Deploy a GPU-Accelerated VI Workload Domain for VMware Private AI Foundation with NVIDIA

As a cloud administrator, in VMware Cloud Foundation, deploy a VI workload domain with the GPU-enabled ESXi where data scientists, MLOps engineers, and DevOps engineers will run AI workloads.

• The VI workload domain is based on the vSphere Lifecycle Manager image containing the host manager driver VIB file.
• The VI workload domain vCenter Server instance is deployed in the vCenter Single Sign-On domain of the management domain.

Note: This documentation is based on VMware Cloud Foundation 5.2.1. For information on the VMware Private AI Foundation with NVIDIA functionality in VMware Cloud Foundation 5.2, see VMware Private AI Foundation with NVIDIA Guide for VMware Cloud Foundation 5.2.

Prerequisites

See Requirements for Deploying VMware Private AI Foundation with NVIDIA.

Procedure

1. For a VMware Cloud Foundation 5.2.1 instance, log in to the vCenter Server instance for the management domain at https://<vcenter_server_fqdn>/ui.
2. Select vSphere Client > Private AI Foundation.
3. If you are using the Private AI Foundation guided deployment workflow for the first time, enter your VMware Private AI Foundation with NVIDIA license.
4. In the Private AI Foundation workflow, click the Set Up a Workload Domain section.
5. Create a network pool to have static IP addresses automatically assigned to vSAN, NFS, iSCSI, and vMotion VMkernel ports of the ESXi hosts in the workload domain.
   See Network Pool Management. The wizard in the guided deployment workflow has the same options as the analogous wizard in the SDDC Manager UI.
6. Commission the ESXi hosts to add them to the inventory of SDDC Manager.
   See Commission Hosts. The wizard in the guided deployment workflow has the same options as the analogous wizard in the SDDC Manager UI.
7. Deploy the VI workload domain.
   The wizard in the guided deployment workflow provides the same options as in the SDDC Manager UI except for the following settings that are specific to VMware Private AI Foundation with NVIDIA:

   • Join the workload domain to the management vCenter Single Sign-On (SSO) domain.
   • For cluster lifecycle management, select Manage this clusters using vLCM images and select the vSphere Lifecycle Manager image with the host driver VIB file from NVIDIA.
   • Use networking only based on NSX.
   • Select hosts whose NVIDIA vGPU state is Ready.
   • Select License Later and assign the VMware Cloud Foundation license to the VI workload domain by using the SDDC Manager UI or the vSphere Client. For vSAN storage, you must also add a VMware vSAN license key.

   For more information on creating a VI workload domain, see Deploy a VI Workload Domain Using the SDDC Manager UI.

8. In the vSphere Client, on the vCenter Server instance for VI workload domain, set the vgpu.hotmigrate.enabled advanced setting to true so that virtual machines with vGPU can be migrated by using vSphere vMotion.

   See Configure Advanced Settings.

Results

After the VI workload domain is created, you see its vCenter Server instance in the inventory in the vSphere Client.