The basic way of defining tasks in Airflow is through directed acyclic graphs (DAGs), i.e., defining relations between operators (sequence or parallelism of tasks), order and dependencies. This means that when creating a workflow, you need to think about how to break it down into tasks that can be performed independently of each other. You can then combine these tasks into a logical whole by arranging them in a chart.

The shape of the graph determines the overall logic of the workflow. Airflow DAG can contain multiple branches, and you can decide which branches to follow and which to skip during workflow execution.

This creates a stable structure because each task can be repeated multiple times if an error occurs. Airflow can even be stopped completely, and running workflows will resume when the last uncompleted task is restarted.