Who responds when the server stops working?
Illustrative scenario: a studio owner finds a cheap server and wants to run n8n on it. A teammate can build workflows, but nobody has restored a database or handled a server outage. The subscription saving is visible; operating effort is missing from the budget.
Self-hosting means the application runs on infrastructure you or your provider are responsible for. The decision is therefore also an agreement about who handles availability, updates, access, and recovery.
What belongs in the budget
| Item | What to include |
|---|---|
| Infrastructure | Server, storage, database, and required backups |
| Initial setup | Installation, security, app connections, and operational handover |
| Routine administration | Updates, failure monitoring, storage checks, and backup checks |
| Recovery | Restore exercises and actual incident response |
| Features and licensing | Required capabilities and permitted use under the current plan and license |
Compare routes with the same people, workflows, and availability requirements. Cloud still needs someone to maintain workflows and app connections. Self-hosting adds infrastructure administration to that work.
A workflow export is not a complete backup
n8n’s recovery documentation distinguishes exports from a full-instance backup. Workflow exports alone do not preserve the complete operating state. Restoring credentials also depends on the encryption key; additional data and configuration depend on the deployment.
Ask the administrator to demonstrate recovery into a separate test instance, including required connections. A file named “backup” is insufficient. Keep backups and keys in approved secure storage, rather than a client-shared folder.
An illustrative monthly budget
Spread initial setup over the same comparison period for both routes. This original example covers the operating portion of costs, rather than a provider quote or the full cost of every workflow. Allocating setup across months does not make its upfront payment an installment plan.
Valued staff time is not automatically an additional cash expense. An existing infrastructure administrator may have usable capacity, but still needs allocated time and backup coverage. For Cloud, enter the actual quote, workflow maintenance, and relevant limits. Do not compare it only with the $25 server line.
When self-hosting is reasonable
Consider it when you already have an experienced administrator, a verified recovery procedure, or a specific requirement the cloud route cannot meet. For a small nontechnical team without that support, start with a managed service or a simpler existing-app route. A lower server bill alone is not a sufficient reason.
Before ordering, ask who updates the system, how they detect outages, what they can restore, how long recovery takes, and who covers their absence. n8n’s monitoring documentation distinguishes server availability from database readiness; neither alone proves a particular workflow delivered its result. This is a decision article, not an installation tutorial or hosting-provider recommendation.
Recovery and operating documentation reviewed October 1, 2026. The budget is illustrative. Check the current license, feature availability, and quote for your intended use.