首页 Data Engineering

课程

Building Data Pipelines with Airflow

高级技能水平

更新时间 2026年6月

Author Dags with the TaskFlow API, asset-based scheduling, and deferrable sensors, and run an end-to-end SQL ETL pipeline with quality checks.

免费开始课程

AirflowData Engineering

4小时

16 视频

60 道练习

4,500 XP

成就证明

深受数千家公司学习者的喜爱

需要团队培训？

企业版试用

课程描述

Want to take your Airflow skills further? This course uses Airflow 3.2, so you'll learn the latest way of doing things. You'll author Dags with the TaskFlow API, schedule them based on data using Assets and the new Asset Partitions, and make them reliable with retries, callbacks, and tests.In the final chapter, you'll build an end-to-end SQL ETL pipeline on DuckDB and add data quality checks directly in Airflow, without any third-party libraries, so the data your pipeline produces stays trustworthy. By the end, you'll know how to take a pipeline from a prototype to something you can actually run in production.

先决条件

Introduction to Apache Airflow in Python

1

Authoring Dags with TaskFlow and XCom

You'll start by meeting the Airflow components, writing your first Dags with the TaskFlow API, and passing data between tasks with XCom.

Airflow core concepts

Writing a minimal TaskFlow Dag

Matching Airflow components to their roles

The TaskFlow API

Converting a classic Dag to TaskFlow

Adding a task and wiring dependencies

Exploring a pipeline in the Airflow UI

Sharing data between tasks

Returning data from a TaskFlow function

Connecting a classic operator to a TaskFlow task

Understanding XCom limitations

2

Dynamic and Data-Aware Pipelines

From there, you'll run tasks in parallel with dynamic task mapping, schedule Dags by data with Assets, and add human approval steps.

Dynamic Task Mapping

Processing multiple endpoints with expand

Fixing a shared parameter with partial

Pairing inputs with zip

Data-aware scheduling with Assets

Signaling data availability with an Asset outlet

Scheduling a consumer Dag on an Asset

Seeing asset-based scheduling in action

Templating, idempotency, and backfilling

Adding date-aware templating

Making a pipeline idempotent

Triggering a backfill

Human-in-the-loop workflows

Collecting form input with HITLEntryOperator

Responding to a Required Action in the UI

Branching with HITLBranchOperator

3

Preparing Dags for Production

In this chapter, you'll handle failures with retries and callbacks, save resources with deferrable sensors, and test your Dags at three levels.

Robust Dags with failure handling and retries

Configuring retries with exponential backoff

Auto-pausing a failing Dag

Choosing the right failure handling strategy

Deferrable operators

Configuring a sensor for production

Making a sensor deferrable

When to use poke, reschedule, or deferrable

Testing Airflow code

Writing a Dag integrity test

Unit testing a task function

Integration testing with dag.test()

Organizing complex Dags with Task Groups

Grouping related tasks

Nesting task groups with display names

Organizing tasks by domain

4

Building a Production SQL ETL Pipeline

In this final chapter, you'll build a SQL ETL pipeline on DuckDB, add partition-aware scheduling with Asset Partitions, and embed data quality checks.

Running SQL workloads

Aggregating sales data with SQL

Parameterizing SQL and extracting it into files

Choosing the safe approach for user input

Partition-aware scheduling with Asset Partitions

Creating a partitioned pipeline

Scheduling on partitioned assets

Watching partitioned assets trigger a downstream Dag

Data quality checks

Adding column-level quality checks

Adding table-level quality checks

Catching data issues with the right operator

From development to production

Triggering a pipeline from the CLI

Backfilling historical data

Production challenges with Airflow

Congratulations!

Building Data Pipelines with Airflow

课程完成

获得成就证明

将此证书添加到您的 LinkedIn 档案、简历或履历中
在社交媒体和绩效评估中分享立即注册

加入超过19百万学习者，今天就开始Building Data Pipelines with Airflow！

通过 DataCamp for Mobile 提升您的数据技能

随时随地通过我们的移动课程和每日 5 分钟编程挑战提升技能。