Edge
n8n logo

For n8n

n8n on Edge,
no per-execution bill

Self-host the workflow automation platform on a real VM with S3-compatible storage and a global CDN. Run unlimited executions, keep your credentials on your network.

# docker-compose.yml — n8n + Postgres

services:

n8n:

image: n8nio/n8n:latest

environment:

DB_TYPE: postgresdb

N8N_HOST: n8n.example.com

N8N_PROTOCOL: https

N8N_BINARY_DATA_MODE: s3

N8N_AVAILABLE_BINARY_DATA_MODES: filesystem,s3

postgres: { image: postgres:16 }

# Front the editor with the CDN

$ edge cdn create n8n.example.com ...

Why teams self-host n8n

Run automation at scale without a per-execution metre running.

Hundreds of integrations

Connect to 400+ services out of the box. Custom HTTP, webhook triggers, AI nodes — n8n does what Zapier and Make do, on your infrastructure.

Node.js + Postgres on a VM

A small VM runs the editor, scheduler and worker. Scale to queue mode (Redis + worker VMs) when execution volume grows.

Binary data to S3

Configure n8n to push binary data (file attachments, exports, generated PDFs) to Edge Storage instead of local disk. Workflows stay portable.

Credentials stay on your network

API keys for Stripe, OpenAI, Salesforce, Slack — they all live in your VM's encrypted credential store. Nothing transits a third-party cloud.

Webhooks behind the CDN

Let inbound webhooks hit the Edge CDN, which forwards to your n8n VM. Caching is bypassed for webhook URLs but you get DDoS protection for free.

No per-execution bills

n8n Cloud charges per workflow execution. Self-hosted on Edge is one fixed monthly bill — run a million executions if you like.

Reference architecture

How n8n maps to Edge

Single VM for low-volume teams, queue mode (Redis + worker VMs) for high-volume — all over private networking.

Compute

n8n (Node.js) + Postgres on a VM, optional Redis + workers for scale

Storage

S3-compatible bucket for workflow binary data and backups

CDN

Optional: webhook ingestion behind the CDN for DDoS protection

DNS

Anycast DNS for `n8n.example.com`

# Queue mode for scale

EXECUTIONS_MODE=queue

QUEUE_BULL_REDIS_HOST=cache.private.edge.network

# Spin up worker VMs

$ edge compute create --plan small \

--count 3 --script ./n8n-worker.sh

✓ Main + 3 workers, scaling horizontally

Common questions

How does this compare to n8n Cloud / Zapier?

n8n Cloud is per-execution; Zapier is per-task. Self-hosted on Edge is flat-fee. The trade-off is one VM to maintain — but the credential security and unlimited execution count usually justify it for any serious automation use.

How does it scale?

Single-VM "main mode" handles thousands of executions per day. For more, switch to queue mode: main process triggers, Redis queue, worker VMs execute. All Edge VMs over private networking.

What about workflow backups?

Workflows live in Postgres. Nightly `pg_dump` to Edge Storage covers them, plus n8n's built-in version-control feature can sync them to a git repo.

Can I use AI nodes?

Yes — n8n has first-class nodes for OpenAI, Anthropic, Ollama, etc. Pair with our Ollama or vLLM stack pages to keep the inference on Edge too.

By Stack

Other stacks on Edge

View all stacks →

Run n8n on your terms

30-day trial. Move off Zapier or Make and unlimited the executions.