Skip to main content
Reference

Data Connection Migration

Migrate legacy data connections in the Model Studio console to the new Connector.

Connector is the upgraded version of data connections in Alibaba Cloud Model Studio. If you previously created data connections in the Model Studio console, those connectors must be migrated to the new version before you can manage them in the Connector console.
The one-click migration entry point is available until September 30, 2026. Complete your migration before that date.

What's new in this version

Legacy data connectionNew Connector
Entry pointThe data connection page in the Model Studio consoleA standalone Connector console
Connectable systemsFiles, tables, MySQL, PostgreSQL, PolarDB-X 2.0, Yuque, OSSEverything above, plus the DingTalk family, Salesforce, MaxCompute, Yunxiao, and more
Consumption methodsAgent skills, workflow nodesEverything above, plus MCP support so that any MCP-compatible client can connect
OrganizationA flat connector listGrouped by app, where one app can hold multiple connectors
Migration does not change the data itself. Files that have already been imported, categories, and configured sync rules all remain available.

Perform the migration

1

Open the migration page

In the left-side navigation of the Connector console, click Data connection migration.
2

Review the list of items to migrate

The page lists the legacy data connections in the current workspace that can be migrated. Each entry includes:
InformationDescription
Connector nameThe name set in the legacy version
Connector IDIn the form conn_file_xxxxxxxx or conn_table_xxxxxxxx
TypeFILE, TABLE, and so on
Creation timeThe creation time in the legacy version
3

Migrate one at a time

Click Migrate for the target connector and follow the prompts.
4

Verify the result

After the migration finishes, go to the Apps page to confirm the result: a connected count appears on the corresponding app card, and the detail page shows the connector instance along with the automatically generated tools.

Some connectors are marked as not migratable

Some entries in the list are marked as not migratable and cannot be handled through this entry point, such as connectors created by the system by default. To handle this type of connector, create a new connector of the same type in Connector, then re-import the data or reconfigure the sync rules.
If you are unsure which case a given connector falls into, rely on the actual label shown on the migration page.

Pre-migration checklist

  • Record the name, ID, and description of each legacy connector so that you can verify them after the migration.
  • Record the configured sync rules and their source credentials. The credentials live on the source system side and are not affected by the migration, but you will need to enter them again when you rebuild the rules.
  • Identify which agents and workflows reference these connectors.
  • Migration is performed within a workspace, so multiple workspaces must be handled one by one.

What to verify after the migration

Check itemHow to verify
The connector is availableConfirm that the status is connected on the App details page
The tools workExpand a tool to review its input and output parameters, then run a test call from a client or the debugging panel
Agents can still make callsOpen an agent application that references the connector and confirm that the connector and its tools are still in the skill list
Workflow nodes are still validOpen the workflow and confirm that the data connector node still points to the correct connector and tool
Sync rules are still runningCheck the sync rule list and its enabled status on the files tab

What happens after the deadline

The one-click migration entry point closes after September 30, 2026. Any legacy data connection that has not been migrated by then must be rebuilt manually in the new version, including re-creating the connector, re-importing the data, and reconfiguring the sync rules, which costs significantly more effort.
If you have a large number of connectors, migrate one non-critical connector first to validate the process and understand the impact, then handle the rest in batches.