Managing governance activities across multiple clusters
After an instance administrator sets up the clusters for distribution of governance activity on multiple clusters, you can begin tracking governance activity of models.
Required permissions
For this task, you must be a service administrator to manage governance activities with the following permissions:
- Create or edit model inventories
- Evaluate and deploy models
- Create or edit AI use cases
Before you begin
Before you can begin model activity management and tracking, you must have the following prerequisites:
- A primary governance cluster
- At least one remote cluster connected to the primary governance cluster
An instance administrator must set up multiple clusters. For instructions see Managing multiple clusters for watsonx.governance.
Create an end-to-end scenario for governing model lifecycles
After your watsonx.governance administrator set up multiple clusters, you start a new process to manage governance activity. You will transfer models from the primary governance cluster to the remote clusters one time, and then perform operations for your models on the remote clusters while their activities are tracked on the primary governance clusters. For example, a service administrator with a multiple cluster setup for watsonx.governance can create models on a primary governance cluster, import the models to different remote clusters, and evaluate or deploy the models on the remote clusters that they live on. The service administrator can then view the model activities from multiple remote clusters on the primary governance cluster.
To govern your model lifecycles from creation to deployment with multiple clusters, you can either:
- Manage your models on multiple clusters with a direct connection
- Manage your models on multiple clusters with a custom governance connector
Contact your instance administrator to learn which type of connection is set up for your remote cluster.