Apache Superset для аналитика

Зачем это знать

Superset — открытый BI из Apache Foundation (создан в Airbnb). Мощнее Metabase, дешевле Tableau. Много russian компаний (Yandex, Ozon) используют. В enterprise и scale-ups часто встречается.

Что такое Superset

Apache Superset — open-source BI для data exploration и visualization.

• Create dashboards
• SQL lab (query editor)
• Rich chart types (40+)
• Alerting, scheduling
• Embed в приложения

Originally built by Maxime Beauchemin в Airbnb (также создатель Airflow).

Архитектура

• Frontend: React
• Backend: Python (Flask)
• DB support: любая через SQLAlchemy
• Caching: Redis

Плюсы

• Огромный chart library
• SQL Lab — powerful query editor
• Dashboards интерактивные
• Self-hosted free
• Active dev (Apache)

Минусы

• Complex setup для production
• Steeper learning curve vs Metabase
• Maintenance overhead для self-host

Setup

# Через Docker
docker pull apache/superset
docker run -d -p 8088:8088 apache/superset
# Создать admin
docker exec -it superset superset fab create-admin

Официальный docker-compose на github.

Основные концепции

Databases

Connect к data source — Postgres, BigQuery, Snowflake, ClickHouse, etc.

Datasets

Definition таблиц / views + columns + metrics (measures).

Charts

Individual visualization. 40+ types:

• Line, bar, area
• Pivot
• Heatmap
• Sankey (flow)
• Country map
• Custom JS

Dashboards

Collection charts + filters + markdown.

SQL Lab

Interactive SQL editor:

• Multiple query tabs
• Results history
• Save query to dashboard

Pre-aggregated metrics

В Dataset definition:

metric: daily_orders_count
expression: COUNT(DISTINCT order_id)

Reuse across charts.

Filters

Dashboard-level filters applied к все charts sharing column.

Cross-filtering — click на chart фильтрует другие.

Alerts и reports

Schedule reports:

• Send to email
• Attach chart screenshots
• CSV exports

Alerting — if metric crosses threshold, notify.

Access control

• Roles (Admin, Gamma, Alpha)
• Database-level permissions
• Row-level security (RLS) — enterprise

vs Metabase

Superset

• More chart types
• Enterprise features
• Larger scale

Metabase

• Easier setup / UX
• Smaller teams
• Quick adoption

Superset — closer к Tableau в capability, но open-source.

vs Tableau

Superset

• Free
• Self-hosted
• Integration с modern data stack

Tableau

• Richer UX
• Expensive
• More polished

Use cases

Internal BI

Dashboards для teams, business.

Data exploration

SQL Lab — powerful для analysts.

Embedded analytics

Superset charts в customer-facing products.

Cross-DB

Single tool для multiple data sources.

Частые задачи

DAU chart

SQL Lab:

SELECT day, COUNT(DISTINCT user_id) FROM events
GROUP BY 1 ORDER BY 1;

Save as chart → line visualization.

Cohort retention

Dataset with cohort_month, days_since_signup, retention_rate. Heatmap visualization.

Funnel

Conditional aggregation + funnel chart type.

Интеграция с другими tools

Airflow

Schedule refresh datasets via Airflow.

dbt

dbt models exposed как Superset datasets.

Jupyter

Embed SQL Lab queries or export to notebook.

Russian context

• Yandex использует Superset (внутри).
• Ozon — internal BI часто Superset.
• Русскоязычные tutorials ограничены — в основном английские.

Performance

Caching

Redis-backed для query results.

Dashboards load

Cache expiration configurable. Balance freshness vs speed.

Queries

Superset — thin layer. Performance зависит от underlying DB.

На собесе

«Superset experience?» Какие dashboards, charts, SQL Lab.

«Superset vs Metabase?» Superset enterprise-grade, Metabase simpler/easier.

«Self-host complexity?» Docker simple для dev. Production — нужны DevOps.

Связанные темы

• Что такое Superset BI
• Что такое Metabase
• Metabase для аналитика
• Tableau vs Metabase
• Looker и LookML

FAQ

Бесплатный full-features?

Yes. Open-source Apache.

Production-ready?

С правильным deployment — yes.

SaaS alternative?

Preset (managed Superset by co-founder).

---

Тренируйте BI — откройте тренажёр с 1500+ вопросами для собесов.