Working with Sources

Sources in dbt represent the raw data tables in your warehouse that serve as the foundation for your transformations. Rather than referencing raw tables directly, dbt allows you to define sources in a centralized way, improving maintainability and enabling powerful features like freshness checking.

What Are Sources?

In dbt, sources represent raw data tables from external systems, such as an operational database, CRM, or third-party APIs. Instead of referencing raw tables directly in models, dbt allows you to define sources in a centralized file (sources.yml) for better organization, maintainability, and documentation.

The sources.yml file is a crucial component in dbt projects, centralizing metadata about raw data tables. This ensures consistency, maintainability, and automatic documentation.

Why Use Sources?

Centralizes raw table definitions – Avoids hardcoded table names across multiple models.
Improves maintainability – If raw table locations change, you only need to update sources.yml.
Enables freshness checks – dbt can monitor source data latency.
Enhances documentation – Automatically generates lineage graphs and model dependencies.

Defining Sources in YAML

Sources are defined in .yml files under the sources: key. Here's a typical example:

version: 2

sources:
  - name: jaffle_shop  # Logical name of the source
    database: raw      # The database where the source is stored (optional)
    schema: jaffle_shop  # Schema containing the source tables
    tables:
      - name: orders
        columns:
          - name: id
            tests:
              - unique
              - not_null
          - name: status
            tests:
              - accepted_values:
                  values: ['placed', 'shipped', 'completed', 'returned']
      - name: customers

In this example:

We've defined a source named jaffle_shop that points to tables in the raw.jaffle_shop schema
We've defined two tables: orders and customers
We've added column-level tests to the orders table

Using Sources in Models

Once sources are defined, you can reference them using the source() function in your dbt models:

-- models/staging/stg_orders.sql
SELECT
  order_id,
  customer_id,
  order_date,
  status,
  amount
FROM {{ source('jaffle_shop', 'orders') }}

This offers several advantages:

Consistency: Source references are standardized across your project
Refactoring: If a source table moves, you only need to update one place
Documentation: dbt automatically builds lineage from sources to models
Testing: You can apply tests to sources for early validation

Best Practices for Source Organization

Organize sources by system or domain:

sources:
  - name: stripe       # Payment processing
    schema: raw_stripe
    tables:
      - name: charges
      - name: customers

  - name: shopify      # E-commerce platform
    schema: raw_shopify
    tables:
      - name: orders
      - name: products

Document Your Sources

Add descriptions to help your team understand the data:

sources:
  - name: google_analytics
    description: "Web analytics data from our marketing site"
    tables:
      - name: sessions
        description: "User sessions with UTM parameters"
        columns:
          - name: session_id
            description: "Unique identifier for the session"

Apply Tests to Sources

Find data quality issues early by testing your sources:

sources:
  - name: crm
    tables:
      - name: customers
        columns:
          - name: customer_id
            tests:
              - unique
              - not_null
          - name: email
            tests:
              - unique
              - not_null

Source Freshness

One of the most powerful features of sources is the ability to check data freshness - ensuring your source data is up-to-date before you build models on top of it.

Configuring Freshness Checks

Add a freshness block and specify a loaded_at_field in your sources definition:

sources:
  - name: sales_data
    schema: raw_sales
    freshness:
      warn_after: {count: 12, period: hour}
      error_after: {count: 24, period: hour}
    loaded_at_field: updated_at
    tables:
      - name: transactions

This configuration:

Uses the updated_at column to determine when data was last loaded
Warns if data is more than 12 hours old
Errors if data is more than 24 hours old

Running Freshness Checks

Check freshness with:

dbt source freshness

The output will show the status of each source:

16:35:31 | Freshness of jaffle_shop.orders: PASS (0 seconds)
16:35:32 | Freshness of jaffle_shop.customers: WARN (13 hours)

Table-Specific Freshness

You can override source-level freshness settings for specific tables:

sources:
  - name: inventory
    freshness:
      warn_after: {count: 12, period: hour}
      error_after: {count: 24, period: hour}
    loaded_at_field: last_updated
    tables:
      - name: daily_stock
      - name: real_time_stock
        freshness:
          warn_after: {count: 15, period: minute}
          error_after: {count: 30, period: minute}
        loaded_at_field: timestamp

In this example, real_time_stock has stricter freshness requirements than other tables in the source.

Advanced Source Configurations

Source Overrides by Environment

You can override source details for different environments by using custom schemas:

sources:
  - name: marketing
    database: "{% if target.name == 'prod' %}analytics{% else %}raw_data{% endif %}"
    schema: "{% if target.name == 'prod' %}production{% else %}{{ target.schema }}{% endif %}"
    tables:
      - name: ad_campaigns

Filtering Source Data

For large source tables, you can define filter conditions:

sources:
  - name: logs
    tables:
      - name: application_logs
        external:
          location: "s3://my-bucket/logs/"
          options:
            format: parquet
        freshness:
          filter: "date_column >= dateadd('day', -3, current_date)"

Automating Source Definitions with DinoAI

With DinoAI (Paradime's AI Agent), you can automatically generate source definitions with appropriate freshness configurations. See step by step instructions for more details.

Benefits of automating Source definitions with DinoAI:

✅ Scans your data warehouse and auto-generates the correct table definitions. ✅ Prevents manual errors in source definitions. ✅ Keeps sources up to date with your evolving data warehouse schema.

Common Source Patterns

Staging Models for Sources

A common pattern is to create staging models that select from sources. These provide a clean interface between raw data and your transformations:

-- models/staging/stg_customers.sql
SELECT
  customer_id,
  first_name,
  last_name,
  email,
  created_at,
  updated_at
FROM {{ source('crm', 'customers') }}

Testing Complex Source Relationships

You can test relationships between source tables:

sources:
  - name: application
    tables:
      - name: users
        columns:
          - name: user_id
            tests:
              - unique
              - not_null
      - name: orders
        columns:
          - name: user_id
            tests:
              - relationships:
                  to: source('application', 'users')
                  field: user_id

By effectively managing sources in dbt, you build a strong foundation for your analytics pipeline, making it easier to maintain, test, and document the origin of your data.

PreviousProject Strucuture NextTesting Data Quality

Last updated 13 days ago

Was this helpful?