ZeusK8s
How it works

Four steps from "this cluster makes me nervous" to "I know exactly what happens when it fails."

No YAML to hand-edit, and nothing you can't inspect. Here's the actual path, and what Zeus is doing under each click.

01

Connect your infrastructure

Most teams reach a point where their cluster exists but nobody is confident it's actually production-ready. Start here: link an AWS or GCP account, or drop a single agent onto your own servers — bare metal, a Proxmox box, a rack in a colo. The agent dials out, so there are no inbound ports to open and nothing to expose. From that moment, that hardware is just another place Zeus can put a cluster.

Zeus · Providers
Connect infrastructure
Link a cloud account, or drop an agent on your own hardware.
AWS linked
EKS · accounts
GCP linked
GKE · projects
Proxmox linked
k3s · VLANs · VMs
Bare metal
dial-out agent
02

Provision a cluster, the same way every time

Pick where it lives and what it's for. Zeus chooses a sane CNI, node groups, encryption, and topology, then builds it. The flow doesn't change between EKS, GKE, and your own metal, so the thing you learned on cloud is the thing you already know on hardware. Every cluster lands in one list.

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
03

Join clusters into one fabric

Select two or more clusters and link them. Zeus wires up the encrypted cross-cloud overlay that usually takes weeks to wire by hand. Now a workload in one cluster reaches a service in another by name, by its real IP, encrypted end to end, and you can finally see how it all connects in one view.

Global fabric
AWS · GCP · private cloud · self-healing
Mesh healthy
pdx-prodAWS · us-west-2sjc-prodAWS · us-west-1den-edgeProxmox · private clouddfw-prodGCP · us-south1oma-prodGCP · us-central1chs-prodGCP · us-east1iad-prodAWS · us-east-1
Seven clusters, one system
AWS, GCP and your own hardware — every request served by the closest healthy region.
04

Put your data everywhere it's needed

Most DR plans are a Notion doc nobody has tested. Zeus changes that: deploy MySQL, PostgreSQL, ClickHouse, or Prometheus and tell it to go global. Zeus places it across the fabric and keeps it replicated, with backups and point-in-time restore configured from the start. Your application keeps talking to an ordinary database endpoint. The geography is no longer your problem.

Zeus · Infrastructure · MySQL (global)
orders-db · MySQL
Global · 2 regions · async replication
Replicating
AWS · us-east-1 Primary
writer
orders-db.zeus.internal ✓ accepting writes
41ms lag
GKE · europe-west3 Replica
reader
orders-db-ro.zeus.internal ✓ in sync
Writes / s
2,480
us-east-1
Reads / s
9,140
europe-west3
Replica lag
41 ms
p99 · healthy
Under the hood

The full surface, not the highlight reel.

Zeus is a working control plane, proven on real hardware, not a wrapper around a single script. This is what's actually in the box.

Provisioning

  • Multi-cloud, multi-account
    AWS and GCP with per-cluster account binding, plus bare-metal and Proxmox/k3s through a dial-out agent. One console, every cluster you own.
  • Intent-first wizards
    You say what you want to run; ZeusK8s picks sane CNI, node groups, encryption, and topology. The knobs are still there under Advanced.
  • Node groups & autoscaling
    EKS-style managed node groups on every flavor, including k3s, with labels, taints, and live scale up/down over the agent tunnel.
  • Clean teardown
    Destroys sweep tagged resources with a keep/delete checklist (EBS, KMS, SGs, load balancers) instead of leaving orphans and a surprise bill.

The global fabric

  • Cross-cluster service discovery
    Services resolve and connect across clusters and clouds by name, over an encrypted overlay, routed to real pod IPs with no NAT in the middle.
  • Default-deny by design
    A freshly connected cluster can reach nothing and is reachable by nothing. Every cross-cluster flow is an explicit, directional grant you can revoke in a click.
  • WAN-aware ingress
    Pin egress IPs, route ingress to the nearest region, and bring traffic in on a stable edge, all without hand-editing a load balancer per cluster.

Stateful data

  • Global database replication
    Multi-region replication for MySQL, PostgreSQL, and ClickHouse, wired up at deploy time. Your apps keep talking to the same endpoint; the geography is ZeusK8s’ problem.
  • Backups & point-in-time restore
    Scheduled backups to S3-compatible storage and PITR restores, including standing up a fresh cluster from a backup.
  • Sizing that isn’t a guess
    Memory-budget and sizing calculators built in, so you provision databases on evidence instead of doubling everything and hoping.

Day-two operations

  • Guided upgrades
    End-of-life banners, preflight checks, and a stepped control-plane → addon → node-group upgrade flow. No more cluster-version roulette.
  • Live pods, logs, and metrics
    Lens-grade workload views, multi-pod log tailing, and metrics pulled straight from Prometheus, without leaving the tool.
  • Infra add-ons as first-class
    Helm-based add-ons (ingress, cert-manager, databases, observability) tracked per cluster with a real config UI, not a values.yaml you paste and pray.

Want to see the four steps run live, end to end?

Book a demo