Dedicated servers for n8n
Bare metal for high-volume automation and local AI
Servers for heavy n8n automation: high webhook volume, file processing and AI agents on local models. Run n8n in queue mode on one powerful server, connect a GPU server for local models over a private VLAN, and keep every workflow, credential and execution on your own hardware.
-
Mirrored storage against drive failure -
NVMe for PostgreSQL -
No noisy neighbors -
GPU servers for local AI
Delivery from 3 hours
Unmetered traffic available
Root access and IP-KVM
IPv4 and IPv6 included
n8n hardware requirements
n8n is not CPU-intensive, and memory runs out first: an idle instance needs about 100 MB, but Code nodes and large payloads copy data in memory. A single instance with PostgreSQL runs well on 16–32 GB, and workflows with files and big data need 64 GB or more.
Under heavy webhook load, a single n8n process starts dropping requests. In n8n's own benchmark, queue mode on a 16 vCPU, 32 GB server held 162 requests per second with no failures, so for high webhook volume plan 8 cores, 16 threads and 32–64 GB of RAM.
Webhooks need an address that never changes, and many APIs accept calls only from allowlisted IPs. Every server comes with a static IPv4 address and IPv6, and an unmetered port option carries constant webhook and API traffic without limits.
The database holds workflows, credentials and the full execution history, and it grows with every run. SQLite is the default for a single instance, while queue mode needs PostgreSQL: run it on NVMe and turn on execution pruning so the history does not fill the disk.
n8n itself needs no GPU when agents call remote models such as OpenAI or Claude. For local models, a GPU server runs Ollama, which n8n supports natively, or vLLM for higher throughput, and VRAM sets how large a model fits on the server.
While n8n is down, webhooks and scheduled runs that fire in the meantime are lost and never recovered. With two NVMe drives in RAID 1, a failed drive does not stop n8n: the server keeps running, and the replacement, which needs a short shutdown, is scheduled for a quiet hour.
Pricing of dedicated servers for n8n
View all configurations
Rent a dedicated server for n8n from INTROSERV
How to order a dedicated server
Select a configuration, add NVMe drives, choose the OS, wait until the server is ready and install n8n.
Read the documentation
Installing n8n with PostgreSQL on a dedicated server
Step-by-step installation of self-hosted n8n with Docker Compose, PostgreSQL and HTTPS behind a reverse proxy.
Read the tutorialServer, setup and backups for n8n from one provider
-
1
Install n8n with PostgreSQL and Redis
Install n8n with Docker Compose, PostgreSQL and Redis behind a reverse proxy with HTTPS. Keep a copy of the encryption key: without it, stored credentials cannot be restored.
-
2
Run in queue mode from the start
Run n8n in queue mode, with worker processes on the same server and a concurrency of 5 or more. Production runs go to the workers, and the editor stays responsive under load.
-
3
Turn on execution pruning
Set how long n8n keeps finished executions. Without pruning, the history in PostgreSQL grows with every run until it fills the NVMe drives and slows the database down.
-
4
Back up with the INTROSERV Backup Service
Back up the whole server with the INTROSERV Backup Service – NAKIVO software and storage in one ready-made service. Restore the entire server or a single file when needed.
n8n licensing and support
Licensing
The n8n Community Edition is free to self-host under the Sustainable Use License, a fair-code license that allows use for your own internal business purposes but not reselling n8n or hosting it for your customers. Business and Enterprise plans add SSO, shared projects and Git version control with a license key sold by n8n, not by INTROSERV.
Support
Support for n8n itself comes from the n8n community forum and documentation, or from n8n with a paid plan. INTROSERV takes care of the Linux servers n8n runs on: monitoring, security, OS updates and incident fixes with a server management plan, and one-time tasks as hourly system administration.
FAQ
-
Do I need a dedicated server for n8n?
Not for a few dozen workflows: n8n is light, and a VPS handles that well. A dedicated server pays off with sustained high webhook volume, workflows that process files, long execution history, or local AI models that need a GPU.
-
How fast is the server delivered?
Instant configurations are ready within 3 hours of payment; custom builds usually take up to 24 hours. The delivery time is shown on each configuration in the catalog.
-
Can the server be tested first?
Yes. A paid test period of up to 5 days is available, enough to install n8n, import your workflows and run them against real data.
-
Do I need a GPU for AI workflows?
Not when your agents call remote models such as OpenAI, Claude or Gemini: n8n only sends the requests. A GPU server is needed when models run locally, for example with Ollama, so prompts and data never leave your infrastructure.
-
Can I use my own domain and HTTPS certificate?
Yes. Point a DNS record at the server and run n8n behind a reverse proxy such as nginx, Caddy or Traefik with a Let's Encrypt certificate. Set the webhook URL to the same domain, so external services call the right address.
-
Can storage be expanded later?
On DC1, DC2 and DC3 configurations, yes: drives for free slots are added by a support ticket, and the monthly price changes accordingly. For other configurations, it is best to choose the drive layout with some headroom at order time.
-
What happens when a drive fails?
With two drives in RAID 1, n8n keeps running on the remaining drive. You open a support ticket, INTROSERV replaces the drive in the data center, and the array rebuilds.
-
Can I move n8n to a bigger server later?
Yes. Order the new server, export workflows and credentials with the n8n command line, import them on the new server and keep the same encryption key, so stored credentials still work.
