Preface

You can deploy Red Hat AI Inference Server in a disconnected OpenShift Container Platform environment that does not have direct access to the internet by mirroring Operator and OCI model container images to a local mirror registry and configuring the cluster to use the mirrored images.

After mirroring the required images, you can install the Node Feature Discovery Operator and NVIDIA GPU Operator from the mirrored sources, then deploy Red Hat AI Inference Server for inference serving the OCI-compliant model.