Use cases
Production-derived scenarios from the blueprint library. Each maps to a docs showcase and an Academy track.
These are not demos. Each use case corresponds to a real blueprint that runs in production, produces structured evidence, and has an associated Academy track for hands-on training.
Cost-optimised SaaS backend
Run steady-state application databases on bare-metal on-prem. Use GCP Cloud SQL replication as the warm standby, activated only during a DR event. Ship costs 60–80% below an always-on cloud equivalent.
onprem/authoritative-foundation@v1dr/postgresql-ha-failover-gcp@v1
- PostgreSQL HA with Patroni leader election on-prem
- pgBackRest backup repo on GCS — incremental, encrypted
- Cloud SQL DMS managed replication as standby
- DNS cutover via the decision service on threshold breach
PostgreSQL HA with automated failover
Three-node Patroni cluster with etcd quorum, pgBackRest backup, and rehearsed failover to a cloud replica — with full evidence output at every step.
onprem/postgresql-ha@v1dr/postgresql-ha-failover-gcp@v1dr/postgresql-ha-failback-onprem@v1
- Automated leader election and replica promotion
- Health probe verification before and after failover
- Evidence envelope produced on each drill execution
- Failback blueprint returns primary to on-prem
WAN edge and cloud hub
Hetzner-hosted WAN edge pair with floating IP, BGP route exchange to a GCP hub, HA VPN tunnels, and DNS-based split routing. Full hybrid connectivity with documented failure-domain behaviour.
networking/wan-hub-edge@v1
- Hetzner edge pair with floating IP for HA failover
- BGP peering between on-prem and GCP Cloud Router
- IPsec / WireGuard VPN tunnel commissioning
- DNS split-horizon for on-prem and cloud name resolution
Lab and training environments
Provision a full HybridOps lab environment from a single blueprint run: NetBox for IPAM, a management network, and seeded inventory. Consistent, repeatable, and evidence-producing — same as production.
onprem/bootstrap-netbox@v1
- NetBox as authoritative IPAM from day one
- Management network and inventory seeded automatically
- Academy lab exercises run against this foundation
- Tear down and rebuild in minutes — no manual steps
Burst-ready on-prem platform
RKE2 Kubernetes cluster on bare-metal with ArgoCD GitOps. The decision service monitors load thresholds and triggers cloud burst when capacity is needed — returning to on-prem when load drops.
onprem/rke2@v1platform/k8s/argocd-bootstrap@v1
- RKE2 cluster provisioned and joined via a single blueprint run
- ArgoCD bootstrapped and pointed at the GitOps repo
- Decision service threshold triggers cloud burst activation
- Cost guard policy prevents runaway cloud spend on burst
DR failover and failback cycle
How use cases 01 and 02 execute end-to-end — from backup replication to DNS cutover and return.
WAN edge topology
The network layer underpinning use case 03 — Hetzner edge pair, BGP peering, and HA VPN tunnels to GCP.
Start with the quickstart or browse all blueprints
The quickstart deploys a working NetBox instance and management network in under 10 minutes. Blueprints are tarball-safe — no internet access required at execution time.