Management model

Choose your management model

Both models run on RKE2 and are priced on the same resource pack basis — Shared or Dedicated CPU. The difference is how much of the operational stack Radibase handles for you.
Self-service cluster
You bring the workloads
We provision the cluster and hand you kubectl access. Everything else is yours to set up using your preferred tools.
✓ Automated RKE2 provisioning — choose master and worker count from the panel
✓ kubectl access via HAProxy — cluster nodes stay on private network, no public IPs
✓ Workload traffic via Caddy load balancer to NodePort — automatic SSL via Let's Encrypt
✓ Second disks on worker nodes ready for Ceph — install the operator yourself
✓ Compatible with FluxCD, ArgoCD, or any GitOps tool
Pricing: Resource pack price only — no added charge for cluster provisioning
Managed Kubernetes
We operate the full stack
We handle the entire operational layer. Your team only needs a Dockerfile and a CI pipeline to deploy to production.
✓ Everything in self-service, plus
✓ ArgoCD + Radibase operator — dependency-aware, ordered deployments
✓ Ceph persistent storage — configured and operated
✓ Vault + ExternalSecrets — secrets deployed automatically, managed via panel
✓ Prometheus + Grafana observability stack
✓ Automatic image updates on new container builds
✓ CI/CD pipeline setup — GitLab runners, registry, build automation
Pricing: Resource packs + contracted fee based on project scope
Architecture

How every cluster is built

All VMs — cluster nodes, load balancers, databases, and helper services — run on the same private client network. Public IPs are assigned only where needed.
kubectl access
HAProxy with public IP fronts the API server. Multiple master nodes supported. Connect from your local machine using kubectl, Lens, or any standard Kubernetes tool.
Workload load balancing
Caddy is the recommended load balancer — automatic SSL via Let's Encrypt, shared certificate storage for multi-region setups. Nginx or other balancers available as alternatives.
Private network
Cluster nodes, databases, cache layers, and any additional services — RabbitMQ, Kafka, Redis — all run on the same private account network. Connect everything with private IPs, expose only what needs to be public.
Resource pack coverage
Packs cover all VMs in your setup — cluster nodes, HAProxy, Caddy, Kubernetes VMs, and helper services. Total pack count depends on your full infrastructure footprint, not just cluster nodes.
VPN access to private-only VMs via OpenVPN or WireGuard — available on request.
Multi-region

Cross-region Kubernetes topologies

Two independent clusters, coordinated at DNS and database layer
Run independent RKE2 clusters in Sofia and Amsterdam. Applications are distributed across regions via DNS — each cluster serves traffic from its region. Cross-region database replication keeps data consistent between regions.
Regional failover is automated via Cloudflare API — switching all traffic from one region to the other is a single triggered operation. Works for VM-based clusters, Kubernetes clusters, or hybrid topologies.
DNS-based routing
Cloudflare API automation
DB replication
Hybrid VM + Kubernetes
Managed stack

What we run in managed tier

All components are proven in production — Radibase itself runs on this exact stack.
ArgoCD GitOps
Ceph storage
HashiCorp Vault
Prometheus
Grafana
Image updater
Helm deployments
ExternalSecrets
GitLab CI/CD

Ready to discuss your cluster setup?

Whether you want a self-service cluster today or a fully managed stack built around your project — we can scope it together.
Book a free architecture review