Many clusters. One private network.
ZeusK8s joins your clusters over an encrypted WireGuard overlay, so a service in one cluster reaches a service in another by name, by its real IP, as if they shared a switch — across clouds, regions, and your own hardware.
A global network, without a six-month project.
Connecting clusters across three clouds and a rack of your own hardware normally means VPNs, peering, NAT gymnastics, and IP ranges that collide the moment you add the second cluster. It’s the part of multi-region that quietly eats a quarter of an engineer’s year.
ZeusK8s makes it a toggle. It allocates every cluster a non-overlapping pod and service range up front, connects them over an encrypted overlay that dials out (no inbound ports, no public exposure), and routes traffic to real destination IPs with no NAT in the middle. You get one map of what can reach what, and a default that’s safe.
From zero to running.
Non-colliding IP space, up front
Network Plans allocate each cluster’s pod and service CIDR from a global pool, so no two clusters ever overlap. The thing that quietly breaks every multi-cluster setup is handled before it bites.
Connect a cluster to the overlay
Enroll a cluster and Zeus mints its credentials, installs a tiny in-cluster routing helper, and advertises its internal ranges over an encrypted WireGuard tunnel. It dials out, so there’s nothing inbound to open.
Grant scoped access
Connecting grants nothing on its own — a fresh cluster is fenced. You explicitly allow one cluster to reach another, in one direction, optionally limited to ports, and revocable in a click. The map is exactly what’s enforced.
Reach services by name
Every cluster gets a mesh DNS name, so a pod in one cluster resolves a service in another and connects to its real IP, encrypted end to end. No load-balancer hairpins, no manual wiring.
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.
Questions you’d actually ask.
What exactly is the “global fabric”?
An encrypted overlay network that Zeus configures and operates for you. Your clusters stay separate Kubernetes clusters, but a service in one can reach a service in another by name and real IP, as if they were on one network.
Is it a service mesh like Istio or Cluster Mesh?
No. There are no sidecars and no per-cluster mesh control plane to operate. It’s a WireGuard overlay: clusters connect at the network layer, traffic is encrypted, and routing uses real IPs because Zeus guarantees the ranges never collide.
Does connecting a cluster expose it?
No. The routing helper dials out over the encrypted tunnel — there are no inbound ports to open and nothing public. And connecting grants no access by itself: every cross-cluster flow is an explicit grant you can revoke.
What keeps the links up?
Tunnels stay warm so the first packets of a conversation aren’t dropped (important for realtime voice, video, and streams), the in-cluster helper re-registers automatically after a network blip, and Zeus reconciles the mesh in the background so node changes heal on their own.
Keep exploring.
All features →Ingress, policy, and DNS on the fabric.
Explore →Global data over the overlay.
Explore →See every cross-cluster flow.
Explore →Start running it today.
Spin up your first cluster free, or get a guided tour from our team.