n8n self-hosted: what it takes to run n8n on your own servers

Every rule and price on this page comes from n8n’s own documentation, license pages and pricing page, read on October 1, 2026. The license FAQ in particular gets new examples over time, so check it before you rely on a borderline case. To compare n8n with Make and Zapier first, see our guide n8n vs Zapier vs Make.

n8n self-hosted or n8n Cloud?

Both run the same core product. The choice is about who runs the server, where the data lives and which features you need. n8n’s documentation is frank about the trade: it “recommends self-hosting for expert users”, because “mistakes can lead to data loss, security issues, and downtime”.

Prices are those shown on n8n’s pricing page on October 1, 2026. The Business plan is, in n8n’s words, “currently only available for self-hosted n8n”, and companies under 20 employees can check whether they qualify for a Start-up Plan at 50 percent off Business.

One line deserves attention before anything else. In the free Community edition, n8n’s edition comparison says that “only the instance owner and the user who creates them can access workflows and credentials”. For a team of several builders, that alone can tip the choice.

What the n8n license allows, in plain terms

The free self-hosted edition is covered by the Sustainable Use License, which n8n created in 2022 and describes as “fair-code”. It is not an open source license, and n8n does not call itself open source. The license FAQ gives the simplest test: “If workflows are only created or modified by you or people in your organization, you can use n8n under the Community license for free.”

The license allows use “only for your own internal business purposes or for non-commercial or personal use”, distribution only free of charge for non-commercial purposes, and no removal of licensing notices. The FAQ then lists concrete cases.

The test is about control of the workflow logic. In n8n’s words, “the key question is not whose credentials are used, but who controls the workflow logic”. And for agencies, the FAQ adds that installing and maintaining n8n on a client’s infrastructure is fine “as long as you don’t offer hosting as well, which would compete with n8n’s own Cloud offering”.

How we apply it. Our n8n agency installs n8n in your infrastructure, or in an instance dedicated to your company in a cloud account in your name, and can operate it for you. We never put a client on a shared platform. When your use calls for a paid license, the Business or Enterprise plan or a commercial agreement with n8n, we take it, and the quote includes it at n8n’s public rates. When a case is unclear, n8n asks you to write to [email protected], and so would we.

Community, Business or Enterprise: what each edition adds

Every self-hosted install runs the same software; a license key turns on extra features. Without a key, n8n runs as the Community edition, which its documentation describes as “almost the complete feature set”.

Two details from n8n’s pricing page are worth knowing before you buy. A Business or Enterprise license key “must ping our license server daily to stay active”, and that ping includes the number of production executions. n8n also collects telemetry by default, which you can switch off. On the Business plan, going over the quota does not stop workflows; overage is billed at 4,000 euros per extra bucket of 300,000 executions.

What a production setup needs

A self-hosted n8n that the business depends on needs more than the one-line install. n8n’s hosting documentation names Docker Compose as the method for “production deployments with databases and additional services”. Here is what we check on every instance.

1. A real database. n8n uses SQLite by default, in the ~/.n8n folder. PostgreSQL is supported, in versions 17 and 18 plus 16 for compatibility as of July 2026, according to n8n’s database page.
2. A persistent encryption key. n8n “creates a random encryption key automatically on the first launch” and uses it to encrypt credentials before they reach the database. Lose it and the stored credentials are unreadable. Keep it in a secret store, and back it up apart from the database.
3. HTTPS and access control. n8n’s security section covers SSL, two-factor authentication, SSO on paid plans, a built-in security audit, SSRF protection, blocking chosen nodes and disabling the public API.
4. Backups you have restored once. The database, the encryption key and the persistent volume, with a restore tested on a spare machine.
5. Monitoring and an error workflow. Execution failures go to a named person; the server itself is watched like any other production system.
6. Room to grow. In queue mode, a main instance receives triggers and webhooks, Redis holds the queue, and workers run the executions. The encryption key must then be shared with every worker.

Where to install it is your call. n8n documents setups on AWS, Azure, Google Cloud, DigitalOcean, Hetzner, Heroku and OpenShift, as well as on your own servers. For a European company, pick a provider and a region that keep the data where your rules say it must stay. Our security page explains how we handle access, keys and data location on client projects.

Updates and upkeep: the part people forget

Self-hosting moves the maintenance from n8n to you. n8n’s Docker page notes that it “releases a new minor version most weeks”, and its update page asks you to “try to update at least once a month”, to read the release notes for breaking changes, and to test the update on a separate environment first.

In practice, a self-hosted instance needs someone who owns it: a monthly update slot, a test copy to try each version on, a check that every workflow still runs after the upgrade, and a person who receives the alerts. When nobody owns that work, the instance drifts several versions behind, and the next upgrade becomes a project.

That is the real cost of “free” n8n. The license costs nothing for internal use; the server, the database, the backups and the hours of upkeep do. Count them before you compare with n8n Cloud.

Self-hosted n8n and AI: keeping the models close

Self-hosting n8n keeps workflow data on your infrastructure, but an AI step still sends its prompt to whichever model it calls. To keep that inside your walls too, run the model yourself.

n8n curates a Self-hosted AI Starter Kit, a Docker Compose template with n8n, Ollama for local models, the Qdrant vector store and PostgreSQL. Its own documentation labels it “for testing only” and asks you to secure and harden it before production. It is a good way to try a fully local agent; a production setup still needs the checks of the previous section.

To see how agents are built inside n8n, with tools, memory and human approval, read our guide to the n8n AI agent. When the model itself has to stay on your servers, our private LLM page covers sizing, hosting and the choice of model. You can also try what a model does with your kind of documents in the knowledge assistant demo.

Your self-hosted n8n, set up and looked after

Imagine your workflows running on your own infrastructure, updated every month, with an alert in the right inbox the moment one fails. In a free 30-minute assessment, tell us what n8n should run and where your data has to stay. We tell you whether self-hosting, n8n Cloud or a paid license fits, and what the setup involves. The scope and price are fixed in writing before we start, and every workflow we deliver ships with an error alert, documentation and accounts in your name.

Book your free 30-minute assessment. We reply within one business day. If your team will run the instance itself, our n8n training takes place on your own setup.