Kubermatic Virtualization for Kubermatic Kubernetes Platform Workloads
Overview
See Kubermatic Virtualization (KubeV) run as the infrastructure layer under the Kubermatic Kubernetes Platform (KKP) on your own hardware. We install KubeV (kube-ovn for networking, KubeVirt for compute) on your hosts and build KKP on top: the master/seed cluster on VMs in a management subnet, and user clusters whose worker nodes are VMs in their own subnets. The Online Boutique microservices demo then proves the stack end to end.
Goal
Hard evidence that KubeV and KKP together give you a complete, isolated Kubernetes platform on bare metal, with results against your success criteria and a list of production go-live prerequisites.
Who it is for
Platform teams that want KKP on their own hardware without a separate hypervisor layer.
Content
- 01
Design validation and sizing
Topology, VLAN and IP plan, KubeV sizing and the KKP subnet layout.
- 02
Platform installation
KubeV with a three-node etcd control plane.
- 03
Networking, isolation and storage
VPC, subnets, IPAM, VLAN attachment, isolation rules and live-migration-ready storage.
- 04
KKP deployment
HA KubeOne cluster on VMs and KKP with your identity provider.
- 05
User clusters on VMs
KubeV registered as a KKP provider, self-service clusters.
- 06
Demonstration
Online Boutique, an 11-service app under generated load.
- 07
Touchdown
Enablement, results and go-live prerequisites.
Typical Activities
- Design validation workshop with your platform and network teams
- Sizing and hardware bill of materials from the KubeV Sizing Tool
- Network policy so the only path between user clusters and management is the one KKP needs
- KKP integration with OIDC or LDAP/AD
- User cluster create, scale, upgrade and delete via UI and API, with no orphaned VMs, volumes or IPs
- Live demos: scaling under load, self-healing, persistent storage, service publication and observability
- Workload kept running while its VM is live-migrated between hosts
Deliverables
| Area | What you receive |
|---|---|
| Design and sizing | Topology, sizing record, hardware bill of materials, frozen success criteria |
| Platform | HA KubeV installation with Git-managed configuration |
| Networking and storage | Configured networks and isolation, live-migration-ready storage, as-built network docs |
| KKP | KubeOne cluster carrying KKP, KubeV registered as an infrastructure provider |
| User clusters | Working user clusters on VM worker nodes |
| Demonstration | Online Boutique scenarios shown and recorded |
| Closure | Runbooks, results report, go-live prerequisites |
Prerequisites
- Hypervisor hosts racked and powered, with VT-x or AMD-V enabled
- Host OS installed (Ubuntu 24.04 LTS or agreed alternative)
- Storage backend with ReadWriteMany or Block volume mode
- VLANs trunked to every host, LACP/MLAG and out-of-band management
- IP addressing and firewall rules in place
- VPN or jump host access and a shared Git repository
- Identity provider reachable from the management subnet
- A project manager plus platform and network counterparts
Why this Accelerator?
- One stack, from metal to Kubernetes, proven on your hardware.
- Isolation by design: management and user clusters separated at the network level.
- A realistic workload that tests scaling, failure recovery and live migration.
- Your success criteria, frozen up front for a clear yes or no.
- A clear path to production via the Kubermatic Virtualization for Kubermatic Kubernetes Platform Workloads (Production) accelerator.
What's not included
Production use or SLA, your own applications, production hardening and backup, VM migration from other hypervisors, multi-site rollout, physical infrastructure and fabric routing. KKP licensing is separate.
Optional extensions
Kubermatic Virtualization for Kubermatic Kubernetes Platform Workloads (Production), VM migration onto KubeV, multi-region rollout, KubeLB, GPU or SR-IOV, and managed 24/7 operations.