Datamarts

​

What is a Datamart?

​

Why Use Datamarts?

​

Creating a Datamart

​

Key Features

​

Tables

• Choose tables from available schemas
• Preview table data before including
• Select multiple tables from different schemas
• Hide tables that aren’t needed

​

Columns

• Select specific columns from each table
• Hide unnecessary columns
• Preview column data
• Configure column visibility

​

Tenancy Configuration

• Table-level tenancy with primary key and client columns
• Database-level tenancy for complete isolation
• Multi-database tenancy (MSSQL only)

​

Relationships

• One-to-many relationships
• Many-to-many relationships
• Self-referential joins
• Configure join types and cardinality

​

Metadata & Descriptions

• Table descriptions
• Column descriptions and labels
• Business-friendly aliases
• Example questions for downstream AI experiences

​

Datamart Structure

• Select tables from multiple schemas
• Configure columns from selected tables
• Define relationships between tables
• Set up multi-tenancy at table, database, or multi-database level

​

Tenancy & Multi-Tenancy

​

Tenancy Levels

• Select a table with primary key and client name columns
• Data is filtered automatically based on client ID
• Suitable for single-database multi-tenant setups

• Use separate databases per tenant
• Each tenant’s data is completely isolated
• Automatically routes queries to the correct database

• Use multiple databases for tenant isolation
• Supports complex multi-tenant architectures
• Configure schema lists per tenant

​

Access Control

• View, Create, Edit, Delete permissions
• Edit Tenancy permission for configuring multi-tenancy

​

Use Cases

​

AI-Powered Analytics

• “Top customers by revenue”
• “Sales trends last quarter”
• “Compare regions”

​

Consistent Metrics

• Standard KPI calculations
• Consistent aggregations
• Unified dimensions

​

Fast Dashboards

• Selected tables and columns for faster queries
• Clear relationships between tables
• Multi-tenant data isolation

​

Multi-Tenant Applications

• Table-level tenancy for single-database setups
• Database-level tenancy for complete isolation
• Multi-database tenancy for complex architectures

​

Best Practices

​

Datamart vs. Data Warehouse

​

Datamart Features

​

Table & Column Management

• Select tables from schemas within your datasource
• Configure which columns are visible
• Preview table data before including
• Hide tables or columns as needed

​

Relationships

• Define relationships between tables in your datamart
• Configure cardinality and join types

​

Metadata & Descriptions

• Table descriptions
• Column descriptions and labels
• Business-friendly aliases

​

Datamart API

​

List Datamarts

curl --request GET \
  --url https://api.usedatabrain.com/api/v2/datamarts \
  --header 'Authorization: Bearer YOUR_API_TOKEN'

​

Create Datamart

curl --request POST \
  --url https://api.usedatabrain.com/api/v2/datamarts \
  --header 'Authorization: Bearer YOUR_API_TOKEN' \
  --header 'Content-Type: application/json' \
  --data '{
    "name": "Sales Datamart",
    "datasourceId": "your-datasource-id"
  }'

​

Update Datamart

curl --request PUT \
  --url https://api.usedatabrain.com/api/v2/datamarts \
  --header 'Authorization: Bearer YOUR_API_TOKEN' \
  --header 'Content-Type: application/json' \
  --data '{
    "id": "your-datamart-id",
    "name": "Updated Name"
  }'

​

Semantic Layer & AI Chat

​

Semantic Layer Basics

​

Business Names

customer_acq_date → Customer Acquisition Date
mrr               → Monthly Recurring Revenue
cac               → Customer Acquisition Cost

​

Descriptions

• What the metric or column measures
• How it’s calculated
• When to use it
• Known limitations

​

Data Types

• Dimensions (categories)
• Measures (numeric values)
• Dates and times
• Text and strings

​

AI Chat Mode Integration

​

How It Works

1. User asks a question – e.g. “What were total sales last month?”
2. AI understands intent – uses datamart metadata and semantic layer to interpret the question
3. Generate SQL – creates an optimized query based on the datamart structure
4. Return results – displays data with a natural language explanation

​

LLM Configuration

• OpenAI GPT
• Claude AI
• Azure OpenAI
• Llama
• Mixtral

​

Datamart Management for Semantic Layer

​

Datamart Description

• Table Descriptions: Add business-friendly descriptions for tables
• Column Descriptions: Document what each column represents
• Relationships: Define relationships between tables
• Examples: Add example questions for AI chat mode
• Synonyms: Configure column synonyms for better AI understanding

​

Semantic Layer Tools

• Lab: Generate and test synthetic queries
• Playground: Test natural language questions against the datamart
• Sample Set: Configure sample data for AI training and evaluation

​

Best Practices for Semantic Layer

​

Next Steps