Self-Hosted Redash: SQL Dashboards Your Whole Team Can Share

Your data lives in Postgres. Your decisions live in Slack screenshots of a SQL client. Every Monday someone asks for "the dashboard version" and three engineers export CSVs by hand.

Redash is the open-source BI tool that puts SQL first: ~29k GitHub stars, BSD-2-Clause, write a query, pick a chart, pin it to a dashboard, schedule refreshes, and share a link the whole company can use. It is older than the shiny agent-analytics wave and still what many teams mean when they say "data driven" without buying a warehouse seat for every intern.

What it actually does

Redash connects to 35+ data sources — Postgres, MySQL, BigQuery, Snowflake, ClickHouse, MongoDB, Elasticsearch, Prometheus, and a long tail of warehouses and SaaS APIs. You compose SQL (or NoSQL where supported) in a browser editor with schema browsing and autocomplete, then visualize results without exporting to Excel.

What teams use daily:

• Queries you can share — peer review SQL, fork a colleague's report, document the metric once.
• Dashboards — drag visualizations together, refresh on a schedule, embed or link internally.
• Alerts — notify when a query crosses a threshold (revenue drop, queue depth, failed jobs).
• REST API — automate everything the UI does; handy for CI or internal portals.
• Parameterised queries — one dashboard, many filters, less copy-paste SQL.

Compared to visual-only BI, Redash assumes someone on the team speaks SQL. Compared to Matomo-style web analytics, it is for your databases, not page-view tags. We covered Metabase recently; Redash leans harder into shared queries and a classic data-team workflow.

Why self-host?

Queries touch production data. Dashboards often run against replicas, but credentials and result sets still matter. Hosting Redash on a Canadian VPS or private network keeps SQL traffic and cached results inside infrastructure you control — relevant for PIPEDA and client contracts.

Predictable cost. Redash Cloud existed; many orgs still run the open stack on Docker to avoid per-user BI pricing. You pay RAM, workers, and DBA time.

Integration freedom. Need a weird ODBC source or an internal API? Redash's query runner model and Python query sources extend farther than most cookie-cutter SaaS tiers.

BSD license. Permissive for internal forks and custom connectors without AGPL contagion worries.

What running it takes

Upstream recommends at least 4 GB RAM for a basic Docker Compose deploy: Redash web/API, background workers, PostgreSQL (app metadata), and Redis (queues). Set REDASH_SECRET_KEY and REDASH_COOKIE_SECRET in a private .env; enable HTTPS in production.

Operational checklist:

• Mail — configure SMTP for invites, password resets, and alert emails; set REDASH_HOST to your public URL.
• Workers — heavy dashboards need enough query runners; scale workers before blaming Postgres.
• Read replicas — point analytics queries at replicas so marketing filters do not lock production rows.
• Upgrades — follow Redash upgrade docs; back up the metadata database before jumping versions.

Official AMIs exist; on Swift Host you typically use Docker Compose on Ubuntu, TLS at nginx, and firewall rules that only expose 443.

Who it's for (and who should skip it)

Good fit: startups with one Postgres app DB, agencies reporting across client schemas (strict permissions per data source), ops teams alerting on Prometheus + SQL joins, anyone who wants a shared query library.

Maybe skip it: if nobody writes SQL and nobody will, a visual explorer (Metabase) may stick better. If you need embedded product analytics or session replay, use PostHog — Redash will not autocapture clicks. If you refuse to maintain Docker and email, managed BI may sleep better.

Hosting it in Canada

Redash pairs naturally with VPS and Docker hosting beside your app: private hostname, nightly Postgres backups for Redash's own DB, outbound access only to databases you whitelist.

Tell us which data sources you query, dashboard count, and alert volume if you want a worker sizing note before you point finance at a public URL.

Want dashboards that start with honest SQL? Talk with Swift Host — we are in St. Albert and actually answer the phone.