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
Scaling Django with Distributed Systems
Search
Andrew Godwin
April 07, 2017
Programming
3
2.2k
Scaling Django with Distributed Systems
A talk I gave at PyCon Ukraine 2017.
Andrew Godwin
April 07, 2017
Tweet
Share
More Decks by Andrew Godwin
See All by Andrew Godwin
Reconciling Everything
andrewgodwin
1
330
Django Through The Years
andrewgodwin
0
220
Writing Maintainable Software At Scale
andrewgodwin
0
460
A Newcomer's Guide To Airflow's Architecture
andrewgodwin
0
370
Async, Python, and the Future
andrewgodwin
2
680
How To Break Django: With Async
andrewgodwin
1
740
Taking Django's ORM Async
andrewgodwin
0
740
The Long Road To Asynchrony
andrewgodwin
0
680
The Scientist & The Engineer
andrewgodwin
1
790
Other Decks in Programming
See All in Programming
ソフトウェア設計の実践的な考え方
masuda220
PRO
4
640
When Dependencies Fail: Building Antifragile Applications in a Fragile World
selcukusta
0
110
SODA - FACT BOOK(JP)
sodainc
1
8.7k
テーブル定義書の構造化抽出して、生成AIでDWH分析を試してみた / devio2025tokyo
kasacchiful
0
280
GC25 Recap: The Code You Reviewed is Not the Code You Built / #newt_gophercon_tour
mazrean
0
110
CSC509 Lecture 08
javiergs
PRO
0
240
AI Agent 時代的開發者生存指南
eddie
4
2.1k
3年ぶりにコードを書いた元CTOが Claude Codeと30分でMVPを作った話
maikokojima
0
630
Range on Rails ―「多重範囲型」という新たな選択肢が、複雑ロジックを劇的にシンプルにしたワケ
rizap_tech
0
6.7k
Domain-centric? Why Hexagonal, Onion, and Clean Architecture Are Answers to the Wrong Question
olivergierke
3
960
登壇は dynamic! な営みである / speech is dynamic
da1chi
0
360
Vueのバリデーション、結局どれを選べばいい? ― 自作バリデーションの限界と、脱却までの道のり ― / Which Vue Validation Library Should We Really Use? The Limits of Self-Made Validation and How I Finally Moved On
neginasu
2
1.4k
Featured
See All Featured
Learning to Love Humans: Emotional Interface Design
aarron
274
41k
Bootstrapping a Software Product
garrettdimon
PRO
307
110k
KATA
mclloyd
PRO
32
15k
A better future with KSS
kneath
239
18k
Rebuilding a faster, lazier Slack
samanthasiow
84
9.2k
Mobile First: as difficult as doing things right
swwweet
225
10k
Measuring & Analyzing Core Web Vitals
bluesmoon
9
630
Connecting the Dots Between Site Speed, User Experience & Your Business [WebExpo 2025]
tammyeverts
10
610
Designing for Performance
lara
610
69k
Build The Right Thing And Hit Your Dates
maggiecrowley
38
2.9k
Improving Core Web Vitals using Speculation Rules API
sergeychernyshev
21
1.2k
Practical Orchestrator
shlominoach
190
11k
Transcript
None
Andrew Godwin Hi, I'm Django core developer Senior Software Engineer
at Used to complain about migrations a lot
Distributed Systems
c = 299,792,458 m/s
Early CPUs c = 60m propagation distance Clock ~2cm 5
MHz
Modern CPUs c = 10cm propagation distance 3 GHz
Distributed systems are made of independent components
They are slower and harder to write than synchronous systems
But they can be scaled up much, much further
Trade-offs
There is never a perfect solution.
Fast Good Cheap
None
Load Balancer WSGI Worker WSGI Worker WSGI Worker
Load Balancer WSGI Worker WSGI Worker WSGI Worker Cache
Load Balancer WSGI Worker WSGI Worker WSGI Worker Cache Cache
Cache
Load Balancer WSGI Worker WSGI Worker WSGI Worker Database
CAP Theorem
Partition Tolerant Consistent Available
PostgreSQL: CP Consistent everywhere Handles network latency/drops Can't write if
main server is down
Cassandra: AP Can read/write to any node Handles network latency/drops
Data can be inconsistent
It's hard to design a product that might be inconsistent
But if you take the tradeoff, scaling is easy
Otherwise, you must find other solutions
Read Replicas (often called master/slave) Load Balancer WSGI Worker WSGI
Worker WSGI Worker Replica Replica Main
Replicas scale reads forever... But writes must go to one
place
If a request writes to a table it must be
pinned there, so later reads do not get old data
When your write load is too high, you must then
shard
Vertical Sharding Users Tickets Events Payments
Horizontal Sharding Users 0 - 2 Users 3 - 5
Users 6 - 8 Users 9 - A
Both Users 0 - 2 Users 3 - 5 Users
6 - 8 Users 9 - A Events 0 - 2 Events 3 - 5 Events 6 - 8 Events 9 - A Tickets 0 - 2 Tickets 3 - 5 Tickets 6 - 8 Tickets 9 - A
Both plus caching Users 0 - 2 Users 3 -
5 Users 6 - 8 Users 9 - A Events 0 - 2 Events 3 - 5 Events 6 - 8 Events 9 - A Tickets 0 - 2 Tickets 3 - 5 Tickets 6 - 8 Tickets 9 - A User Cache Event Cache Ticket Cache
Teams have to scale too; nobody should have to understand
eveything in a big system.
Services allow complexity to be reduced - for a tradeoff
of speed
Users 0 - 2 Users 3 - 5 Users 6
- 8 Users 9 - A Events 0 - 2 Events 3 - 5 Events 6 - 8 Events 9 - A Tickets 0 - 2 Tickets 3 - 5 Tickets 6 - 8 Tickets 9 - A User Cache Event Cache Ticket Cache User Service Event Service Ticket Service
User Service Event Service Ticket Service WSGI Server
Each service is its own, smaller project, managed and scaled
separately.
But how do you communicate between them?
Service 2 Service 3 Service 1 Direct Communication
Service 2 Service 3 Service 1 Service 4 Service 5
Service 2 Service 3 Service 1 Service 4 Service 5
Service 6 Service 7 Service 8
Service 2 Service 3 Service 1 Message Bus Service 2
Service 3 Service 1
A single point of failure is not always bad -
if the alternative is multiple, fragile ones
Channels and ASGI provide a standard message bus built with
certain tradeoffs
Backing Store e.g. Redis, RabbitMQ ASGI (Channel Layer) Channels Library
Django Django Channels Project
Backing Store e.g. Redis, RabbitMQ ASGI (Channel Layer) Pure Python
Failure Mode At most once Messages either do not arrive,
or arrive once. At least once Messages arrive once, or arrive multiple times
Guarantees vs. Latency Low latency Messages arrive very quickly but
go missing more Low loss rate Messages are almost never lost but arrive slower
Queuing Type First In First Out Consistent performance for all
users First In Last Out Hides backlogs but makes them worse
Queue Sizing Finite Queues Sending can fail Infinite queues Makes
problems even worse
You must understand what you are making (This is surprisingly
uncommon)
Design as much as possible around shared-nothing
Per-machine caches On-demand thumbnailing Signed cookie sessions
Has to be shared? Try to split it
Has to be shared? Try sharding it.
Django's job is to be slowly replaced by your code
Just make sure you match the API contract of what
you're replacing!
Don't try to scale too early; you'll pick the wrong
tradeoffs.
Thanks. Andrew Godwin @andrewgodwin channels.readthedocs.io