Create AI Self-Service Catalog Items in VMware Aria Automation for Private AI Ready Infrastructure for VMware Cloud Foundation

Use the catalog setup wizard in VMware Aria Automation to create catalog items for GPU-enabled deep learning VMs and for GPU-enabled Tanzu Kubernetes Grid clusters with NVIDIA GPU Operator deployed and configured.

The Private AI Automation Services wizard in VMware Aria Automation can be run multiple times for the following scenarios:

• Add a new vCenter Cloud Account and an associated Supervisor.

• Accommodate a change in your NVIDIA AI Enterprise license, including license server and client configuration token.

• Add or remove VM Classes and content libraries.

• Create the catalog items for a new project.

Prerequisites

• VMware Aria Automation has been deployed with version 8.16+ refer to (LINK TO ILIANA'S ARIA DEPLOYMENT) and Implementation of Private Cloud Automation for VMware Cloud Foundation.

Procedure

1. Log in to VMware Aria Automation as an administrator.
2. Under My Services, click Quickstart.
3. In the Private AI Automation Services card, click Start.
4. Add a cloud account.
   1. Select the vCenter Server for the VI workload domain.
   2. Select the Supervisor.
   3. Select NVIDIA Cloud License Service (CLS) for connected environments or Self hosted - NVIDIA Delegated License Service (DLS) for disconnected environments.

      Refer to NVIDIA's documentation on how to setup the Delegated License Service (DLS) for disconnected environments.

   4. For connected environments, enter the contents of your NVIDIA Client Configuration token file (.tok) and your NVIDIA Licensing Portal API Key, and click Next Step.
   5. For disconnected environments, enter the contents of your NVIDIA Client Configuration token file (.tok) and a vGPU guest driver local URL, and click Next Step.
5. Configure the catalog items and click Next Step.
   1. Select the VM image name to use for provisioning deep learning VMs.
   2. Select the GPU-enabled and non-GPU enabled VM classes for your deep learning VMs and Tanzu Kubernetes Grid cluster nodes.
   3. Select the storage policy that you created earlier .
   4. For connected environments, select Cloud - NVIDIA NGC Catalog, for disconnected environment, Local - self-hosted registry and click Next Step.
6. Select the users and user groups to have access to the project that will be created by the Quickstart wizard, and click Next Step.
   See Personas in Private AI Ready Infrastructure for VMware Cloud Foundation.
7. Review the configuration and click Run Quickstart.
8. Verify that the catalog items are created successfully.
   1. Select Service Broker > Catalog.
   2. Verify that the catalog contains two items - AI Kubernetes Cluster and AI Workstation.