dataflow source change from aws redshift to databricks

[bhavyamalik1]

Hi, I've some reports which use dataflows. Right now, those dataflows are using amazon redshift as the source. I want to use databricks instead. How can i change that within the dataflow? Please guide. I've downloaded the json file for that dataflow so do i need to copy that content and make amendments to that or how can it be done. Please help.

[v-achippa]

Hi @bhavyamalik1,

Thank you for reaching out to Microsoft Fabric Community.

Thank you @rohit1991, @Nabha-Ahmed and @Mauro89 for the prompt response. 

As we haven’t heard back from you, we wanted to kindly follow up to check if the solution provided by the user's for the issue worked? or let us know if you need any further assistance.

Thanks and regards,

Anjan Kumar Chippa

[v-achippa]

Hi @bhavyamalik1,

We wanted to kindly follow up to check if the solution provided by the user's for the issue worked? or let us know if you need any further assistance.

Thanks and regards,

Anjan Kumar Chippa

[Nabha-Ahmed]

Hi @bhavyamalik1 

I hope you are doing well!!

You don’t need to modify the exported JSON file. Dataflow definitions aren’t designed to be edited directly, and manual changes may break the dataflow.

The correct way is to update the connection through Power Query Online:

1. Open the Power BI Service

2. Navigate to the workspace that contains the dataflow

3. Select Edit dataflow

4. In Power Query Online, open each query that currently uses Amazon Redshift

5. Update the Source step to use the Azure Databricks connector instead

6. Provide the Databricks connection details (server, HTTP path, authentication)

7. Save the dataflow and validate the refresh

If the schema between Redshift and Databricks is different, it’s often cleaner to:

Create a new query using Databricks as the source

Reapply the existing transformation steps

Replace the old query once validated

Editing the JSON file is not supported and is not recommended.

This approach keeps the dataflow stable, supported, and easy to maintain.

If this help you mark as "Solution" and put kudo to help other 

[Mauro89 · Accepted solution]

Hi @bhavyamalik1,

You can change the data source in your dataflow, but it requires recreating the connection. Here's how you can try to approach it:

Option 1: Edit directly in Power BI Service (Recommended)

1. Go to your workspace and open the dataflow for editing
2. In Power Query Editor, select the query connected to Redshift
3. Click Data source settings or Configure connection
4. Unfortunately, you cannot switch connector types (Redshift → Databricks) directly
5. You'll need to delete the current source step and add a new Databricks connection
6. Right-click the Source step → Delete
7. Get Data → Databricks → enter your connection details
8. Reapply your transformation steps (they should still be there)

Option 2: Use the JSON file (Advanced)

1. Open the downloaded JSON file
2. Find the data source section (look for "Amazon Redshift" references)
3. Replace with Databricks connection string format:
   • Change the connector from AmazonRedshift to Databricks
   • Update server, database, and authentication details
4. This is risky - one syntax error breaks the dataflow

Option 3: Recreate the dataflow (Safest)

1. Create a new dataflow
2. Connect to Databricks
3. Copy/paste the M code from your existing queries (from the JSON or Advanced Editor)
4. Update reports to point to the new dataflow

Best regards!

PS: If you find this post helpful consider leaving kudos or mark it as solution

[bhavyamalik1]

not getting it as databricks on top but redshift

[Mauro89]

What about Option 3? Connecting to Databricks and move the remaining Power Query code into the new Dataflow?

Best regards!

[bhavyamalik1]

data is reflecting in dataflow but not in pbi

[Mauro89]

Have you actually run the dataflow successfully and in case you used import mode in PBI refreshed the data?

[rohit1991]

Hii @bhavyamalik1 

You should not edit the exported dataflow JSON. Changing sources via JSON is unsupported and can break the dataflow. The correct approach is to edit the dataflow in Power BI Service (Power Query Online), update the Source step from Amazon Redshift to Databricks in the Advanced Editor, then update credentials and refresh. As long as the schema matches, downstream transformations and reports will continue to work.