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
Building Resilient Frontend Systems (@frontendne)
Search
Ianfeather
April 06, 2018
1
200
Building Resilient Frontend Systems (@frontendne)
Ianfeather
April 06, 2018
Tweet
Share
More Decks by Ianfeather
See All by Ianfeather
Building Resilient Front End Systems (Smashingconf)
ianfeather
0
120
Building Resilient Frontend Systems (All Day Hey)
ianfeather
1
510
Testing Without Assertions
ianfeather
0
140
Building Resilient Frontend Systems - NationJS
ianfeather
0
210
Establishing Performance Contexts
ianfeather
0
130
Reducing complexity with a Component API
ianfeather
0
200
Web Fonts and Performance
ianfeather
0
210
Featured
See All Featured
Building an army of robots
kneath
306
46k
Making Projects Easy
brettharned
117
6.3k
Docker and Python
trallard
45
3.5k
Documentation Writing (for coders)
carmenintech
73
5k
The Language of Interfaces
destraynor
160
25k
The Web Performance Landscape in 2024 [PerfNow 2024]
tammyeverts
9
780
Principles of Awesome APIs and How to Build Them.
keavy
126
17k
Bash Introduction
62gerente
614
210k
GitHub's CSS Performance
jonrohan
1031
460k
Into the Great Unknown - MozCon
thekraken
40
2k
Build your cross-platform service in a week with App Engine
jlugia
231
18k
Agile that works and the tools we love
rasmusluckow
329
21k
Transcript
BUILDING RESILIENT FRONTEND SYSTEMS Ian Feather - BuzzFeed - @ianfeather
None
RESILIENCE IS FUNCTION IN A HOSTILE ENVIRONMENT
UNDERSTAND YOUR TIERS OF USER EXPERIENCE
GUARANTEE THE MOST BASIC LEVEL OF UX
1. HOW OUR SYSTEMS FAIL 2. DESIGNING FOR FAILURE 3.
MITIGATING RISK 4. LEARNING FROM FAILURE
HOW OUR SYSTEMS FAIL SECTION 1
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE
HTTPS IS TABLE STAKES
HTTPS IS TABLE STAKES
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY
CONTROL YOUR POINTS OF FAILURE
2016
2016 DYN DNS 5 HRS
2016 DYN DNS 5 HRS AWS s3 9 HRS 2017
2016 DYN DNS 5 HRS AWS s3 9 HRS 2017
Fastly CDN 1 HR
2016 DYN DNS 5 HRS AWS s3 9 HRS 2017
Fastly CDN 1 HR AWS S3 2 hrs
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY 3. DEVELOPER ERROR
ADD SLIDE ABOUT SENTRY
SLACK ALERTS
KNOWING IT’S BROKEN BEFORE TWITTER DOES
THEORY VS PRACTICE
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY 3. DEVELOPER ERROR
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY 3. DEVELOPER ERROR 4. THE NETWORK
THEORY VS PRACTICE
THEORY VS PRACTICE
1.04% OF REQUESTS FOR JAVASCRIPT WILL TIMEOUT
13 MILLION OF REQUESTS FOR JAVASCRIPT WILL TIMEOUT
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY 3. DEVELOPER ERROR 4. THE NETWORK
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY 3. DEVELOPER ERROR 4. THE NETWORK 5. USER’S PRIVILEGE
8.9% OF USERS USE SOME FORM OF CONTENT BLOCKER
3.84% WON’T SUCCESSFULLY DOWNLOAD OUR FONTS
38 MILLION PAGEVIEWS PER MONTH
HOW OUR SYSTEMS FAIL 1. MALICIOUS INTERFERENCE 2. 3RD PARTY
AVAILABILITY 3. DEVELOPER ERROR 4. THE NETWORK 5. USER’S PRIVILEGE
HOPE FOR THE BEST?
DESIGN FOR FAILURE SECTION 2
DESIGN FOR FAILURE 1. PRIORITIZE CRITICAL PARTS OF THE PAGE
User FONTS html IMAGES DATA (xhr) IMAGES CSS JS IMAGES
User FONTS html IMAGES DATA (xhr) IMAGES CSS JS IMAGES
Images
User FONTS html IMAGES DATA (xhr) IMAGES CSS JS IMAGES
HTML
None
None
None
DESIGN FOR FAILURE 1. PRIORITIZE CRITICAL PARTS OF THE PAGE
DESIGN FOR FAILURE 1. PRIORITIZE CRITICAL PARTS OF THE PAGE
2. MAKE ERRORS A FIRST CLASS CITIZEN
SOMETHING BROKE! SHOULD I TELL THEM?
IT BROKE. SHOULD I TELL THEM?
IT BROKE. SHOULD I TELL THEM? ✘
IT BROKE. SHOULD I TELL THEM? ✘
IT BROKE. SHOULD I TELL THEM?
None
DESIGN FOR FAILURE 1. PRIORITIZE CRITICAL PARTS OF THE PAGE
2. MAKE ERRORS A FIRST CLASS CITIZEN
MITIGATE RISK SECTION 3
MITIGATE RISK 1. LOCK YOUR RUNTIME DEPENDENCIES
CONTROL YOUR POINTS OF FAILURE
None
MITIGATE RISK 1. LOCK YOUR RUNTIME DEPENDENCIES
MITIGATE RISK 1. LOCK YOUR RUNTIME DEPENDENCIES 2. BUILD IN
REDUNDANCY
HAVE TWO OF EVERYTHING
Asset SERVER 1
Asset SERVER 1 www.asset-server-one.com/styles.css
Asset SERVER 1 www.asset-server-one.com/styles.css
✖ Asset SERVER 1 www.asset-server-one.com/styles.css
✖ Asset SERVER 1 Asset SERVER 2 www.asset-server-one.com/styles.css
✖ Asset SERVER 1 Asset SERVER 2 www.asset-server-two.com/styles.css www.asset-server-one.com/styles.css
✖ Asset SERVER 1 Asset SERVER 2 www.asset-server-two.com/styles.css www.asset-server-one.com/styles.css
Asset SERVER 1 Asset SERVER 2 Proxy service
Asset SERVER 1 Asset SERVER 2 www.asset-server.com/styles.css Proxy service
Asset SERVER 1 Asset SERVER 2 www.asset-server.com/styles.css Proxy service
Asset SERVER 1 Asset SERVER 2 www.asset-server.com/styles.css Proxy service
MITIGATE RISK 1. LOCK YOUR RUNTIME DEPENDENCIES 2. BUILD IN
REDUNDANCY
MITIGATE RISK 1. LOCK YOUR RUNTIME DEPENDENCIES 2. BUILD IN
REDUNDANCY 3. SERVE STALE CONTENT
SERVER
SERVER CDN
SERVER CDN
SERVER CDN
SERVER CDN
SERVER CDN
SERVER CDN
CDN SERVER
CDN ✖ SERVER
CDN ✖ SERVICE WORKER SERVER
CDN ✖ SERVICE WORKER SERVER
MITIGATE RISK 1. LOCK YOUR RUNTIME DEPENDENCIES 2. BUILD IN
REDUNDANCY 3. SERVE STALE CONTENT
LEARN FROM MISTAKES SECTION 4
LEARN FROM MISTAKES 1. POSTMORTEMS
BLAMELESS
HOW DID WE HANDLE IT AS A TEAM?
HOW COULD IT HAVE BEEN PREVENTED?
LEARN FROM MISTAKES 1. POSTMORTEMS
LEARN FROM MISTAKES 1. POSTMORTEMS 2. FIRE DRILLS & CHAOS
TESTING
FIRE DRILLS ARE A SAFE SPACE TO PRACTICE
1. LIMIT IMPACT 2. BE DECISIVE 3. DELEGATE EARLY
CHAOS TESTING
DELIBERATELY INTRODUCE FAILURE TO ENSURE YOUR SYSTEMS ARE RESILIENT
LEARN FROM MISTAKES 1. POSTMORTEMS 2. FIRE DRILLS & CHAOS
TESTING
IN SUMMARY
KNOW WHAT’S IMPORTANT TO YOUR USERS
IDENTIFY HOW YOUR SYSTEM WILL DEGRADE
IDENTIFY POINTS OF FAILURE AND BUILD IN FAIL-SAFES
LEARN FROM EVERY FAILURE
THANK YOU IAN FEATHER - BUZZFEED - @IANFEATHER