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
Engineering Large Systems When You're Not Google Or Facebook (test in prod)
Search
Charity Majors
April 30, 2018
Technology
20
5.3k
Engineering Large Systems When You're Not Google Or Facebook (test in prod)
lightning talk at Clever, 4/30/18
Charity Majors
April 30, 2018
Tweet
Share
More Decks by Charity Majors
See All by Charity Majors
Case Studies: Modern Development Practices In Highly Regulated Environments
charity
3
3.4k
Compliance & Regulatory Standards Are NOT Incompatible With Modern Development Best Practices
charity
7
5.2k
Perils, Pitfalls and Pratfalls of Platform Engineering (QCon NYC, 2023)
charity
0
220
The Death of DevOps Has Been Greatly Exaggerated, but Platform Engineering Is Here To Stay
charity
1
280
The Future of Ops Jobs (PlatformCon 2023)
charity
2
110
Observability and the Glorious Future (with Liz Fong-Jones)
charity
0
220
The Engineer/Manager Pendulum (QCon SF 2022)
charity
0
810
Being On Call Does Not Have to Suck.
charity
1
1.2k
The Paradox of Alerts
charity
5
6.6k
Other Decks in Technology
See All in Technology
KubeCon EU: Unlocking new Platform Experiences with Open Interfaces
salaboy
1
380
MongoDB Atlas Vectorsearchではじめる生成AIアプリ開発
chie8842
3
510
GitHub最新情報キャッチアップ 2024年3月
dzeyelid
17
3.2k
オーティファイ会社紹介資料 / Autify Company Deck
autifyhq
7
100k
オブジェクト指向宗教史
tanakahisateru
14
12k
エンジニアブランディングチームの KPI / KPI's of engineer branding team
chaspy
1
140
単回帰分析について数式を追いながら実装してみた
kentaitakura
0
510
Introducing Pkl
enomotok
0
240
暗黙知を集積するプラットフォーム : 「健常者エミュレータ事例集」の取り組み
sora32127
1
170
統計的学習理論読み Chapter 2
kmatsui
1
290
SaaS型Webサービス「カオナビ」のチーム開発でPackage by Featureを取り入れた話/Implementing Package by Feature in kaonavi
kaonavi
0
100
AWS アーキテクチャクイズ
yuu26
2
700
Featured
See All Featured
Done Done
chrislema
178
15k
Adopting Sorbet at Scale
ufuk
66
8.5k
Bash Introduction
62gerente
604
210k
Imperfection Machines: The Place of Print at Facebook
scottboms
257
12k
Understanding Cognitive Biases in Performance Measurement
bluesmoon
6
950
Large-scale JavaScript Application Architecture
addyosmani
501
110k
Optimizing for Happiness
mojombo
369
69k
Designing Dashboards & Data Visualisations in Web Apps
destraynor
225
51k
Reflections from 52 weeks, 52 projects
jeffersonlam
343
19k
10 Git Anti Patterns You Should be Aware of
lemiorhan
644
57k
What's new in Ruby 2.0
geeforr
335
31k
Stop Working from a Prison Cell
hatefulcrawdad
265
19k
Transcript
Engineering Large Systems When You’re Not Google Or Facebook Some
Advice By Charity Majors
None
I blame this guy: Testing in production has gotten a
bad rap.
None
how they think we are how we really are
but *why*?
monitoring => observability known unknowns => unknown unknowns LAMP stack
=> distributed systems
“Complexity is increasing” - Science
Many catastrophic states exist at any given time. Your system
is never entirely ‘up’
We are all distributed systems engineers now the unknowns outstrip
the knowns why does this matter more and more?
Distributed systems are particularly hostile to being cloned or imitated
(or monitored). (clients, concurrency, chaotic traffic patterns, edge cases …)
Distributed systems have an infinitely long list of almost-impossible failure
scenarios that make staging environments particularly worthless. this is a black hole for engineering time
unit tests integration tests functional tests basic failover test before
prod: … the basics. the simple stuff. known-unknowns
behavioral tests experiments load tests (!!) edge cases canaries rolling
deploys multi-region test in prod: unknown-unknowns
test in staging? meh
unit tests integration tests functional tests “What happens when …”
(you know the answer) “What happens when …” (you don’t) behavioral tests experiments load tests (!!) edge cases canaries rolling deploys multi-region test before prod: test in prod:
Only production is production. You can ONLY verify the deploy
for any env by deploying to that env
1. Every deploy is a *unique* exercise of your process+
code+system 2. Deploy scripts are production code. If you’re using fabric or capistrano, this means you have fab/cap in production.
Staging is not production.
Why do people sink so much time into staging, when
they can’t even tell if their own production environment is healthy or not?
That energy is better used elsewhere: Production. You can catch
80% of the bugs with 20% of the effort. And you should. @caitie’s PWL talk: https://youtu.be/-3tw2MYYT0Q
feature flags (launch darkly) high cardinality tooling (honeycomb) canary canary
canaries, shadow systems (goturbine, linkerd) capture/replay for databases (apiary, percona) also build or use: plz dont build your own ffs
Failure is not rare Practice shipping and fixing lots of
small problems And practice on your users!!
Failure: it’s “when”, not “if” (lots and lots and lots
of “when’s”)
Does everyone … know what normal looks like? know how
to deploy? know how to roll back? know how to canary? know how to debug in production? Practice!!~
None
None
None
• Charity Majors @mipsytipsy