Hybrid Mesh Platform
One Git repo governs three OpenShift clusters — secure connectivity, fleet GitOps, AI Computer Vision at the Edge, and optional factory telemetry across hub and spokes.
Primary demo (v2.2+): NeuroFace full stack on east/west spokes — face detection (OVMS ModelMesh), PPE safety (YOLO/KServe), Kafka events, Grafana — federated via Skupper and Gateway API 50/50 from the hub (
neuroface,neuroface-cvroutes).
Optional: Industrial Edge (MQTT → Kafka → line-dashboard, sensors) is disabled by default. Uncomment IE apps in
charts/region/*/values.yamlto enable.
Multi-cluster GitOps using Red Hat Advanced Cluster Management, OpenShift GitOps (Argo CD), ambient Service Mesh, Connectivity Link (Kuadrant), Skupper, Grafana, ACS, Developer Hub, and OpenShift AI.
AI Computer Vision at the Edge (primary)
| Surface | URL | What it shows |
|---|---|---|
| NeuroFace app | https://neuroface.<hub-domain>/ | Full UI — 50/50 east/west via hub Gateway |
| NeuroFace CV | https://neuroface-cv.<hub-domain>/ | PPE gateway (YOLO on spokes) |
| Developer Hub | https://developer-hub.<hub-domain>/create | AI Computer Vision at the Edge template |
| Grafana | https://grafana.<hub-domain>/ | NeuroFace east/west + hub gateway metrics |
Deep dive: NeuroFace & CV journey · Workshop modules 13–16
Quick links
| Topic | Page |
|---|---|
| Architecture | Architecture |
| Install (ACM-first) | Getting Started |
| RHDP catalog orders | RHDP field content |
| Region paths (hub/east/west) | Region strategy |
| ACM + GitOps | Deploy with ACM and GitOps |
| GitOps chain | Deployment chain |
| Fleet domain sync | fleet-values-sync |
Secrets (values-secret.yaml) | Secrets configuration |
| Red Hat products | Products index |
| Scaffolding | Scaffolding |
| Industrial Edge (optional) | Industrial Edge |
| Troubleshooting | Troubleshooting |
| Workshop Showroom | Hybrid Mesh AI Workshop |
Cluster sizing
| Role | Recommended | Minimum demo |
|---|---|---|
| Hub | 4 × 16 vCPU × 64 GiB | 3 × 8 × 32 GiB |
| Spoke (CPU) | 3 × 8 vCPU × 32 GiB | 2 × 4 × 16 GiB |
| Spoke (GPU, optional) | 3 × 8 vCPU × 32 GiB + 1× T4/A10G | — |
Full tables, workload budgets, and GPU operators: Bill of Materials — Cluster sizing.
Verify: bash scripts/verify-node-capacity.sh · ROLE=spoke bash scripts/verify-node-capacity.sh
Reference
| Topic | Page |
|---|---|
| Bill of Materials | Operator versions |
| Validation Guide | Verify deployment |
| Support Policy | Community support |
Repository layout
| Path | Purpose |
|---|---|
charts/region/hub/ | Hub cluster bootstrap + clusterGroup values |
charts/region/east/ | East spoke bootstrap + clusterGroup values |
charts/region/west/ | West spoke bootstrap + clusterGroup values |
charts/all/ | Cross-cluster Helm charts (50+ charts) |
values-global.yaml | Pattern-wide globals |
overrides/ | Platform-specific value overrides |
docs/ | GitHub Pages documentation |
Key charts (default deploy)
| Chart | Purpose |
|---|---|
spoke-neuroface | Full NeuroFace stack on spokes + OVMS ModelMesh + spoke Gateway |
spoke-neuroface-cv | YOLO PPE InferenceService, Kafka, MinIO data path |
neuroface-gateway | Hub routes 50/50 east/west + CV gateway |
acm-operator | ACM MultiClusterHub installation |
acm-hub-spoke | Spoke registration + GitOpsCluster |
developer-hub | Catalog, software templates, scaffolder |
fleet-values-sync | Cross-cluster domain sync |
Pattern repo: github.com/maximilianoPizarro/hybrid-mesh-platform
Next → NeuroFace & AI CV