# Onboard data sources & datasets

Before you can monitor, test, or enforce data contracts in Soda, you need to **connect your data source and onboard your datasets**.

**Onboarding** is the process of:

* **Connecting Soda to your data source** (e.g., Snowflake, BigQuery, Databricks, PostgreSQL, etc.)
* **Discovering** available datasets (tables or views)
* **Selecting** which datasets to monitor and/or validate
* **Configuring** observability, testing, and scheduling behavior

How you onboard datasets depends on how you deploy Soda.

***

### Choose your onboarding path

Soda supports two main ways to onboard datasets:

{% columns %}
{% column %} <a href="onboard-data-sources-and-datasets/onboard-datasets-on-soda-cloud" class="button primary">Via Soda Cloud</a>

This approach is ideal if you want:

* No-code dataset onboarding
* Automated dataset discovery
* Built-in metric monitoring
* Data contracts created in the UI
* Centralized scheduling and alerting
  {% endcolumn %}

{% column %}

#### <a href="onboard-data-sources-and-datasets/via-soda-core" class="button primary">Via Soda Core</a>

This approach is ideal if you want:

* Full control in code
* Git-based workflows
* CI/CD integration
* Contract verification inside pipelines
* Support for in-memory datasets
  {% endcolumn %}
  {% endcolumns %}

You can also combine both approaches: use Soda Cloud for **centralized observability and governance**, and Soda Core for **pipeline-level validation**.