Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
A Newcomer's Guide To Airflow's Architecture
Search
Andrew Godwin
July 12, 2021
Programming
0
350
A Newcomer's Guide To Airflow's Architecture
A talk I gave at Airflow Summit 2021.
Andrew Godwin
July 12, 2021
Tweet
Share
More Decks by Andrew Godwin
See All by Andrew Godwin
Reconciling Everything
andrewgodwin
1
310
Django Through The Years
andrewgodwin
0
200
Writing Maintainable Software At Scale
andrewgodwin
0
440
Async, Python, and the Future
andrewgodwin
2
660
How To Break Django: With Async
andrewgodwin
1
720
Taking Django's ORM Async
andrewgodwin
0
720
The Long Road To Asynchrony
andrewgodwin
0
650
The Scientist & The Engineer
andrewgodwin
1
760
Pioneering Real-Time
andrewgodwin
0
420
Other Decks in Programming
See All in Programming
#kanrk08 / 公開版 PicoRubyとマイコンでの自作トレーニング計測装置を用いたワークアウトの理想と現実
bash0c7
1
650
git worktree × Claude Code × MCP ~生成AI時代の並列開発フロー~
hisuzuya
1
510
ReadMoreTextView
fornewid
1
490
Kotlin エンジニアへ送る:Swift 案件に参加させられる日に備えて~似てるけど色々違う Swift の仕様 / from Kotlin to Swift
lovee
1
260
Cline指示通りに動かない? AI小説エージェントで学ぶ指示書の書き方と自動アップデートの仕組み
kamomeashizawa
1
600
ソフトウェア品質を数字で捉える技術。事業成長を支えるシステム品質の マネジメント
takuya542
0
250
イベントストーミング図からコードへの変換手順 / Procedure for Converting Event Storming Diagrams to Code
nrslib
1
550
Node-RED を(HTTP で)つなげる MCP サーバーを作ってみた
highu
0
110
Cursor AI Agentと伴走する アプリケーションの高速リプレイス
daisuketakeda
1
130
ふつうの技術スタックでアート作品を作ってみる
akira888
0
220
Railsアプリケーションと パフォーマンスチューニング ー 秒間5万リクエストの モバイルオーダーシステムを支える事例 ー Rubyセミナー 大阪
falcon8823
4
1k
0626 Findy Product Manager LT Night_高田スライド_speaker deck用
mana_takada
0
130
Featured
See All Featured
Principles of Awesome APIs and How to Build Them.
keavy
126
17k
YesSQL, Process and Tooling at Scale
rocio
173
14k
Six Lessons from altMBA
skipperchong
28
3.9k
KATA
mclloyd
30
14k
The Invisible Side of Design
smashingmag
300
51k
Fight the Zombie Pattern Library - RWD Summit 2016
marcelosomers
233
17k
Large-scale JavaScript Application Architecture
addyosmani
512
110k
Bash Introduction
62gerente
614
210k
Sharpening the Axe: The Primacy of Toolmaking
bcantrill
44
2.4k
Fashionably flexible responsive web design (full day workshop)
malarkey
407
66k
Intergalactic Javascript Robots from Outer Space
tanoku
271
27k
Distributed Sagas: A Protocol for Coordinating Microservices
caitiem20
331
22k
Transcript
A NEWCOMER'S GUIDE TO ANDREW GODWIN // @andrewgodwin AIRFLOW'S ARCHITECTURE
Hi, I’m Andrew Godwin • Principal Engineer at • Also
a Django core developer, ASGI author • Using Airflow since March 2021
None
High-Level Concepts What exactly is going on? The Good and
the Bad Or, How I Learned To Stop Worrying And Love The Scheduler Problems, Fixes & The Future Where we go from here
Differences from things I have worked on? (An eclectic variety
of web and backend systems)
"Real-time" versus batch The availability versus consistency tradeoff is different!
Simple concepts, hard to master In Django, it's the ORM. In Airflow, scheduling. It's all still distributed systems Which is fortunate, after fifteen years of doing them
Airflow grew organically It started off as an internal ETL
tool
None
DAG ➡ DagRun One per scheduled run, as the run
starts Operator ➡ Task When you call an operator in a DAG Task ➡ TaskInstance When a Task needs to run as part of a DagRun
Scheduler Works out what TaskInstances need to run Executor Runs
TaskInstances and records the results
Scheduler LocalExecutor Webserver Database DAG Files
Scheduler CeleryExecutor Webserver Database DAG Files Redis/Queue Workers
The Executor runs inside the Scheduler Its logic, at least,
and the tasks too for local ones
Everything talks to the database It's the single central point
of coordination
Scheduler, Workers, Webserver All can be run in a high-availability
pattern
Scheduler Works out what TaskInstances need to run Executor Runs
TaskInstances and records the results
Scheduler Works out what TaskInstances need to run Executor Runs
TaskInstances and records the results
Timing Dependencies Retries Concurrency Callbacks ...
Scheduler Works out what TaskInstances need to run Executor Runs
TaskInstances and records the results
Celery or Kubernetes Our two main options, currently
Scheduler CeleryExecutor Webserver Database DAG Files Redis/Queue Workers
Scheduler KubernetesExecutor Webserver Database DAG Files Kubernetes Task Pods
None
Tasks are the core part of the model DAGs are
more of a grouping/trigger mechanism
Very flexible runtime environments Airflow's strength, and its weakness
Airflow doesn't know what you're running This is both an
advantage and a disadvantage.
What can we improve? Let's talk about The Future
More Async & Eventing Anything that involves waiting!
Scheduler CeleryExecutor Webserver Database DAG Files Redis/Queue Workers Triggerer
Removing Database Connections APIs scale a lot better!
I do like the database, though There's a lot of
benefit in proven technology
Software Engineering is not just coding Any large-scale project needs
documentation, architecture, and coordination
Maintenance & compatibility is crucial Anyone can write a tool
- supporting it takes effort
Airflow is forged by people like you. Coding, documentation, triage,
QA, support - it all needs doing.
Thanks. Andrew Godwin @andrewgodwin
[email protected]