N8N benchmarking throughput - max 30 req/s?

Questions
,
1

I’m doing some performance testing on N8N to find out what the limits are.

I’m running N8N in queue mode on a Kubernetes (GKE) with 2 workers and 2 webhook instances. Response times seem to be doing extremely well up until 30 requests / s (50ms response time). From there response time starts aggregating upwards. I tried increasing the amount of workers (and concurrency)/webhooks, but this doesn’t seem to decrease the speed of response time aggregation. I’m using a GKE provided ingress. The trigger is a simple GET Webhook.

The reason why I’m reaching out is to find out what is causing this increase in latency, and if there’s a way to overcome it. It’s purely from theoretical interests to understand up until where n8n can be considered to be a viable solution for integration.

I’ve benchmarked against edge functions to create a performance baseline for the api performance testing tool I’m using (apiDog), and I can confirm there’s no latency on their side.

See benchmark underneath.

image
image1476×876 64.3 KB

Happy to hear any thoughts on the matter.

2

It looks like your topic is missing some important information. Could you provide the following if applicable.

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
3

Following up with

  • **n8n version:1.24.1
  • **Database (default: SQLite):Postgres
  • **n8n EXECUTIONS_PROCESS setting (default: own, main):-
  • **Running n8n via (Docker, npm, n8n cloud, desktop app):Kubernetes
  • **Operating system:-
1 Like
5

Hi @Cees_de_Kort, are you storing execution data for your benchmark workflow? If so, the latency might come down to the increased database load. In a first step you could disable storing execution data for the workflow in question and verify if this has an impact.

6

Right observation, database was throthling on CPU usage. Hitting 90 Req/s now, everything is scaling well.

2 Likes
7

Sweet, thanks for confirming!

8

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.