Importing spaces and projects into existing deployment spaces in Watson Machine Learning
You can import a deployment space or a project (in .zip format) into an existing deployment space. This way, you can automatically add assets or update existing assets (for example, replacing a model with its newer version).
Note:
When you export a project or space, the contents, including model assets, are included in the export package. You can then import the project or space to another server cluster. Note that the underlying architecture
must be the same or you might encounter failures with the deployment of your machine learning models. For example, if you export a space from a cluster running the Power platform, then import to a cluster running x86-64, you may be unable to
deploy your machine learning models.
To import a space or a project when you are creating a new deployment space:
- Click New deployment space.
- Choose Create an empty space.
- Enter the details for the space.
- Find or drop the exported .zip file that contains data assets.
- Click Create. The assets from the .zip file will be added as space assets.
To import a space or a project into an existing space:
- From the space, click the export space () icon.
- Click Import space. If the space that you are importing is encrypted, you will be asked to provide the password.
- Click Browse, find the exported .zip file, and click Open.
Resolving issues with asset duplication
The importing mechanism compares assets that exist in your space with the assets that are being imported. If it encounters an asset with the same name and of the same type:
- If the asset type supports revisions, the importing mechanism creates a new revision of the existing asset and patches the new revision.
- If the asset type does not support revisions, the importing mechanism patches the existing asset.
This table describes how import works to resolve cases where assets are duplicated between the import file and the existing space.
Your space | Space being imported | Result |
---|---|---|
No assets with matching name / type | One or more assets with matching name / type | All assets imported. If multiple assets in the import file have the same name, they will be imported as duplicate assets in the target space. |
One asset with matching name / type | One asset with matching name / type | Matching asset will be updated with new version. Other assets imported normally. |
One asset with matching name / type | More than one asset with matching name / type | The first matching asset processed is imported as a new version for the existing asset in the space, additional assets with matching name will be created as duplicates in the space. Other assets imported normally. |
Multiple assets with matching name / type | One or more assets with matching name / type | Assets with matching name will fail to import. Other assets imported normally. |
For more information on revisions, refer to Managing asset versions.
Existing deployments get updated differently, depending on deployment type:
- If a batch deployment was created by using the previous version of the asset, the next invocation of the batch deployment job will refer to the updated state of the asset.
- If an online deployment was created by using the previous version of the asset, the next "restart" of the deployment will refer to the updated state of the asset.
Learn more
- For an example of how to export and import a deployment space by using the Watson Machine Learning Python client, view or download this sample notebook.
- For details on adding data assets to a space, refer to Adding data assets to a deployment space.
- For details on promoting data assets to a space, refer to Promoting assets to a deployment space.
- For details on importing models to a space, refer to Importing models into Watson Machine Learning.
- For details on exporting assets from a deployment space, refer to Exporting space assets.
Parent topic: Adding assets to a deployment space