Anyone Successfully Achieved True Parallel Processing of Subworkflows?

[Freddy_T]

Hi everyone,

I’m working on an n8n workflow that requires parallel execution of two basic LLM Chains and I’ve run into some roadblocks trying to make this work.

I understand that native parallel processing isn’t officially supported but I’ve been testing a couple of workarounds to simulate it, which I found here in forum posts. Specifically, I’ve tried:

Running two subworkflows using the Execute Workflow node.
Triggering the subflows via HTTP Webhook calls from the main workflow.

Unfortunately, both approaches still run the subflows sequentially, not in parallel as hoped. Even when I switched the execution mode from v1 to Legacy (v0) there was no change—the subflows are processed one after the other from the main flow.

Has anyone here managed to get true parallel execution working reliably in n8n? Would love to hear about any tricks, node configurations, or architectural patterns that made it work.

Thanks in advance for your insights!

Freddy

[hubschrauber]

If you call the “sub workflow” synchronously, you’ll still have the serialized behavior. The only way I’ve found to do this is to call via http asynchronously and use a Wait w/ Resume on Webhook Callback. So, same answer as this recent post which has a link to a template that shows a flow that:

1. Asynchronously starts all the sub-processes
2. Loops the same number of times to wait for the callbacks from each started process.

• Note this template depends on n8n being set up such that it will run more than one workflow at a time. With nodejs, this is managed using a single thread with an event loop. If you wanted even more parallelism, you could run n8n in queue mode (i.e. with multiple workers).