Buyer guide · Updated 2026-10-27

n8n VPS Requirements: How to Size CPU and RAM from Your Workload

There is no universal n8n hardware minimum. Use payload size, concurrency, observed memory, and current VPS plan terms to choose a monitored starting point.

✓

Quick answer
There is no official one-size-fits-all CPU or RAM minimum for self-hosted n8n: workload size, binary data, Code nodes, and concurrent executions all change memory use. A practical VPS choice is therefore a measured starting point, not a guarantee. On DigitalOcean Basic Droplets, current USD monthly caps include 1 vCPU/2 GiB RAM at $12, 2 vCPUs/2 GiB at $18, 2 vCPUs/4 GiB at $24, and 4 vCPUs/8 GiB at $48. These are ongoing listed rates, not intro offers; no separate renewal price is published. Source check: October 27, 2026.

Original sizing framework. It is a decision process, not a performance benchmark or hardware guarantee.

Why n8n sizing depends on your workflows

n8n does not publish one CPU/RAM minimum that guarantees a given number of workflows or users. Its memory guidance says that JSON and binary payload size, node count, memory-heavy Code or Function nodes, manual runs, and simultaneous executions can increase memory use. When memory is insufficient, the documented choices are to provision more memory or reduce the work held at once, for example by processing smaller batches.

That makes “n8n needs X GB” a poor buying rule. Start by listing your own workflow patterns: small scheduled API calls differ from workflows that download files, transform large datasets, or run many jobs together. Include the operating system, Docker, database, reverse proxy, and monitoring in your capacity plan too; the VPS total is not all available to n8n.

Current VPS choices and what they cost

The table uses DigitalOcean Basic Droplets, a shared-CPU virtual machine product, as a transparent example rather than a claim that it is the only suitable provider. Prices are the provider's currently listed USD monthly caps, not a limited-time discount. DigitalOcean says Droplets are billed per second with a 60-second minimum and monthly cap. The public price table does not list a separate renewal amount for these ongoing Basic rates; confirm the plan and final invoice in the official checkout before ordering. Source checked October 27, 2026.

Basic plan example	CPU	RAM	Disk / transfer	Current USD monthly cap	Weekly backup add-on*
Small candidate	1 shared vCPU	2 GiB	50 GiB / 2,000 GiB	$12	$2.40; $14.40 total
More CPU, same RAM	2 shared vCPUs	2 GiB	60 GiB / 3,000 GiB	$18	$3.60; $21.60 total
More memory	2 shared vCPUs	4 GiB	80 GiB / 4,000 GiB	$24	$4.80; $28.80 total
Larger candidate	4 shared vCPUs	8 GiB	160 GiB / 5,000 GiB	$48	$9.60; $57.60 total

*Backup arithmetic: DigitalOcean's basic weekly backup plan adds 20% of Droplet cost per month: for example, $24 × 0.20 = $4.80 and $24 + $4.80 = $28.80, before tax or other services. Daily basic backups add 30%. Weekly backups are retained for four weeks and daily backups for seven days. These are provider disk images; they do not replace an independent copy of n8n's data or a restore test.

Pick CPU and RAM by the bottleneck you can observe

Your measured situation	What to compare	Purchase decision
Mostly small payloads, low concurrency, no memory warnings in representative runs	Available RAM under normal and peak runs; persistent database and disk growth	A lower-cost candidate can be a test starting point if you can monitor it and resize. This is not a promise that 2 GiB is sufficient.
Memory warnings, large JSON/binary data, Code nodes, or overlapping jobs	Whether the host has enough unused RAM during the actual peak; whether batching or reduced concurrency can shrink demand	Compare a higher-RAM plan or redesign the workflow. Do not buy extra CPU alone to fix a memory shortage.
Executions are CPU-bound while memory has headroom	Host CPU load during real runs; shared-vCPU plan description and any provider limits	Compare CPU options and benchmark your own workload. DigitalOcean describes Basic CPU as shared and variable; it does not guarantee a specific n8n throughput.
Heavy files, many concurrent workflows, or strict uptime/recovery expectations	File handling, concurrency, storage, backup recovery, and acceptable downtime	Evaluate a larger or different architecture only after testing. A single VPS plus optional disk images is not a high-availability or managed recovery service.

On n8n's official guidance, an out-of-memory event can be indicated by messages such as “Execution stopped at this node” or “JavaScript heap out of memory.” n8n identifies payload size, binary data, number/type of nodes, manual executions, and simultaneous work as relevant factors. Test with a representative peak workflow before committing important production jobs.

Region, backups, renewal, and refund terms

Data-center location: Basic Droplets are listed as available in NYC1/NYC2/NYC3, Amsterdam (AMS3), SFO2/SFO3, Singapore (SGP1), London (LON1), Frankfurt (FRA1), Toronto (TOR1), Bangalore (BLR1), and Sydney (SYD1), with availability limits for some CPU variants in some locations. Pick only a region that the creation screen currently offers for your selected size.
Renewal price: these are listed ongoing monthly-cap rates, not a discounted introductory term. DigitalOcean's public pricing page does not publish a separate renewal rate for this Basic plan table. Check the live rate before creating the Droplet; do not infer a future price from a coupon.
Backups: optional system-level disk images; basic weekly is +20% and daily +30% of Droplet monthly cost. DigitalOcean's docs say weekly retention is four weeks and daily retention is seven days. Verify that the schedule and retention meet your recovery needs.
Refunds: DigitalOcean's published policy says it does not offer refunds; it says customers with extenuating circumstances may contact support. If a refund window is required for your purchase decision, this policy does not meet that requirement.

Who should choose self-hosted n8n on a VPS?

Potentially suitable: an operator who can administer Ubuntu and Docker, has a real reason to control the host, can secure public endpoints, and will monitor workload memory, maintain updates, and keep independent backups. These are operational requirements of running the server; the n8n documentation does not certify a particular Droplet tier for them.

Not a good fit: someone who needs a managed service to handle server patching and infrastructure, cannot tolerate hands-on recovery, needs a guaranteed hardware minimum for their unmeasured workload, or requires a provider refund guarantee. For those cases, compare n8n Cloud or a managed hosting option directly and check its published plan limits; do not assume its server CPU, RAM, or region when they are not disclosed.

Before you pay: a short sizing worksheet

Write down the largest realistic payload and whether workflows handle files or use Code nodes.
Record the highest number of workflows that overlap, not just the total number of workflows saved.
Run a representative test and record memory, CPU, execution errors, and disk growth during the peak.
Choose a current plan as a monitored starting candidate; include the operating system and services in the available-resource estimate.
Price the chosen region, monthly cap, backup add-on, taxes, and any extra storage in checkout. Confirm the provider's current refund and renewal terms before paying.
Repeat the test after changing payload size or concurrency. Resize or redesign when observed evidence shows the current choice is insufficient.

Official sources

n8n: Fix memory issues — factors that affect memory use and ways to respond.
n8n: Scaling — scaling modes and execution-data controls.
DigitalOcean: Droplet Pricing — current Basic plan resource sizes, USD rates, and billing basis.
DigitalOcean: Droplet Availability — region and plan availability.
DigitalOcean: Backups Pricing and Backup Quickstart — add-on rates and retention.
DigitalOcean: Can I have a refund? — published refund policy.

Source check: October 27, 2026. DigitalOcean figures describe its current Basic Droplet product, not an n8n sizing recommendation. Suitability guidance above is conditional on measured workload and the operator's ability to administer the host. Provider terms and availability can change; verify the selected plan in checkout.