Quick Start#
Goal: Get access to the Omniverse Storage APIs and service adapter framework, then choose a deployment path: local or full deployment, or development path: custom adapter development.
The Omniverse Storage APIs are distributed through the NGC catalog. This page walks you through setting up NGC access, then choosing one of three deployment paths: local development, full production-style, or custom adapter development.
Before You Begin#
Regardless of which deployment path you choose, you need:
An NGC account
The NGC CLI installed on your workstation
Each deployment path has additional requirements listed on its own requirements page.
Set Up NGC Access#
Create an NGC account – Sign in to the NGC portal and select your NGC org.
Generate an API key – In the NGC portal, generate a new API key with the following permissions:
NGC Catalog
NVIDIA Private Registry
Install the NGC CLI – Follow the NGC CLI installation guide.
Verify access – Download the Storage API specifications to confirm your setup works:
ngc registry resource download-version "nvidia/omniverse/storage-api:1.0.0-beta.4"
A successful download confirms your NGC access is configured correctly.
Choose your journey#
Path |
What You Deploy |
Best For |
|---|---|---|
Discovery + Storage service (Python example) on MicroK8s |
Learning the deployment process, building and testing custom adapters locally |
|
Discovery, Storage (S3/Azure), Notifications, Permissions, Navigator, and more |
Multi-user production environments with external client access |
|
Your own service adapter from the API specifications |
Integrating with proprietary or unsupported storage backends |
Local Development Deployment#
Deploy Discovery and a Python-based Storage service adapter on MicroK8s. Connect an Omniverse Kit application to verify the deployment, the same way you would connect to Enterprise Nucleus Server.
Additional requirements: MicroK8s, Docker, Python 3.10+, Poetry
Full Deployment#
Deploy the full Omniverse Storage APIs stack: Discovery, Storage (S3/Azure), Notifications, and more. This deployment represents the feature set of Omniverse Enterprise Nucleus Server and supports multiple simultaneous users connecting via Kit or other clients.
Services can be deployed incrementally – you do not need to deploy everything at once.
Additional requirements: Kubernetes 1.31+, kubectl, Helm 3.0+
Custom Adapter Development#
Build a custom service adapter from the Omniverse Storage API specifications. Download the API specs (gRPC and REST), implement your adapter, and validate it against a local MicroK8s deployment.
Start custom adapter development