Skip to main content
Moving to a new data warehouse, for example from Snowflake to Databricks, does not mean rebuilding your journeys. In Unified Mapping, every Inflection field is separate from the source that feeds it. You swap the source underneath the field, and every journey, audience, and token that uses the field keeps working.
Do not delete fields to switch warehouses. Deleting a custom field is blocked while journeys or audiences use it. Removing the old source from the field is all you need, and it has no dependency check.

Before You Start

Prerequisites
  • Admin access to your Inflection workspace.
  • Credentials for the new warehouse. See Import from Data Warehouse and the setup guide for your warehouse.
  • A table or view in the new warehouse for each object you are moving, with a column for each field you map today.
Your warehouse data comes in as objects, such as Contacts, Product users, Accounts, or Organizations. Each object can be read by only one connection at a time. Decide which kind of move you are making. It changes step 2.

Step 1: Remove the Old Source in Unified Mapping

  1. Go to Connections > Unified Mapping in the left nav.
  2. Select the fields for the object that reads from the old warehouse, for example Product User Fields.
  3. Click Edit mapping.
  4. On each field fed by the old warehouse, click the × on the old warehouse’s source block, for example Snowflake. You can remove all of them before saving.
  5. Click Save changes.
The page shows Mapping saved, and fields with no other source read No sources configured. The fields themselves stay in place. Repeat for every object that reads from the old warehouse. For a full swap, that is every object on the old connection. Removing the Snowflake source from each Product User field in Unified Mapping and saving
Removing a source only stops new data from coming in from that source. It does not delete data already in Inflection. Existing values stay and your journeys and audiences keep running as they are. New updates resume once you map the new source. Learn more in Unified Mapping.

Step 2: Clear the Old Connection

The new connection cannot read an object while the old connection still holds it. Until you finish this step, the new connection shows This object is already mapped by another connection and cannot be configured here.

Full Swap: Delete the Old Connection

  1. Go to Connections > Data Connections in the left nav.
  2. Find the old connection and delete it.
  3. Click Delete to confirm.
Delete connection confirmation for the old warehouse connection If any field in Unified Mapping still reads from the connection, the delete is blocked and the mappings are listed. Click each listed mapping to open it, remove the old source as in step 1, then delete the connection again. Delete connection blocked, listing the Product User Mapping that still references the connection

Partial Swap: Remove the Moving Objects

  1. Open the old connection and go to Set up your data.
  2. Select the object in the left rail, for example Product users.
  3. Click Remove from connection, then click Remove to confirm.
Inflection stops syncing the object and drops its mapping. The rest of the connection keeps running, and the tables in your warehouse are not touched. Repeat for each object you are moving. Removing Product users from the old connection in Set up your data If the object’s fields are still mapped in Unified Mapping, the removal is blocked and the mapping is listed. Remove the old source as in step 1, then remove the object. Remove from connection blocked because Product users fields are still mapped in Unified Mapping
You cannot remove the last object on a connection. If every object is moving, delete the connection instead (full swap).

Step 3: Set Up the New Connection

Create the new connection and set up the objects you are moving. Follow Import from Data Warehouse for the full wizard: Connect, Set up your data, and Save & activate. When you activate, the page shows This connection is live. Save & activate step for the new Databricks connection showing This connection is live and Manual sync
You do not need to run a sync before mapping. The new connection’s columns are available in Unified Mapping as soon as you save the connection. We recommend finishing the mapping in step 4 before the first sync, so the first sync fills in your field values. If the first sync runs before you finish mapping, the values populate on the next sync, which re-reads the full table.

Step 4: Map the New Source in Unified Mapping

  1. Go to Connections > Unified Mapping in the left nav.
  2. Select the object’s fields, for example Product User Fields.
  3. Click Edit mapping.
  4. On each field, click Add source and select the new warehouse, for example Databricks.
  5. Select the column to read from.
  6. Click Save changes.
Repeat for every object you are moving. Your journeys, audiences, and tokens now read from the new warehouse with no other changes. To bring data in now, return to the new connection’s Save & activate step and click Manual sync. Otherwise the data arrives on the next scheduled sync. Adding Databricks as the source for Product User fields in Unified Mapping
After the first sync from the new warehouse, review the journeys and audiences that rely on the moved fields to confirm the values look as expected.

Troubleshooting

You do not need to delete it. Remove the old warehouse as a source on the field instead (step 1). The field and its journey references stay in place.
The message This connection can’t be deleted yet. Its data is still in use. means a field in Unified Mapping still reads from it. Click each listed mapping, remove the source, and try again.
The message can’t be removed yet means the object’s fields are still mapped in Unified Mapping. Click the listed mapping, remove the source, and try again.
You cannot remove the last object on a connection. Delete the connection instead.
The old connection still holds the object. Finish step 2, then return to the new connection.

Frequently Asked Questions

No. Data already synced stays in Inflection. Removing a source only stops new updates from it.
No. They reference Inflection fields, not warehouse columns, so they keep working when you swap the source.
Pausing stops syncs but does not free the objects for the new connection. Delete the connection or remove the objects instead.

Additional Resources