What is TractorStack

Planned design. TractorStack is in development. This page describes how it is designed to work; nothing here is available to install yet. See the roadmap for the status of each part.

TractorStack is in development. Nothing described here is released or installable yet. These pages describe the planned design, and they will change as the work moves forward.

TractorStack is an on-premise platform being built by InfraTractor. It is a set of tested building blocks for running applications and data on servers you own: one server running Docker Compose at the start, and Kubernetes later if and when you need it.

Every part is a well-known open-source tool, configured as code in a Git repository you keep. The design needs no cloud account and no subscription to our servers to run.

Who it is for

TractorStack is aimed at organizations that run, or want to run, their own hardware:

Small and mid-sized organizations with a few applications and a database, and no appetite for a full platform team.
Managed service providers who look after several customers and want each one built the same way.
Regulated businesses and defense suppliers that need data to stay on their own premises and want evidence of how systems are configured.
Anyone running servers in their own building or in a Tier III colocation facility.

If your workloads already live comfortably in a public cloud, TractorStack probably isn’t for you.

The core idea

Start standalone. Grow to Kubernetes when you actually need it.

Most organizations don’t need a cluster on day one, and many never will. Stage 1 runs everything on one server with Docker Compose, but it follows the same rules a Kubernetes deployment needs: no state on local disk, configuration from environment variables, health checks, pinned images and backups to S3 storage. That way, moving up is a change of runtime rather than a rewrite, and your data doesn’t have to move in a hurry.

The same repository and the same tools carry through every stage. OpenTofu builds the virtual machines and networks, Ansible configures the hosts and services, and your applications run on Compose or Kubernetes depending on the stage.

What “in development” means

Each part of TractorStack has one of three statuses:

In design: being specified and built next.
Planned: committed, and built after the stage before it.
Exploring: being evaluated, and may change or be dropped.

When these docs say TractorStack “will” do something, read it as the current plan. When they describe what PostgreSQL or OpenTofu can do, that is about the tool itself, which you can check today.

How to read these docs

Architecture explains the six layers, the per-customer Git repository and how a change is reviewed and applied.
Roadmap and status lists every stage and cross-cutting item with its current status.
The stage pages go into detail: Stage 1: Standalone, Stage 2: Resilient standalone, Stage 3: Kubernetes for applications and Stage 4: Everything on Kubernetes.
Moving between stages covers when to move up, and how each component moves.
Kubernetes-ready rules lists the habits that make moving up possible.

If you only have ten minutes, read the architecture page and the stage 1 page.

Becoming a pilot

We are looking for a few organizations to pilot TractorStack on their own hardware, with InfraTractor engineers working alongside your team. A pilot means you get the early design applied to your real environment, and we get honest feedback about what works and what doesn’t.

To talk about a pilot, email sales@infratractor.com with a short description of what you run today: how many servers, which hypervisor, which applications and databases, and any rules your data has to follow.

Last updated 25 October 2026 · Questions or corrections: sales@infratractor.com