time APIs |
| Databricks | ✓ | ✓ | ✓ | — | All three, enterprise-priced |
Databricks acquired Redash in 2020. SQLDesk is building, from that lineage, toward a
subset of what Databricks does — with one docker compose up, no lakehouse
commitment and no per-seat bill.
Breadth is the bet, and breadth has a cost. If one of these is your whole problem, use the specialist.
Open-source notebooks and dashboards with real-time multiplayer editing, shipping now. SQLDesk's Notes are still ahead of us on the roadmap.
A purpose-built streaming engine with push-based dashboards over SSE. SQLDesk starts with bounded reads from Kafka, which is a smaller claim honestly made.
The best point-and-click question builder in open source. SQLDesk assumes somebody on the team writes SQL.
If your dashboards are about infrastructure and alerting on metrics, that is what Grafana was built for and SQLDesk was not.