ZeusK8s
Private cloud

Your own hardware, run like a cloud.

Turn a Proxmox host or a rack of servers into managed Kubernetes: provisioned, meshed, and operated with the same console and workflow as your cloud clusters.

The cloud experience, on infrastructure you own.

There are good reasons to run your own hardware: cost at scale, data residency, latency, or simply already owning the machines. The usual price is hand-rolling everything yourself — 23 YAML files, a Helm chart nobody fully understands, and one engineer who is the single point of failure for all of it. That's the hidden cost most teams only discover after the first outage.

ZeusK8s removes that tax. Provision k3s clusters on Proxmox with a dedicated, exposed control plane and EKS-style worker node groups, then join those clusters to the same encrypted overlay as your EKS and GKE clusters. A lightweight agent dials out from your hardware, so there are no inbound ports to open and nothing to expose to the internet.

Zeus · Clusters
Clusters
6 clusters · 3 providers
Live New Cluster
NameProviderRegionVersionNodesStatus
production-usAWS EKSus-east-11.306Ready
production-euGKEeurope-west31.304Ready
stagingAWS EKSus-west-21.29→1.30 ⚠3Ready
edge-apacGKEasia-east11.302Ready
HQ Denverk3sDenver, COv1.35.55Ready
prox-test01ProxmoxProvo, UTv1.34.1Provisioning
How it works

From zero to running.

01

Drop in the agent

Install one static agent on your Proxmox host or servers. It dials out over an encrypted tunnel, with no inbound firewall changes and no public endpoints.

02

Provision k3s clusters

Zeus stands up VMs on Proxmox, installs k3s with a dedicated tainted control plane (1/3/5 nodes) and worker node groups, ready to run workloads.

03

Join the cloud

Connect private clusters to the same encrypted overlay as EKS and GKE. Cross-cluster, cross-site service discovery just works.

04

Operate identically

Scale node groups, upgrade, deploy services, and watch pods live, all from the same console you use for cloud clusters.

The specifics

Built by people who run this in production.

No hand-waving. Here’s what’s actually under the hood: the kind of detail you’d expect from a platform you’re going to trust with production.

Hypervisor
Proxmox VE: VM lifecycle managed via a connector agent
Distribution
k3s: dedicated control plane + worker node groups
Networking
Standard CNI; kube-vip control-plane VIP for HA
Connectivity
Dial-out agent: no inbound ports, no exposure
Storage
CSI over the node LAN; raw-disk pool init (LVM/ZFS/NFS)
IP planning
Static cloud-init IPs + Network Plans to avoid collisions
UniFi integration
Optional: auto-create cluster VLANs + firewall zones
HA
PVE HA rules, CRS, affinity, managed from the console
Straight answers

Questions you’d actually ask.

Do I have to expose my servers to the internet?

No. The agent dials out to ZeusK8s over an encrypted tunnel. There are no inbound ports to open and nothing public, a key reason teams run private clusters in the first place.

Is this real Kubernetes?

Yes. It’s k3s, a fully conformant, CNCF-certified Kubernetes distribution. Workloads, manifests, and tooling are standard.

Can private clusters share data and services with my cloud clusters?

Yes. They join the same encrypted overlay, so a service on your hardware and a service in EKS reach each other by name and real IP. Global databases replicate across the boundary too.

What about networking and firewalls on-prem?

Zeus has an optional UniFi integration that creates per-cluster VLANs and firewall zones automatically, or you can wire networking yourself and Zeus stays out of the way.