|Index|Using SUSE AI Factory with NVIDIA|What deployment types are available?
Applies to SUSE AI Factory with NVIDIA

1 What deployment types are available?

 

You can deploy AI applications and blueprints using three primary methods: direct Fleet bundles, GitOps-driven Fleet publishing, or manual Helm installations. Select your preferred deployment type during the application installation wizard.

Fleet Bundle

Deploy the application directly as a Fleet bundle managed by Rancher Fleet. This option is ideal when using Fleet without maintaining a separate Git repository. For more details, see the Rancher Fleet documentation.

Publish to Fleet Git

Publish the application configuration to an external Git repository monitored by Fleet. Fleet automatically deploys, monitors and reconciles the application using a GitOps workflow.

Helm

Deploy the application using Helm commands or the Rancher Apps UI. Use this method for direct installation and manual lifecycle management.

Note
Note: Helm chart size limitation

Large Helm charts may not be supported in SUSE Rancher Prime. This is due to a Kubernetes object-size limit of approximately 1 MB. Because SUSE Rancher Prime stores Helm chart content in Kubernetes objects, charts whose packaged content exceeds this limit may fail to deploy or upgrade.