## Data Connections and Integration

URL: https://docs.atlan.com/faq/data-connections-and-integration

> General questions about connecting Atlan to your data sources, what metadata Atlan collects, and how to extend connectivity beyond native connectors.

Atlan connects to your data sources—warehouses, lakehouses, BI tools, pipelines, and SaaS—and crawls their metadata into one governed context layer for AI.

General questions about connecting Atlan to your data sources, what metadata Atlan collects, and how to extend connectivity beyond native connectors. For questions about a specific connector, see that connector's documentation under [Connectors](https://docs.atlan.com/llms/catalog/connector-capabilities/llms.txt).

### What types of data sources can Atlan connect to?

Atlan connects to the full spectrum of enterprise data infrastructure—both structured and unstructured:

- **Systems of Data**: Data warehouses and lakes (Snowflake, Databricks, BigQuery, Redshift, S3), databases (Postgres, MySQL, MongoDB, Oracle), and streaming systems (Kafka)
- **Systems of Semantics**: BI and analytics tools (Tableau, Power BI, Looker, Sigma, Mode)
- **Systems of Record**: Operational databases and SaaS platforms accessed via API
- **Systems of Knowledge**: Documentation tools, notebooks, and workflow systems (Jupyter, Airflow, dbt, Fivetran, Matillion)
- **Unstructured sources**: Via the open API architecture—any source that exposes metadata programmatically can be connected

For a complete list of native connectors, see [Supported sources](https://docs.atlan.com/llms/catalog/connector-capabilities/supported-sources/llms.txt).

### How can I connect Atlan to my tools and systems?

There are four integration layers, depending on what you're trying to do:

**Native connectors**: Pre-built, crawl-based integrations for the most common data platforms. Managed through [Connections](https://docs.atlan.com/llms/catalog/connector-capabilities/llms.txt), which handles workflow scheduling, preflight validation, query mining, and error analysis. For the full list of supported sources, see [Supported sources](https://docs.atlan.com/llms/catalog/connector-capabilities/supported-sources/llms.txt).

**App framework**: For automation beyond metadata crawling: bulk imports, lineage enrichment, data model ingestion, and custom governance workflows. Apps run as scheduled or on-demand batch processes. See [Apps](https://docs.atlan.com/llms/platform/packages/llms.txt) to browse what's available and build your own using the same framework.

**APIs and SDKs**: For integrating Atlan with any web application or building custom connectors. Atlan exposes REST APIs for all platform functionality plus Python, Java, and JavaScript SDKs. You can programmatically catalog any asset type—databases, schemas, tables, dashboards, or fully custom types—for sources without a native connector. See the [developer portal](https://docs.atlan.com/llms/platform/build-apps/llms.txt) for guides and examples.

**MCP server**: For connecting AI agents (Claude, Cursor, Windsurf, Microsoft Copilot Studio) directly to Atlan's metadata. The [Atlan MCP server](https://docs.atlan.com/llms/governance/atlan-mcp-overview/llms.txt) exposes asset search, lineage exploration, metadata management, tagging, and data quality enforcement as tools that agents can call in real time.

### What data does Atlan actually collect from connected sources?

Atlan extracts metadata only—never your actual business data:

- **Schema information**: Table structures, column names, data types, relationships
- **Lineage**: Data flow and transformation logic across pipelines
- **Usage statistics**: Query patterns and user access logs (where available)
- **Quality signals**: Metrics from connected data quality tools
- **Semantic metadata**: Business definitions, BI layer semantics, report structures
- **Custom metadata**: Tags, descriptions, business glossary terms you've applied

Atlan never stores your actual row-level data. It catalogs the metadata that lets you understand, govern, and activate your data landscape.

### What happens to asset metadata when source assets are removed or renamed?

Asset continuity in Atlan is tied to the `qualifiedName`, the unique identifier for each asset:

- **Same `qualifiedName`**: Asset retains all metadata, tags, glossary links, and lineage
- **New `qualifiedName`** (for example, server migration, rename): Creates a new asset in Atlan—previous metadata isn't automatically transferred
- **Asset removed from source**: Automatically archived (soft-deleted) in Atlan during the next crawler run—it disappears from search results but remains [accessible through discovery](https://docs.atlan.com/llms/catalog/discovery/access-archived-assets/llms.txt)

### What does "API only" mean in connector documentation?

In the _Where in Atlan_ column of _What does Atlan crawl from (connector)?_ reference pages, **API only** means that property has been crawled and stored in Atlan but isn't surfaced or searchable in the product UI. It's available via API for custom integrations.

### Is offline extraction still available?

No. Offline extraction has been sunset and is no longer available.

For on-premises or network-restricted environments, use the **Agent** extraction method with [Self-Deployed Runtime](https://docs.atlan.com/llms/platform/self-deployed-runtime/llms.txt) instead. Self-Deployed Runtime runs within your organization's network and connects directly to your data sources without requiring public internet access from Atlan.

---