Skip to content

Demo: Edge / Single Node OpenShift

Start here. Take bare x86_64 hardware from power-on to a running OCP Virt demo — AAP provisioning VMs on OpenShift Virtualization, with CIS L1 hardening, Grafana Cloud observability, and nightly teardown — using two public repos and no cloud dependency.
Length Setup: ~90 minutes hands-on, ~45 minutes waiting. Demo: same 30-minute OCP Virt talk track
Audience Platform engineers and edge infrastructure architects evaluating SNO for edge deployments
Reader The Red Hat pre-sales engineer setting up and presenting it
Needs a live environment? Yes — this IS the environment. The whole point is showing it on hardware you own
Status Ready — proven end-to-end on a NUC, 2026-09-09

The four documents

File Read it when
run-sheet.md While building. Step-by-step checklist from bare hardware to demo-ready
talk-track.md Before the meeting. Why bare metal matters, and the words that land
architecture.md When asked "how does that work". The flow, the timing, the storage layout
objections.md Before you go in. What this audience asks, answered from what the kit really does

Build from the run sheet. Present the OCP Virt demo from its own run sheet once the environment is up.

The 60-second version

Two repos. Three phases. One command per phase.

  1. Build the ISO (image.builder.pipeline) — generate-iso.sh takes your hardware details (IP, MAC, disk, NIC), your pull secret, and your SSH key, and produces a bootable Agent-Based Installer ISO. Day 0 manifests bake in AAP 2.7, OpenShift Virtualization, LVMS, and the Compliance Operator.
  2. Boot the hardware — write the ISO to USB, boot from it, wait ~45 minutes. The cluster installs itself, fully unattended.
  3. Configure the platform (sales.demos) — install LVMS storage, deploy AAP from its operator, install CNV, run a compliance scan, and prove the environment by building and timing a real VM. Five playbooks today; a single setup_edge.yml that chains them is sales.demos#406.

After phase 3, the environment is identical to an RHDP sandbox — same playbooks, same job templates, same demo. The difference is you own it.

What the demo is actually about is not that we automated a NUC. It is that the same automation that runs in the cloud runs on hardware under a desk, and the customer can hold both in their hands.

Why bare metal

Three things RHDP cannot show:

  1. Persistence. RHDP environments expire. This one does not. You can build a demo on Monday and show it on Friday without re-provisioning.
  2. The install. RHDP gives you a running cluster. A customer evaluating edge wants to see how the cluster gets there — the ABI ISO, the unattended boot, the Day 0 operators appearing without intervention.
  3. The physical thing. A NUC on the table is a different conversation than a URL on a slide. "That box is running everything you just saw" is a sentence you cannot say with RHDP.

If you want to run it

Two repos, both public, both cloned to your laptop:

git clone https://github.com/ericcames/image.builder.pipeline.git
git clone https://github.com/ericcames/sales.demos.git

Then follow the run sheet — it is a step-by-step checklist with exact commands.

Prerequisites you will need:

  • A Red Hat pull secret from console.redhat.com
  • An SSH key pair
  • x86_64 hardware with at least 32 GB RAM, 120 GB disk, 8 CPUs
  • A laptop on the same network running Fedora or RHEL (for DNS)
  • A USB drive (8 GB minimum)