Apache Superset — Open Source Data Visualization & Exploration

What Superset Does

• 40+ Chart Types: Bar, line, area, scatter, pie, heatmap, map, treemap, sunburst, pivot table, and more
• SQL Editor: Full SQL IDE with autocomplete, saved queries, and result visualization
• Dashboard Builder: Drag-and-drop dashboard creation with cross-filtering and drill-down
• Data Exploration: No-code exploration with visual query builder for slicing and dicing data
• 30+ Database Connectors: PostgreSQL, MySQL, ClickHouse, BigQuery, Snowflake, Redshift, Trino, Druid, etc.
• Access Control: Role-based access with row-level security and dataset permissions
• Caching: Query result caching with configurable TTL for dashboard performance
• Alerts & Reports: Scheduled reports and threshold-based alerts via email/Slack
• Embedding: Embed dashboards and charts into external applications

Architecture

┌──────────────┐     ┌──────────────┐     ┌──────────────┐
│  Browser     │────▶│  Superset    │────▶│  Your Data   │
│  Dashboards  │     │  Server      │     │  PostgreSQL  │
└──────────────┘     │  (Python/    │     │  ClickHouse  │
                     │   Flask)     │     │  BigQuery    │
                     └──────┬───────┘     │  Snowflake   │
                            │             │  30+ more    │
                  ┌─────────┼─────────┐   └──────────────┘
                  │         │         │
           ┌──────┴──┐ ┌───┴───┐ ┌───┴───┐
           │  Redis  │ │Celery │ │Metadata│
           │ (Cache) │ │(Async)│ │  DB    │
           └─────────┘ └───────┘ └────────┘

Self-Hosting

Docker Compose

services:
  superset:
    image: apache/superset:latest
    ports:
      - "8088:8088"
    environment:
      SUPERSET_SECRET_KEY: your-long-secret-key
      DATABASE_URL: postgresql+psycopg2://superset:superset@db:5432/superset
      REDIS_URL: redis://redis:6379/0
    depends_on:
      - db
      - redis

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

  redis:
    image: redis:7-alpine

volumes:
  pg-data:

Key Features

No-Code Exploration

1. Select dataset (table/view)
2. Choose chart type (bar, line, etc.)
3. Drag metrics (SUM, COUNT, AVG)
4. Add dimensions (group by)
5. Apply filters
6. Customize colors, labels, formatting
7. Save to dashboard

SQL Lab (SQL IDE)

-- Full SQL editor with schema browser
SELECT
  DATE_TRUNC('month', order_date) as month,
  product_category,
  SUM(revenue) as total_revenue,
  COUNT(DISTINCT customer_id) as unique_customers
FROM orders
WHERE order_date >= '2024-01-01'
GROUP BY 1, 2
ORDER BY 1, 3 DESC

-- Results can be instantly visualized as any chart type
-- Save queries for reuse
-- Share with team members

Dashboard Features

• Cross-filtering: Click on one chart to filter all others
• Filter bar: Global filters affecting all charts
• Drill-down: Click data points to explore deeper
• Annotations: Mark events on time-series charts
• CSS customization: Custom styling per dashboard
• Auto-refresh: Configurable refresh intervals

Jinja Templating

-- Dynamic queries with user context
SELECT * FROM orders
WHERE region = '{{ current_user().region }}'
  AND order_date >= '{{ from_dttm }}'
  AND order_date <= '{{ to_dttm }}'

{% if filter_values('status') %}
  AND status IN ({{ "'" + "', '".join(filter_values('status')) + "'" }})
{% endif %}

Superset vs Alternatives

常见问题

Q: Superset 和 Metabase 怎么选？ A: Metabase 更适合非技术用户（可视化查询构建器更友好）。Superset 更适合数据团队（SQL Lab 更强大、图表类型更多、性能更好）。如果团队习惯写 SQL，选 Superset。

Q: 可以处理大数据吗？ A: Superset 本身不存储数据，性能取决于底层数据库。搭配 ClickHouse、Druid 或 Trino 等 OLAP 引擎可以查询数十亿行数据。Superset 的缓存层（Redis）也能显著提升仪表盘加载速度。

Q: 自托管难度大吗？ A: Docker 部署相对简单。生产环境需要配置 Redis（缓存）、Celery（异步任务）和 PostgreSQL（元数据）。Apache 官方提供详细的生产部署指南和 Helm chart。

来源与致谢

• GitHub: apache/superset — 72.3K+ ⭐ | Apache-2.0
• 官网: superset.apache.org