Can large flows cause slowdowns?

I have a flow that I consider quite large

I notice that in this flow the nodes take longer than in other flows that are smaller, the flows that take the longest are postgres, bubble and the last node, it can be http request connected with an immediate response webhook, or rabbitMQ, this last node always they take an average of 7000ms, these postgres nodes sometimes take 50000ms when the same node when used in a smaller flow does not exceed 50ms.

Captura de Tela 2024-03-07 às 15.15.28
Captura de Tela 2024-03-07 às 15.13.40

I’m sure I have the amount of memory, cpu and replicas, more than necessary for my operation, my doubt is:

The N8N was not designed to be used with large flows and is it better to divide it into several small ones or is this a possible BUG?

Information on your n8n setup

  • **n8n version:1.29.1 queue
  • **Database (default: SQLite):Postgres
  • **n8n EXECUTIONS_PROCESS setting (default: own, main):
  • **Running n8n via (Docker, npm, n8n cloud, desktop app):docker swarm
  • **Operating system:ubunto 20

Is everything else really identical? So same number of items (best 1), similar item data size, calling same server, …?

yes, same table and similar amount of data, calling on the same server.

I did a test now these two nodes did exactly the same thing, started in the same second, the only difference is that one is in a large flow and the other in a small flow.

@jan Is this a bug or should I split it into smaller streams?

Sounds like a bug. Is the node execution in the beginning or the end of the workflow?

In this example that I sent here, it is at the end of the flow, but there are other nodes that also take a long time.

This flow runs on average about 10 times per minute and its 9 workers reach a minimum of 80% with 9 CPU and 10 GB of RAM dedicated to this service. conflict, since in the last updates a lot of things related to Postgres were changed.