Iapetus is in beta — free Pro for 3 months, no credit card. Get started free →
Iapetus
Now in beta

Several real machines. Real networks. On your laptop.

Iapetus gives you several real Linux VMs on your workstation, wired into networks you control — deploy across them, cut a node's network, throttle a NIC, and prove your system recovers. Disposable, repeatable — one tool instead of Vagrantfile + shell scripts + tc + hope.

$ curl -fsSL https://get.iapetus.dev | bash

Registering gets you a free Pro license for 3 months — no credit card. Runs on Ubuntu 24.04+ — other distros coming soon.

topology — kafka-broker-loss.scenario fault active
23 ms kafka-1 ubuntu-24.04 kafka-2 ubuntu-24.04 node isolated zookeeper debian-13
kind and docker-compose give you containers sharing one kernel and a network you can't break.
Iapetus gives your services real machines with real NICs — then lets you cut the cable.

The lab

Build it. Break it. Watch it survive.

One tool for designing, running, and fault-testing realistic distributed environments — on your own workstation, not a shared staging cluster.

Real Linux VMs

Create, snapshot, and reset Ubuntu and Debian VMs in seconds. Cloud-init provisioning from profiles — the default lab profile ships Docker and dev tooling on every machine; slimmer profiles are one click away.

Container orchestration

Deploy Docker Compose stacks to any VM or the host, backed by a built-in image registry and nginx reverse proxy.

Real networks, real failures

Your VMs have actual NICs, so faults are real: inject latency, jitter, packet loss, and bandwidth caps, or cut a node off the network entirely — tc and iptables under the hood.

Playbooks & fault drills

Save fault sequences — cut a node, restore, throttle, wait — with reusable command templates, and replay them identically every run. A failed run unwinds the faults it applied.

Agents drive the same surface you do

UI, HTTP API, and MCP share the same core surface. An AI agent can design, deploy, fault-test, and tear down the whole topology through the same API you use.

Your laptop, not a cloud

Local-first by design: licenses verify offline, environments live and die on your hardware, and it all works with the WiFi off.

Workflow

From zero to a broken network in three steps

Three steps from an empty workstation to a multi-machine lab you're allowed to break — no cloud account, no ticket to a shared staging cluster.

  1. 01

    Spin up real VMs

    Launch Ubuntu or Debian VMs in seconds, each provisioned with cloud-init from a profile you choose — the default lab profile includes Docker — ready to connect.

  2. 02

    Wire the topology

    Connect VMs into bridged subnets that mirror the environment your system actually runs in.

  3. 03

    Break it on purpose

    Dial in latency and loss, isolate a node, block its internet — and watch how your services behave while playbooks replay the drill identically.

Platform support: the Iapetus app currently runs on Ubuntu 24.04+ — other distros coming soon. Guest VMs can run Ubuntu Server 22.04, 24.04 & 26.04 and Debian 13.

Would your system survive losing a node?

Find out before production does. Registration takes a minute and gets you a free Pro license for 3 months.