DataWorks:Overview

When to use Deploy Center

Use Deploy Center when you need to:

• Promote validated work to production: Push nodes and their dependencies from a development workspace to a production workspace after testing.

• Isolate dev and prod for basic-mode workspaces: Set up a separate workspace as the production environment for a basic-mode workspace, and deploy objects across them.

• Migrate across regions, accounts, or cloud platforms: Move an entire pipeline from one cloud environment to another when using the legacy Data Studio.

Deployment types

Choose the deployment type that matches your workspace setup:

For step-by-step instructions, see Same-workspace deployment, Cross-workspace deployment, and Cross-cloud deployment.

How it works

When you deploy nodes from a source workspace to a target workspace, Deploy Center automatically:

• Replaces name prefixes: Updates all node input and output names, swapping the source workspace name prefix with the target workspace name prefix.

• Remaps cross-workspace dependencies: If you configure dependency mappings, any cross-workspace dependency on a source workspace is redirected to the mapped target workspace.

• Updates MaxCompute task code: For MaxCompute engine tasks (ODPS SQL, ODPS Spark), replaces all references to the source workspace name in the task code with the target workspace name.

• Enforces dependency order: Deploys upstream nodes before downstream nodes. If an upstream node has not been deployed yet, its downstream nodes wait.

Name prefix replacement

After deployment, every input and output name that started with the source workspace prefix is updated to use the target workspace prefix.

Example: deploying from `project1` to `project2`

Before deployment	After deployment
Input: project1_root	Input: project2_root
Output: project1.task_A	Output: project2.task_A

Cross-workspace dependencies

If any node in the source workspace depends on a node in a different workspace, the behavior after deployment depends on whether you have configured a dependency mapping.

Without dependency mapping

The cross-workspace dependency is carried over as-is to the target workspace.

Example: project1.task_A depends on project2.task_A. All nodes are deployed to project3.

After deployment:

• All project1 prefixes change to project3 in node input and output names.

• project3.task_A retains its cross-workspace dependency on project2.task_A.

With dependency mapping

A dependency mapping redirects cross-workspace dependencies from a source workspace to a different target workspace.

Example: project1.task_A depends on project2.task_A. All nodes are deployed to project4, with a dependency mapping from project2 to project3.

After deployment:

• All project1 prefixes change to project4 in node input and output names.

• project4.task_A's dependency is remapped from project2.task_A to project3.task_A.

To configure dependency mappings, see Configure a deployment environment. To configure scheduling dependencies, see Configure scheduling dependencies.

The output name format WorkspaceName.NodeName is used as an example. Actual output names may vary.

MaxCompute engine task code changes

For MaxCompute engine tasks — such as ODPS SQL or ODPS Spark — Deploy Center rewrites the task code, replacing all references to the source workspace name with the target workspace name.

Example: task_A is an ODPS SQL node in project1. Its query reads:

SELECT * FROM project1.tableA

After deploying to project2, the query becomes:

SELECT * FROM project2.tableA

Avoid system-generated output names for cross-workspace dependencies

If a node's cross-workspace dependency uses a system-generated output name, the deployment may fail. This happens when the Scheduling > Output Parameter of Ancestor Node list references a system-generated output name from another workspace.

System-generated output names follow this format (in workspaces not using the new Data Studio):

WorkspaceName.FileID_out

For example: shanghai_simple02.504822000_out

Use a custom output name instead. Recommended formats:

What's next

• Same-workspace deployment

• Cross-workspace deployment

• Cross-cloud deployment

• Achieve environment isolation in basic mode