All Products
Search
Document Center

Dataphin:Cross-tenant deployment

Last Updated:Jun 23, 2026

Cross-tenant deployment migrates data between tenants so you can adapt to different application scenarios and maximize data asset value.

Prerequisites

  • You must purchase the cross-tenant deployment add-on and enable the module for the current tenant. To export or import objects like label schemas, labels, or quality rules, you must also enable their respective features.

  • Ensure the source and destination tenants run the same version of Dataphin and use the same type of compute engine. Otherwise, deploying compute source objects will fail. For more information about compute engine settings, see Compute Settings.

  • Ensure the source and destination tenants use the same cluster for their compute engine. If they use different compute engines, you must first migrate the compute engine's metadata and business data.

  • Ensure that the destination tenant has network connectivity to the compute engine and business databases.

  • Ensure that maintenance mode for cross-tenant deployment is enabled in both the source and destination tenants.

  • The source and destination tenants must use the same credentials for cross-tenant deployment.

Limitations

  • For custom data sources, you must manually create the data source in the destination environment and ensure its type code matches the one in the source environment. Inconsistencies may cause unknown exceptions during deployment.

    Note

    Custom data sources created in Dataphin versions earlier than v3.10 do not support cross-tenant deployment.

  • Configuration files for data sources and compute sources cannot be exported and must be re-uploaded during deployment.

  • Among real-time compute sources, only Fusioninsight Flink cannot be exported.

  • Real-time compute templates and real-time tasks created from templates cannot be exported.

  • Data service APIs cannot be exported and must be migrated manually.

  • Ad hoc query files cannot be exported.

  • During a cross-tenant deployment, users other than system administrators and cross-tenant deployment users can access only the following menus.

    • Data Development & Construction > Development (Data Integration, Data Development, Task Publishing, Task O&M)

    • Asset Governance Management > Governance (Asset Inventory)

    • Asset Analysis & Application > Analysis (SQL Query, Notebook)

    • More > Management Center (Permission Management)

    • More > My (Personal Center, Personal Data Center, Alert Center, Message Center, Task Center)

    • User-defined Menu

  • Deployment packages from a single-engine tenant can only be imported into another single-engine tenant. Similarly, deployment packages from a multi-engine tenant can only be imported into another multi-engine tenant.

Background

Dataphin fully isolates data between tenants for security. If you need to build a data ecosystem or enforce strict controls—such as separate development, testing, and production environments across tenants—you would typically have to redevelop all data assets in the new tenant, reducing efficiency.

Cross-tenant deployment solves this by providing a secure, efficient migration platform.

Cross-tenant deployment is suitable for the following scenarios:

  • Environment migration, such as moving from an on-premises deployment to public cloud SaaS.

  • Isolation between development and production environments (instances), such as dev/prod isolation in the financial industry.

image

Overview

Cross-tenant deployment deploys objects from a deployment package in bulk. The system automatically determines the deployment order based on object dependencies to ensure a smooth process. Dataphin supports deploying various object types across tenants, including planning, data architecture, development, and asset-related objects.

  • Global objects for tenant deployment: Includes statistical periods, global variables, public calendars, offline scheduling templates, data domains, compute sources, projects, data sources, object attributes, and identification features.

  • Data architecture objects: Includes subject areas, business objects, and business activities.

  • Development objects: Includes integration tasks, modeling development objects, metric development objects, resource functions, and real-time and offline compute tasks.

  • Label schema objects: Includes label entities and label entity IDs.

  • Label objects: Includes offline labels (metric mappings, rule combinations) and offline views.

  • Data standard objects: Includes standard objects (standard set directories, standard sets, standard templates, data standards, common standard attributes), compliance mapping rules, mappings, code tables, and word roots.

  • Data quality objects: Includes rule templates and quality rules.

  • Data security objects: Includes security rules (data classification, data sensitivity levels, identification features), identification results, and keys.

Key concepts

Tenant: In a Dataphin instance, the workspace can be divided into isolated tenants where you define business domains and projects. With cross-tenant deployment, you export development objects from a source environment into a deployment package and import that package into a destination environment.

Source environment: The environment from which you migrate data.

Destination environment: The environment to which you deploy objects. Source and destination are relative concepts. For example, if you deploy data from Environment A to Environment B, Environment A is the source and Environment B is the destination.

Cross-tenant deployment user: A user role with permissions for cross-tenant deployment.

Procedure

image

Step

Description

Step 1: Prepare for cross-tenant deployment

  • In the source and destination environments, use a super administrator or system administrator account to assign cross-tenant deployment users. For details, see Add Dataphin Member.

  • Use a super administrator or system administrator account to enable maintenance mode for cross-tenant deployment. For details, see Maintain/Upgrade Dataphin.

Step 2: Configure cross-tenant deployment settings

In the source and destination environments, the cross-tenant deployment user configures the retention period for deployment packages, sets cross-tenant deployment credentials, and configures permission validation and process settings. For details, see Cross-Tenant Deployment Settings.

Step 3: Export and download the deployment package from the source environment

In the source environment, the cross-tenant deployment user exports and downloads the deployment package. You can export a deployment package using full export, incremental export, or by selecting specific objects.

  • Full export: Exports all objects from the source environment. For details, see Export Full Package.

  • Incremental export: Exports objects that have changed in the source environment. For details, see Export Incremental Package.

  • Export by selecting specific objects: Exports specific objects from the source environment. For details, see Export Package by Object.

Note

The development module in the exported deployment package does not include instance data generated by O&M.

Step 4: Import the deployment package into the destination environment

In the destination environment, the cross-tenant deployment user imports the deployment package. For details, see Import Package.

Step 5: Deploy objects in the destination environment

In the destination environment, select an imported deployment package and then deploy its objects. For details, see Objects to be Deployed.

Considerations

  • During a cross-tenant deployment, you can create new tasks in both the source and destination tenants. However, you cannot deploy them to the production environment (or submit tasks in a Basic environment). Any draft or development tasks created during this period must be migrated manually to sync them with the destination environment.

  • Dataphin generates a global dependency graph based on data lineage to determine the correct deployment order and ensure dependency accuracy. When you use the One-Click Publish feature, you do not need to manage the deployment order.

    Before you use One-Click Publish, we recommend that you first manually deploy compute sources and data sources.

  • During a cross-tenant deployment, the system generates a globally unique ID (GUID) for each object, ensuring it is uniquely identifiable across environments.