Best VPS for n8n in 2026
n8n's server needs are driven by one thing: how much data your workflows hold in memory at once. Get the RAM right and everything else follows. Here is how to size it, and the plans we actually run n8n on.
RAM is the whole story
n8n idles cheaply — around 300-400 MB. The memory spikes come from workflow data: a node that pulls a 50 MB API response, a loop over thousands of items, or several workflows running at once each hold their working set in RAM. That is why two people can have wildly different needs on the "same" tool. Size for your heaviest concurrent workflow, not the idle baseline.
Sizing by workload
| Your usage | RAM | Why |
|---|---|---|
| Personal automations, a few workflows, low volume | 2 GB | n8n + Postgres with headroom. The comfortable entry point. |
| Small business, moderate volume, some big API payloads | 4 GB | Room for concurrent executions and data-heavy nodes. |
| High volume or queue mode with workers | 8 GB+ | Main + Redis + Postgres + multiple workers. |
Note what is not here: CPU cores. n8n is rarely CPU-bound outside heavy data transforms, so two vCPUs is plenty for almost everyone. Do not overpay for cores you will not use — spend it on RAM.
What else matters
- NVMe disk. Postgres and execution history live on disk; NVMe keeps the UI and queries snappy. Avoid old spinning-disk plans.
- Location. If external services call your webhooks, put the server near them to cut round-trip time. EU-based automations belong on EU hardware.
- In-place upgrades. n8n usage grows. A provider that resizes with a reboot beats one that forces a migration.
- Bandwidth. Data-heavy workflows move real traffic; check the allowance isn't tiny.
Our picks
Best overall: Inferno 4 GB
This is what we run production n8n on. NVMe, generous bandwidth, eight EU + Asia locations and DDoS protection included, with in-place resizing when a workflow set grows. The 4 GB plan comfortably handles Postgres plus real concurrent workflows, and you can start smaller and scale up without moving anything.
Recommended: Inferno VPS for n8n
Start on the 2 GB plan for personal automations ($3.49/mo) or the 4 GB for business workflows. NVMe, eight locations, DDoS protection, and one-click upgrades when you outgrow it.
Budget entry: 2 GB anything NVMe
For a personal instance, any reputable 2 GB NVMe plan works. The recommended VPS above starts here; the key is not going below 2 GB, where Postgres plus n8n starts swapping.
Heavy / queue mode: 8 GB
Once you move to queue mode with multiple workers, jump to 8 GB. Main, Redis, Postgres and a couple of workers at concurrency 10 will use it.
What to avoid
- 1 GB plans. They look fine until Postgres and a real workflow run together, then swap. False economy.
- Spinning-disk "storage" VPS. Cheap per GB, painful for database latency.
- Oversold shared-CPU bargains. A stalled core during an execution shows up as a hung workflow.
What to do next
- Install n8n on your new VPS — the full walkthrough.
- Cloud vs self-hosted — confirm self-hosting is right for you.
- Best VPS for Docker — if n8n is one of several containers.
Author: Marek Vainionpää — SRE, Helsinki. Last tested: 2026-05-14 on n8n 1.62 across Inferno 2/4/8 GB plans.