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
Asynchronous Processing for Fun and Profit
Search
Mike Perham
November 02, 2012
Technology
11
870
Asynchronous Processing for Fun and Profit
Some pro tips and an overview of Sidekiq for great async victory!
Mike Perham
November 02, 2012
Tweet
Share
More Decks by Mike Perham
See All by Mike Perham
Computer Science vs Software Engineering
mperham
0
330
A Tribute to Ezra Zygmuntowicz
mperham
1
1k
Other Decks in Technology
See All in Technology
Culture Deck
optfit
0
420
2024.02.19 W&B AIエージェントLT会 / AIエージェントが業務を代行するための計画と実行 / Algomatic 宮脇
smiyawaki0820
13
3.4k
Larkご案内資料
customercloud
PRO
0
650
開発組織のための セキュアコーディング研修の始め方
flatt_security
3
2.4k
インフラをつくるとはどういうことなのか、 あるいはPlatform Engineeringについて
nwiizo
5
2.6k
PHPカンファレンス名古屋-テックリードの経験から学んだ設計の教訓
hayatokudou
2
290
エンジニアのためのドキュメント力基礎講座〜構造化思考から始めよう〜(2025/02/15jbug広島#15発表資料)
yasuoyasuo
17
6.7k
JEDAI Meetup! Databricks AI/BI概要
databricksjapan
0
100
Data-centric AI入門第6章:Data-centric AIの実践例
x_ttyszk
1
400
関東Kaggler会LT: 人狼コンペとLLM量子化について
nejumi
3
580
人はなぜISUCONに夢中になるのか
kakehashi
PRO
6
1.7k
Cloud Spanner 導入で実現した快適な開発と運用について
colopl
1
650
Featured
See All Featured
Keith and Marios Guide to Fast Websites
keithpitt
411
22k
The Art of Delivering Value - GDevCon NA Keynote
reverentgeek
10
1.3k
Helping Users Find Their Own Way: Creating Modern Search Experiences
danielanewman
29
2.4k
実際に使うSQLの書き方 徹底解説 / pgcon21j-tutorial
soudai
174
51k
Designing for humans not robots
tammielis
250
25k
Unsuck your backbone
ammeep
669
57k
Put a Button on it: Removing Barriers to Going Fast.
kastner
60
3.7k
No one is an island. Learnings from fostering a developers community.
thoeni
21
3.1k
Measuring & Analyzing Core Web Vitals
bluesmoon
6
240
Navigating Team Friction
lara
183
15k
Java REST API Framework Comparison - PWX 2021
mraible
28
8.4k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
rmw
33
2.8k
Transcript
Async Processing for Fun and Profit Mike Perham @mperham
Me Director of Engineering, TheClymb.com
Agenda • Basics • Protips • Sidekiq
Why? • User-Perceived Performance • I/O is SLOOOOOOOOOOOW • I/O
is unreliable
What? • Optional work • Anything not required to build
the HTTP response
How? • Client puts message on a queue • Server
pulls message off queue • Worker executes code based on message
How? async do # perform some work end
How? • Marshalling a Proc • Need to serialize closure
How? async(instance, :method, args)
How? • Marshal instance • Marshal args
How? async(Class, :method, args)
How? • Serialize just the class name • Marshal the
arguments
Congratulations! This is exactly how Resque and Sidekiq work.
How?
How?
How?
Tip #1 Small, Stateless Messages
Stateless • Database holds objects (nouns) • Queue holds actions
(verbs) • “Perform X on Object 123”
Avoid State • Bad • @user.delay.sync_images • Good • User.delay.sync_images(@user.id)
Simple Types • Small & Easy to read • Cross-platform
• Sidekiq / Resque use JSON
Debugging
Tip #2 Idempotent, Transactional Units of Work
Idempotent • Fancy computer science term • “Work can be
applied multiple times without changing the result beyond the initial application”
Idempotent • Canceling an order • Updating user’s email address
Not Idempotent! • Charging credit card • Sending email
Idempotent • Your code has bugs! • Sidekiq will retry
jobs that raise • Design jobs to be retried • e.g. verify you need to perform action before performing it
Transactional • Infinite credit?
Tip #3 Embrace Concurrency
Concurrency • Resque/DJ = 4 workers • Sidekiq = 100
workers • Mike, what is best in life?
“To crush their servers, see them smoking before you and
hear the lamentations of their admins.”
Concurrency • Use connection_pool gem to limit client connections •
Split work into small batches • 100 items => 10 jobs of 10 items
Concurrency • Thread safety rarely an issue • Most gem
maintainers very responsive • Recently fixed: • cocaine, typheuos
Theory, meet Practice
Sidekiq • Simple, efficient message processing • Like Resque, but
10x faster
MODERN RUBY IS NOT SLOW SINGLE THREADING IS SLOW
Concurrency • To scale single-threaded, create lots of processes. •
HORRIBLY RAM INEFFICIENT
Threads vs Processes • Example: 400MB single-threaded process • 25
processes = 10GB RAM = EC2 xlarge • 25 threads = 1GB RAM = EC2 small • $60/month vs $480/month • 160 dynos => 10 dynos • 150 dynos * $35/month =~ $5000/month
Quick Rant • GIL + poor GC • C extension
API must DIE • What’s larger: 50% or 800%?
Client Your App Client Middleware Redis Sidekiq Client API Rails
Process
Server Processor Server Middleware Redis Worker Sidekiq Process Processor Server
Middleware Worker Processor Server Middleware Worker Processor Server Middleware Worker Fetcher Manager
Versions • Sidekiq - Free, LGPLv3 • Sidekiq Pro -
more features, support, $ • motivation!
Features Sidekiq Resque DJ Concurrency Store Hooks Web UI Scheduler
Retry Delay Batches Threads Processes Processes Redis Redis DB middleware callbacks callbacks ✓ ✓ ? ✓ ? ✓ ✓ ? ✓ ✓ ? ✓ Pro ? - ? - optional
Future • Nicer, more functional Web UI • APIs for
managing queues / retries • Rails 4 Queue API
Pro Future • Enterprise-y features • Workflow • Notifications
Conclusion • small, stateless messages • idempotent / transactional •
concurrency • sidekiq
Questions? @mperham
[email protected]