How Is It Orchestrated?

The Cube runs a Red Hat OpenShift Single‑Node (SNO) cluster that bundles control‑plane and worker services on the appliance. Key operators are installed automatically at first boot:

• LVM Storage Operator – Creates thin‑provisioned local PVs on the Cube’s NVMe drives for fast, fully‑local persistence.

• Node Feature Discovery (NFD) Operator – Detects hardware capabilities (CPU flags, PCI devices, GPUs), labeling the node so workloads can request them. (docs.redhat.com)

• NVIDIA GPU Operator – Pulls the matching CUDA driver, device plugin, DCGM exporter, and monitoring stack, so GPU‑accelerated containers work out of the box. (docs.redhat.com)

The cluster is managed by the RUN service, which applies continuous updates, rollbacks, and health checks via Palantir Apollo, so your cube stays current without manual kubectl steps.

---