N8n cloud instance extremely slow on big workflows

Questions
,
1

:snail: Issue Description

My workflows become extremely slow when they get large and start processing a lot of text data.

I understand that some slowdown is expected with bigger workflows, but I’d like to know if there are any best practices or tips to improve performance. I’ve noticed that the biggest drop in speed seems to happen when n8n loads the output data from each node — especially when dealing with heavy or lengthy text.

:information_source: n8n Setup

  • n8n version: 1.99.1
  • Deployment: n8n Cloud
  • Operating system: Windows 11
  • Browser: Google Chrome
2

Best is to split the processing up into smaller batches. and then process those batches in a subworkflow. after a subworkflow ends it will return the data from the last node executed and release everything from memory. So if you make sure not to return data from the subflow the main flow will remain fast and snappy :slight_smile:

2 Likes
3

Ok, thank you!

It could make sense.

It’s just not the way I would like to structure the workflow :slight_smile:

1 Like
4

Hi @joao_antunes, welcome to the community!

Anyway, here’s some tips for you.

  1. Use sub-workflow, use it if you will use after run AI Agent
  2. Use MCP Client and create MCP Server Triggers in another workflow, use this if you want flexibility.

You can read these documentations to support your flows:

  1. Sub-Workflows
  2. MCP Server Trigger
  3. MCP Client Tool (For AI Agent)
2 Likes