When do I need these?

Around the 20th workflow, or when incidents spike.

Real but small — the payoff is at scale.

Move to a full code service. But most teams don't need to.

Extract your most-copied logic into a sub-workflow this week.

Patterns10 min readUpdated 2026-06-29

Advanced Apache Airflow Patterns in n8n

Sub-workflows, dynamic routing, and event-sourced Apache Airflow pipelines in n8n — for the ambitious.

Once your Apache Airflow workflows outgrow single-flow linear logic, these patterns keep them maintainable at 10x scale.

Extract every reusable Apache Airflow action into a sub-workflow. Version them like functions. Call from anywhere.

Use a routing table (Postgres or JSON) to decide which sub-workflow handles each Apache Airflow event. Add types without redeploying.

Persist every raw Apache Airflow event to a log. Rebuild any downstream projection by replaying the log. Debugging becomes trivial.

Queue mode plus per-key concurrency locks. Two workflows never step on the same Apache Airflow record.

When do I need these?: Around the 20th workflow, or when incidents spike.
Overhead?: Real but small — the payoff is at scale.
Alternative?: Move to a full code service. But most teams don't need to.
Where to start?: Extract your most-copied logic into a sub-workflow this week.