# Metabase — Open Source Business Intelligence & Analytics

> Metabase lets everyone in your organization ask questions and learn from data. Connect any database, build dashboards, and share insights — no SQL knowledge required.

## Install

Save in your project root:

## Quick Use

```bash
docker run -d --name metabase -p 3000:3000 metabase/metabase:latest
```

Open `http://localhost:3000` — connect your database and start building dashboards.

## Intro

**Metabase** is the easiest way to let everyone in your company work with data. It provides a beautiful web interface where anyone can ask questions about your database data, build interactive dashboards, and share insights — without writing SQL (though SQL is supported for power users).

With 46.8K+ GitHub stars, Metabase is the most popular open-source BI tool, used by over 60,000 organizations worldwide from startups to Fortune 500 companies.

## What Metabase Does

- **Visual Query Builder**: Click-based interface to ask questions without SQL
- **SQL Editor**: Full SQL editor with autocomplete for advanced users
- **Dashboards**: Interactive dashboards with filters, drill-down, and auto-refresh
- **Charts**: 15+ visualization types (bar, line, pie, map, funnel, pivot table, etc.)
- **Alerts**: Automated alerts when metrics cross thresholds
- **Embedding**: Embed charts and dashboards into your own application
- **Data Modeling**: Define metrics, segments, and relationships in a metadata layer
- **Permissions**: Granular access control by group, database, table, and column
- **Collections**: Organize dashboards and questions into shareable collections

## Architecture

```
┌──────────────┐     ┌──────────────┐     ┌──────────────┐
│  Browser     │────▶│  Metabase    │────▶│  Your Data   │
│  (Dashboard) │     │  Server      │     │  PostgreSQL  │
└──────────────┘     │  (Java/Clj)  │     │  MySQL       │
                     └──────────────┘     │  BigQuery    │
                                          │  Snowflake   │
                                          │  Redshift    │
                                          │  MongoDB     │
                                          └──────────────┘
```

## Self-Hosting

### Docker Compose (Production)

```yaml
services:
  metabase:
    image: metabase/metabase:latest
    ports:
      - "3000:3000"
    environment:
      MB_DB_TYPE: postgres
      MB_DB_HOST: metabase-db
      MB_DB_PORT: 5432
      MB_DB_DBNAME: metabase
      MB_DB_USER: metabase
      MB_DB_PASS: metabase
    depends_on:
      - metabase-db

  metabase-db:
    image: postgres:16-alpine
    environment:
      POSTGRES_USER: metabase
      POSTGRES_PASSWORD: metabase
      POSTGRES_DB: metabase
    volumes:
      - mb-db:/var/lib/postgresql/data

volumes:
  mb-db:
```

## Supported Data Sources

| Category | Databases |
|----------|-----------|
| SQL | PostgreSQL, MySQL, MariaDB, SQL Server, SQLite |
| Cloud DW | BigQuery, Snowflake, Redshift, Databricks |
| NoSQL | MongoDB |
| Analytics | ClickHouse, Druid, Presto/Trino, Spark SQL |
| Flat Files | CSV uploads |
| APIs | Google Analytics, Stitch, custom drivers |

## Key Features

### Visual Query Builder

No SQL required:
```
Table: Orders
├── Filter: Created At is in the last 30 days
├── Summarize: Sum of Total, grouped by Product Category
├── Sort: Total descending
└── Visualization: Bar Chart
```

This generates SQL behind the scenes:
```sql
SELECT product_category, SUM(total)
FROM orders
WHERE created_at >= CURRENT_DATE - INTERVAL '30 days'
GROUP BY product_category
ORDER BY SUM(total) DESC
```

### Dashboard Filters

```
Dashboard: Sales Overview
├── Date Filter: Last 30 days / This quarter / Custom range
├── Region Filter: Dropdown (North, South, East, West)
├── Product Filter: Search + multi-select
└── All charts update based on filter selections
```

### Embedding

Embed Metabase in your own app:

```javascript
// iframe embedding
const iframeUrl = "http://metabase:3000/embed/dashboard/eyJ..."

// JWT-signed embedding (secure)
const jwt = require('jsonwebtoken');
const METABASE_SECRET = "your-secret-key";

const payload = {
  resource: { dashboard: 1 },
  params: { customer_id: currentUser.id },
  exp: Math.round(Date.now() / 1000) + (10 * 60),
};
const token = jwt.sign(payload, METABASE_SECRET);
const embedUrl = `http://metabase:3000/embed/dashboard/${token}`;
```

### Alerts & Subscriptions

```
Alert: When daily revenue drops below $10,000
  → Send Slack notification to #revenue channel
  → Email CFO

Subscription: Weekly sales report
  → Every Monday 8am
  → Email to sales-team@company.com
  → Include dashboard as PDF attachment
```

## Metabase vs Alternatives

| Feature | Metabase | Redash | Grafana | Looker | Tableau |
|---------|----------|--------|---------|--------|---------|
| Open Source | Yes | Yes (BSD) | Yes (AGPL) | No | No |
| No-SQL queries | Visual builder | SQL only | PromQL | LookML | Drag & drop |
| Ease of use | Very easy | Moderate | Moderate | Complex | Moderate |
| Embedding | Yes | Limited | Yes | Yes | Yes |
| Self-hosted | Yes | Yes | Yes | No | Enterprise |
| Pricing | Free (OSS) | Free | Free | $5K+/mo | $70/user/mo |

## 常见问题

**Q: Metabase 适合什么规模的团队？**
A: 从个人到数千人的组织都适用。开源版没有用户数限制。对于大型企业，Metabase Pro/Enterprise 提供缓存、审计、SAML SSO 等高级功能。

**Q: 查询会影响生产数据库性能吗？**
A: Metabase 建议连接只读副本（read replica）而非直接查询生产库。也可以配置查询超时和并发限制来保护数据库。

**Q: 可以连接多个数据库吗？**
A: 可以。一个 Metabase 实例可以连接多个不同类型的数据库，在同一个仪表盘中展示来自不同数据源的数据。

## 来源与致谢

- GitHub: [metabase/metabase](https://github.com/metabase/metabase) — 46.8K+ ⭐
- 官网: [metabase.com](https://metabase.com)

---
Source: https://tokrepo.com/en/workflows/a1cd7314-34d0-11f1-9bc6-00163e2b0d79
Author: AI Open Source