Why does it need a second Postgres database?

Metabase stores its own state — users, dashboards, saved questions, permissions — in an application database. By default that is an embedded H2 file, which is fine for a quick trial but not for production. This deployment sets `MB_DB_TYPE=postgres` and ships a companion `postgres:16` container so your Metabase configuration is durable and backup-friendly. This metadata Postgres is separate from the database you analyse.

How much RAM does Metabase need?

Metabase runs on the JVM, so it is heavier than a typical Go or Rust app. Upstream's minimum is 1 CPU / 1 GB, but that leaves little headroom once dashboards load — a 2 GB VPS is the realistic floor and 4 GB is comfortable for a small team, adding roughly 2 GB per 20 concurrent users. You can cap the heap with `JAVA_OPTS=-Xmx1g` on a 2 GB box.

Which databases can Metabase connect to?

Out of the box Metabase supports PostgreSQL, MySQL/MariaDB, SQL Server, SQLite, ClickHouse, BigQuery, Redshift, Snowflake, Presto/Trino and more via official drivers. On a VPS the common pattern is to point it at your own application's PostgreSQL or MySQL — ideally a read replica — over a private Docker network so analytics load never touches your production primary.

Is Metabase really free and open source?

The Metabase Open Source Edition is licensed under the AGPL and is fully self-hostable with no per-seat fee — that is the edition this template deploys. Metabase also sells paid Pro/Enterprise editions (advanced permissions, SSO, extra embedding) under a commercial license, but you never need those to run dashboards on your own VPS.

Do I need a domain to run Metabase?

No. Metabase answers on `http:// :3000` immediately after setup. A domain is only needed if you want HTTPS and a clean URL — put a Caddy or Nginx reverse proxy in front, point a DNS A record at your VPS, and restrict port 3000 to the proxy so the app is never exposed unencrypted.

Disposer rapidement d'une plateforme de données.

Metabase

Open-source business intelligence on top of any SQL database — dashboards and charts in the browser, no SQL required.

2 Go (4 Go recommandé) RAM 1 vCPU (2 recommandé) Port 3000 Disponible

Code source

Stack technique

DockerJava/JVMPostgreSQL

RAM minimum2 Go (4 Go recommandé)

CPU minimum1 vCPU (2 recommandé)

Port par défaut3000

OS compatiblesubuntu-24.04

Metabase is the easy-to-use open-source Business Intelligence tool (AGPL, ~46k GitHub stars). Point it at a database you already run — PostgreSQL, MySQL/MariaDB, SQL Server, ClickHouse, BigQuery and dozens more — and non-technical users build questions, charts and interactive dashboards through a visual query builder, while analysts drop into raw SQL when they need to. Results can be scheduled to email or Slack, embedded in your own app, or shared behind granular row-level permissions.

Deployed on a ServOrbit VPS, Metabase stops your BI dashboards from living on a foreign SaaS you pay per-seat in USD. It runs as two containers — Metabase itself plus a companion Postgres that holds its own metadata (questions, dashboards, users) — behind your domain, on infrastructure you control, at a fixed monthly cost. Connect it to your production database over a private network or a read replica and your entire analytics layer stays on your own server, with your data never leaving Morocco.

Fonctionnalités clés

Visual query builder — non-technical users build questions and charts by clicking, with a native SQL editor available when analysts need full control.

Interactive dashboards with filters, drill-through and cross-filtering, refreshed on a schedule or in real time.

Connects to 20+ databases — PostgreSQL, MySQL/MariaDB, SQL Server, ClickHouse, SQLite, BigQuery, Redshift, Snowflake and more.

Scheduled alerts and subscriptions — deliver dashboards or 'when a number crosses a threshold' alerts to email or Slack.

Granular permissions — group- and collection-level access plus row/column controls so each team sees only its own data.

AGPL open source — self-hostable with no per-seat SaaS fee, its own metadata stored in a Postgres database you own.

Quand utiliser cette solution ?

1

Self-hosted product & revenue dashboards

Connect Metabase to your application's production database (or a read replica) and give the whole team live dashboards for signups, revenue, churn and usage — without shipping a byte of customer data to a third-party SaaS or paying per seat in USD.

2

Ad-hoc analytics for non-technical teams

Marketing, ops and finance answer their own questions through the visual query builder instead of queuing tickets with the data team. Save frequent questions into shared collections so the whole company works from the same numbers.

3

Embedded analytics in your own product

Use Metabase's embedding to drop signed, filtered dashboards straight into your customer-facing app or internal admin — self-hosted BI as a feature of your product, running on the same VPS, with no external analytics vendor in the loop.

Déployer Metabase sur votre VPS

Guide optimisé pour les VPS Cloud ServOrbit.

01

Order a VPS

Pick a ServOrbit VPS with at least 2 GB RAM (Metabase runs on the JVM; 4 GB is comfortable for a small team). Ubuntu 22.04, Docker pre-installed by the provisioner. Metabase and its metadata Postgres deploy together as a two-service Docker Compose stack.

02

Deploy the stack

The provisioner brings up two containers: metabase/metabase on port 3000 and a companion postgres:16 that holds Metabase's own application data (questions, dashboards, users). Metabase is wired to Postgres via MB_DB_TYPE=postgres and the MB_DB_* connection variables — no external database to provision first.

03

Finish the setup wizard

Open http://:3000 and complete the one-time setup: create the admin account, then connect the database you actually want to analyse (your app's PostgreSQL/MySQL, a read replica, etc.). Metabase auto-scans the schema so tables and columns are query-ready in seconds.

04

Build your first dashboard

Use the visual query builder to create a few questions (a count over time, a breakdown by category), save them into a collection, and pin them onto a dashboard with filters. Analysts can switch any question to the native SQL editor for full control.

05

Add HTTPS and lock it down

Put Metabase behind a Caddy or Nginx reverse proxy for a real domain and a Let's Encrypt certificate (bi.yourdomain.com { reverse_proxy localhost:3000 }). Restrict port 3000 with ufw so only the proxy reaches it, and set a strong admin password — Metabase exposes all connected data to anyone who can log in.

Questions fréquentes

Metabase is an open-source Business Intelligence tool. You connect it to a SQL database and build charts, questions and dashboards through a point-and-click interface — no SQL required for everyday use, with a full SQL editor available for analysts. It is aimed at teams who want self-service analytics without a per-seat SaaS BI bill.

Solutions complémentaires

Bases de Données

NocoDB

Open-source Airtable alternative — turn any database into a smart spreadsheet with grid, kanban, gallery and form views, REST and GraphQL APIs and built-in automations. Self-hosted, no per-seat pricing.

Intelligence Artificielle

Qdrant

Self-hosted vector database for semantic search and RAG. Store, index and query high-dimensional embeddings at sub-millisecond latency — the missing infrastructure layer for your LLM stack.