Overview

This Deployment Guide for NVIDIA Enterprise Reference Architectures provides a step-by-step process to deploy Upstream Kubernetes (K8s) on a cluster running NVIDIA Base Command Manager (BCM) 10.x. Kubernetes is required for scalable AI inference and machine learning workloads in enterprise environments.

While multiple Kubernetes distributions exist, including commercial and community supported options, this guide focuses specifically on upstream Kubernetes, as it is the deployment option available through NVIDIA BCM. The full BCM Deployment Guide is available on the NVIDIA Enterprise Reference Architecture Docs webpage under Deployment.

The following steps support successfully deploying a standard Kubernetes cluster environment by using BCM 10:

• Master nodes as K8s Master

• Compute nodes as K8s Workers

• Ready for application deployment and cluster management

• Extensible with GPU support, private registries, ingress controllers, and more