Data Source

Overview

The data_source property specifies which configured data source a cube should use. This enables multi-database architectures where different cubes query different warehouses.

Example

cubes:
  - name: orders
    sql_table: public.orders
    data_source: default

  - name: analytics_events
    sql_table: events
    data_source: clickhouse_analytics

Syntax

Single Data Source

cubes:
  - name: users
    sql_table: users
    data_source: postgres_main

Default Behavior

If data_source is omitted, the cube uses the default data source:

cubes:
  - name: orders
    sql_table: orders
    # Uses "default" data source

Multi-Database Architecture

Different cubes can query different databases:

cubes:
  # Transactional data from Postgres
  - name: orders
    sql_table: public.orders
    data_source: postgres

  # Analytics events from ClickHouse
  - name: events
    sql_table: analytics.events
    data_source: clickhouse

  # ML features from Snowflake
  - name: predictions
    sql_table: ml.predictions
    data_source: snowflake

Data Source Configuration

Data sources are configured in your Bonnard project:

# .bon/datasources.yaml
datasources:
  - name: default
    type: postgres
    host: localhost
    database: mydb

  - name: analytics
    type: snowflake
    account: myaccount
    database: ANALYTICS

Cross-Database Joins

Cubes from different data sources cannot be directly joined. Use views or pre-aggregations to combine data from multiple sources.