# Apache Airflow alternative: node-flow vs Airflow

> node-flow vs Apache Airflow — event-driven, long-running business workflows with polyglot workers, compared with Airflow's scheduled Python data pipelines.

Source: https://node-flow.dev/docs/alternatives/airflow

**node-flow is an Airflow alternative for business and service workflows —
not a replacement for data pipelines.** Apache Airflow schedules batch DAGs
written in Python: run this pipeline every night, backfill last month. node-flow
runs *per-event* workflows: an order was placed, a document arrived, a user
signed up — thousands of concurrent runs, each with its own inputs, retries,
waits and human steps.

## node-flow vs Airflow at a glance

|                                | node-flow                                                           | Apache Airflow                                    |
| ------------------------------ | ------------------------------------------------------------------- | ------------------------------------------------- |
| Typical workload               | Per-event business processes, microservice orchestration, AI agents | Scheduled batch and data pipelines                |
| Workflow is…                   | A JSON DAG                                                          | A Python DAG                                      |
| Triggered by                   | API calls, cron, webhooks, message-broker events                    | Schedules, datasets/assets, API                   |
| Run inputs                     | Every run has its own typed input                                   | DAG run configuration                             |
| Latency to start a step        | Workers long-poll; steps start as soon as they are ready            | Scheduler loop and executor                       |
| Languages                      | Any, over HTTP                                                      | Python operators (others via containers)          |
| Long waits and human approvals | First-class                                                         | Sensors and deferrable operators                  |
| Infrastructure                 | PostgreSQL 18                                                       | Scheduler, metadata database, executor, webserver |

## Choose which

**Choose Airflow** for ETL and ELT, backfills, and data-team-owned scheduled
pipelines — its operator ecosystem is unmatched there.

**Choose node-flow** when each run is a transaction in your product: a payment,
an onboarding, a claim, a document pipeline triggered per upload, an LLM agent
loop.

## Frequently asked questions

### Is node-flow a replacement for Airflow?

For event-driven business workflows and microservice orchestration, yes. For scheduled batch data pipelines and backfills, Airflow remains the better tool; the two are often used together.

## Next

* [Concepts](/docs/guide/concepts)
* [System tasks](/docs/guide/system-tasks)