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
Resilient Software Design
Search
Swanand Pagnis
January 21, 2015
Technology
1
260
Resilient Software Design
Lessons learned from Michal Nygard's "Release It!"
Swanand Pagnis
January 21, 2015
Tweet
Share
More Decks by Swanand Pagnis
See All by Swanand Pagnis
Principles of Project Management Flow Part 1 and 2
swanandp
0
110
The Road to Continuous Deployments
swanandp
0
440
Fearless Joins
swanandp
0
270
The Deep Work Divide: Restoring the balance between collaboration and deep work
swanandp
1
800
The Well Rounded Engineer
swanandp
23
12k
The Deep Work Mentality
swanandp
6
1.8k
Combinator Parsing
swanandp
0
600
It's 2017, and I still want to sell you a graph database
swanandp
2
700
Architecture Over Framework: Rethink Your App Structure
swanandp
1
200
Other Decks in Technology
See All in Technology
2024/4/26 コンピュータ歴史博物館解説告知
toshi_atsumi
0
220
長期運用プロジェクトでのMySQLからTiDB移行の検証
colopl
2
820
コンパウンドスタートアップのためのスケーラブルでセキュアなInfrastructure as Codeパイプラインを考える / Scalable and Secure Infrastructure as Code Pipeline for a Compound Startup
yuyatakeyama
4
4.7k
Postman v10リリース後を振り返る
nagix
0
170
日本におけるデータエンジニアリングのこれまでとこれから
foursue
16
4.1k
エンジニアのキャリアをちょっと楽しくする3本の軸/Three Pillars to Make an Engineer's Career More Enjoyable
kwappa
0
2.6k
Google Cloud の AI を支える裏側のインフラを垣間見る!
maroon1st
0
330
開発生産性向上サービスを作るFindyが自分たちで開発生産性を爆上げした組織づくりの歩み / Findy's path to boosting its own development productivity 2024-04-17
ma3tk
3
600
プロデザ! BY リクルート vol.18_リクルートのリサーチ実践組織「リサーチブーストコミュニティ」
recruitengineers
PRO
3
270
MLOpsの「壁」を乗り越える、LINEヤフーの Data Quality as Code
lycorptech_jp
PRO
4
300
カオナビの利用実績をアウトカムへつなげる旅 / example-of-data-management-startup-in-kaonavi
kaonavi
0
130
プラットフォームってつくることより計測することが重要なんじゃないかという話 / Platform Engineering Meetup #8
taishin
1
340
Featured
See All Featured
StorybookのUI Testing Handbookを読んだ
zakiyama
13
4.6k
4 Signs Your Business is Dying
shpigford
175
21k
Dealing with People You Can't Stand - Big Design 2015
cassininazir
357
22k
[RailsConf 2023 Opening Keynote] The Magic of Rails
eileencodes
9
8.3k
Intergalactic Javascript Robots from Outer Space
tanoku
266
26k
Optimizing for Happiness
mojombo
370
69k
CSS Pre-Processors: Stylus, Less & Sass
bermonpainter
352
28k
Code Review Best Practice
trishagee
55
15k
Responsive Adventures: Dirty Tricks From The Dark Corners of Front-End
smashingmag
244
20k
Facilitating Awesome Meetings
lara
42
5.6k
実際に使うSQLの書き方 徹底解説 / pgcon21j-tutorial
soudai
120
39k
Visualization
eitanlees
136
14k
Transcript
Resilient Software Design Building software that doesn’t give up!
1. What’s all this about? 2. Anti-patterns: What not to
do! 3. Patterns: Make your life easier.
Find the odd one 1. :cloud_factory 2. "CloudFactory" 3. CloudFactory
4. CLOUD_FACTORY
What will this code print? if fork puts "I won
the lottery!" else puts "I am bankrupt!" end
What’s all this about?
Stability
It Just Works™
It Should Work™
Bad things happen
Good things happen as well
But those are rare!
Anything that can go wrong, will go wrong. - Murphy’s
Law
Developers think positive
Too much so!
Be negative.
Our goal is to build software
Our goal is also to minimise pain
Our goal is also to save money
Resilient software saves money by not breaking when needed
Resilient software saves money by using optimum infrastructure
Resilient software saves money by keeping developers happy
Anti-patterns
1. Integration points
Integration is not what you think™
Database is an integration
Third party services are integration.
Your cache layer is an integration
None
Networks fail more often
Socket based protocols have a special way of failing
Refused connection is bad.
Hanged connection is worse.
Micro-services that talk to each other, will stop talking abruptly
2. Unbalanced Capacities
Specially applicable to micro-services
3. Slow responses
4. Unbounded Result Sets
Major anti-pattern, overlooked by many
What is the size of an HTTP cookie?
Patterns
1. Use timeouts!
What is the default timeout on Ruby’s net/http?
Now and forever, networks will always be unreliable. - Michael
T Nygard
Every network call in your system must have a timeout
This includes database calls
This includes API calls
This includes cache lookups
What to do when the timeout occurs depends on where
it occurred
Highly context specific, so the dev team should make that
decision
ProTip: Do not use Ruby’s “timeout” module
Instead, depend on libraries for the timeout
If you’re a library author, just use net/http’s timeout
2. Circuit Breaker
None
3. Bulkheads
A ship is divided into several water- tight compartments
In case there is a leakage in one section, water
doesn’t flood into other sections
Same principle!
Use resource pools
Use rate limiting
Consideration: Capacity
Bulkheading often conflicts increasing or variable capacity
Consideration: Performance
Bulkheading often results in slightly reduced performance
It’s worth it, trust me™
4. Fail fast
Remember guard clauses in Ruby?
# class Event def closest_event return unless self.location # …
# … end
Same principle!
Any Ruby libraries?
Not a lot :(
shopify/semian
Thank you!
Questions?
Swanand Pagnis Principal Engineer @ First
Swanand Pagnis @_swanand on Twitter
Swanand Pagnis @swanandp on GitHub