Sources

Sources help you organize data sources to which Ataccama ONE can connect, such as databases, data lakes, object storages, file systems, BI and analytics tools, and more.

See Supported Connectors for the full list of supported connectors.

Understanding sources and connections

A source is a business representation of a data source, while a connection is the actual technical integration.

For example, your Marketing department might have a single source called "Marketing Analytics" containing multiple connections:

Salesforce connection for customer and campaign data.
Google Drive connection for market research data.
Tableau connection for performance dashboards.

Meanwhile, your Data Platform team might have a "Cloud Data Warehouse" source with connections to Snowflake (production data), Amazon S3 (raw data lake), and Databricks (analytics workloads).

How to organize sources and connections

You can organize your data sources in different ways, for example:

Single source, single connection: Simple setups where one business unit uses one data source (for example, a "Sales CRM" source for Salesforce data).
Single source, multiple connections: When logically related data lives in different systems (for example, a "Customer Data" source connecting to PostgreSQL for operational customer data, Amazon S3 for customer data extracts, and Tableau for customer analytics dashboards).
Multiple sources: For clear organizational boundaries or different stewards (for example, separate sources for each department).

Connect to a data source

To connect to a data source:

Create a new source or edit an existing one. To edit an existing source, go to Sources > [your source] > Connections.
Add a connection to your data source, including:
- Connection parameters
- Authentication credentials
- Optional advanced settings.
  
  For detailed guides and reference information, see Supported Connectors.
Test the connection to verify credentials and access.
Save and publish your source.
Start populating your Catalog. See Discovery Actions in Catalog.

Create source

Go to Catalog > Sources.
Select Create.
Fill in the following:
- Name: Give your source a clear name.
- Description: Explain what the source is used for or provide other useful information.
- Deployment (Optional): Choose the deployment type.
Save your changes.

Test connection

To test and verify whether the data source connection has been correctly configured, select Test Connection.

If the connection is successful, save and publish your connection. Otherwise, verify that your configuration is correct and that the data source is running.

Save and publish

Once your connection is ready, save and publish your changes. The connection is now available for other users in the platform.

In case your configuration is missing required fields, a list of detected errors is displayed. Review your configuration and resolve the issues before continuing.

Edit connection

Before editing an existing connection, review Good to know.

To edit a connection, go to Sources > [your source] > Connections and select the connection you want to modify. In the Connection detail section, open the three-dot menu and select Edit.

Alternatively, on the source Connections tab, open the three-dot menu next to the connection and select Edit.

For connector-specific configuration, see Supported Connectors.

Good to know

Saving changes to an existing connection:

Removes all configured credentials as a security precaution. You’ll need to re-enter credentials before you can test or publish the connection again.

If you opened the connection just to review it, leave the page without saving to keep your credentials intact.
Changes how already-imported catalog items are processed if you update the JDBC connection path (specifically, the target database). Because item names must be unique per connection, items re-import under the new path and can fail with validation errors.

To avoid this, duplicate the connection instead, or remove the existing catalog items before editing.

Removing catalog items deletes their associated data; review carefully before proceeding to avoid losing data.

Compliance mode

Compliance mode controls which edge processing results can leave your edge environment and be stored in ONE. It applies only to connections that use an edge instance; other connections are processed in Ataccama Cloud.

The following modes are available:

Strict (no data leaves Edge): No sensitive data leaves your edge environment. DQ evaluation and profiling still run on the edge, and ONE receives only metadata, such as DQ scores and record counts.

Some features are limited or unavailable. See Strict mode.
Normal (DQ results visible): DQ results are available in ONE, including invalid record samples and full profiling results. Invalid record samples and full profiling results stay stored on the edge and are loaded only when you view them. Data processed on the edge can’t be loaded to reference data.
Allow exports: The same as Normal (DQ results visible), and in addition, data processed on the edge can be loaded to reference data. See Load edge data to reference data.

Strict mode

When a connection uses Strict (no data leaves Edge), the following information about its data stays available in ONE:

Names of databases, schemas, tables, and columns.
DQ scores and the number of records that passed and failed.
Record counts and other aggregate statistics that don’t expose individual values.

Individual data values, and statistics derived from them, stay on the edge. In Strict (no data leaves Edge) mode, different processing actions behave as follows:

Feature	Behavior in Strict mode
DQ evaluation	DQ evaluation runs, and scores and charts are available in the DQ monitor. No invalid record samples are stored or shown: Collect invalid samples is unavailable in the monitor settings, and the monitor’s Records tab is grayed out. See Invalid records.
Export of invalid records	Exporting invalid records to your own storage is not affected. See Export Invalid Records.
Profiling	Profiling runs with standard statistics only: Shown: Counts (DQ insights), text length statistics. Not shown: Frequency, frequency group, quantile, mask, pattern analyses and numerical statistics. On the attribute Profiling tab, a Compliance mode is on card explains why other metrics are not available. See Understand Profiling Results.
Anomaly detection	Anomaly detection doesn’t run, because it needs profiling values that can’t leave your edge environment. Anomaly detection is unavailable in the monitor settings. See Anomaly Detection.
Data transformations	Transformation plans don’t run on the connection: a plan that reads from it fails with a message that the connection’s compliance mode doesn’t allow it.
Data preview by query	Not available on the connection, because the preview would show its data values. The request fails with an error message.
Load to reference data	Not allowed. See Load edge data to reference data.

Feature

Behavior in Strict mode

DQ evaluation

DQ evaluation runs, and scores and charts are available in the DQ monitor. No invalid record samples are stored or shown: Collect invalid samples is unavailable in the monitor settings, and the monitor’s Records tab is grayed out. See Invalid records.

Export of invalid records

Exporting invalid records to your own storage is not affected. See Export Invalid Records.

Profiling

Profiling runs with standard statistics only:

Shown: Counts (DQ insights), text length statistics.
Not shown: Frequency, frequency group, quantile, mask, pattern analyses and numerical statistics.

On the attribute Profiling tab, a Compliance mode is on card explains why other metrics are not available. See Understand Profiling Results.

Anomaly detection

Anomaly detection doesn’t run, because it needs profiling values that can’t leave your edge environment. Anomaly detection is unavailable in the monitor settings. See Anomaly Detection.

Data transformations

Transformation plans don’t run on the connection: a plan that reads from it fails with a message that the connection’s compliance mode doesn’t allow it.

Data preview by query

Not available on the connection, because the preview would show its data values. The request fails with an error message.

Load to reference data

Not allowed. See Load edge data to reference data.

Change the compliance mode

To change the compliance mode, edit the connection and select a different option in Compliance mode. Re-enter the credentials and publish the connection again. See Good to know.

Changes to the compliance mode are recorded in the audit log.

If you switch a connection from Strict (no data leaves Edge) to another mode, results stored before it was set to Strict, such as invalid record samples, are shown again. Runs made in Strict mode collect no invalid record samples or full profiling statistics, so they have nothing to show after the switch.

Load edge data to reference data

Data processed on the edge can be loaded to reference data only if the connection uses Allow exports compliance mode. This applies when:

A transformation plan writes to reference data using the Reference data output step.
You create a reference data table from a catalog item on this connection. See Create a table from catalog data.

With any other compliance mode:

Create reference data is grayed out on catalog items from this connection.
Catalog items from the connection are hidden when selecting a catalog item to create a reference data table from.
A transformation plan that writes to reference data fails with the error: Can’t export from this edge connection. Ask your administrator to enable edge export for this connection.

Next steps

Assign Stewardship: Assign an owner group to your new source.

Was this page useful?