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
React Performance Tuning
Search
Recruit Technologies
April 19, 2019
Technology
1
470
React Performance Tuning
2019/4/24 React Tokyo with ZEITでの、辻の講演資料になります
Recruit Technologies
April 19, 2019
Tweet
Share
More Decks by Recruit Technologies
See All by Recruit Technologies
障害はチャンスだ! 障害を前向きに捉える
rtechkouhou
1
670
Flutter移行の苦労と、乗り越えた先に得られたもの
rtechkouhou
3
11k
ここ数年間のタウンワークiOSアプリのエンジニアのチャレンジ
rtechkouhou
1
1.5k
大規模環境をAWS Transit Gatewayで設計/移行する前に考える3つのポイントと移行への挑戦
rtechkouhou
1
1.9k
【61期 新人BootCamp】TOC入門
rtechkouhou
3
42k
【RTC新人研修 】 TPS
rtechkouhou
1
41k
Android Boot Camp 2020
rtechkouhou
0
41k
HTML/CSS
rtechkouhou
10
51k
TypeScript Bootcamp 2020
rtechkouhou
9
45k
Other Decks in Technology
See All in Technology
"TEAM"を導入したら最高のエンジニア"Team"を実現できた / Deploying "TEAM" and Building the Best Engineering "Team"
yuj1osm
1
120
OPENLOGI Company Profile
hr01
0
60k
データエンジニアリング領域におけるDuckDBのユースケース
chanyou0311
9
2.1k
Perlの生きのこり - エンジニアがこの先生きのこるためのカンファレンス2025
kfly8
2
270
2/18 Making Security Scale: メルカリが考えるセキュリティ戦略 - Coincheck x LayerX x Mercari
jsonf
0
170
OPENLOGI Company Profile for engineer
hr01
1
20k
IAMのマニアックな話2025
nrinetcom
PRO
1
150
Snowflakeの開発・運用コストをApache Icebergで効率化しよう!~機能と活用例のご紹介~
sagara
1
420
MIMEと文字コードの闇
hirachan
2
1.4k
スキルだけでは満たせない、 “組織全体に”なじむオンボーディング/Onboarding that fits “throughout the organization” and cannot be satisfied by skills alone
bitkey
0
170
実は強い 非ViTな画像認識モデル
tattaka
2
1.2k
分解して理解する Aspire
nenonaninu
2
1k
Featured
See All Featured
How to Create Impact in a Changing Tech Landscape [PerfNow 2023]
tammyeverts
49
2.3k
KATA
mclloyd
29
14k
A Tale of Four Properties
chriscoyier
158
23k
Build your cross-platform service in a week with App Engine
jlugia
229
18k
Principles of Awesome APIs and How to Build Them.
keavy
126
17k
Bash Introduction
62gerente
611
210k
Product Roadmaps are Hard
iamctodd
PRO
50
11k
Evolution of real-time – Irina Nazarova, EuRuKo, 2024
irinanazarova
6
570
Producing Creativity
orderedlist
PRO
344
40k
Cheating the UX When There Is Nothing More to Optimize - PixelPioneers
stephaniewalter
280
13k
GraphQLの誤解/rethinking-graphql
sonatard
68
10k
Build The Right Thing And Hit Your Dates
maggiecrowley
34
2.5k
Transcript
React Performance Tuning Kento TSUJI @maxmellon Recruit Technologies Co., Ltd.
React Meet up with zeit / 2019.04.19
Welcome to Japan!!!!! At asakusa © 2019 The Hotelier Group
Akasaka K.K. https://anaintercontinental-tokyo.jp/location/asakusa/
Profile • Github: @maxmellon • Twitter: @maxmellon_9039 • Start working
second year at the recruit-tech • Yosuke Furukawa’s subordinate *BN/FXCJF 'SPOUFOE&OHJOFFS ✨ ✨
Let me introduce one My favorite food
Udon ͏ͲΜ
:VVVVN
Let’s talk about main subject
What am I doing in R-tech • Develop “AirSHIFT” •
Develop new features • Improve performance • Performance Hackson in frontend • In other products than AirSHIFT
What is the “AirSHIFT” ?
"EKVTUNFOU 4IJGUT AirSHIFT is web service for store managers of
part time staffs ※ “Store Manager” manages all schedule of part time job in Japanese $SFBUFXPSL 4DIFEVMF -JTUVQTIJGUT $PMMFDUTIJGUT GSPNQBSUT 3FNJOE 1SJOUPVU 3FRVFTU XPSL
So rich UI as Desktop Application
Architecture
BFF (express) Client API Isomorphic Session Data Notification (socket.io) Redis
FCM wrapper (React/Redux) Fetchr CSR SSR DB Push Notification WebSocket OAuth CellPhone Application For Part time worker
͜Ε·ͰͷύϑΥʔϚϯεվળ Performance improvements so far
̎ͭͷΞϓϩʔνͰվળ Improved performance with two approaches https://speakerdeck.com/maxmellon/reactzhi-spa-niokeru-pahuomansutiyuningu ࠶ϨϯμϦϯάͷ࠷దԽ 0QUJNJ[BUJPOSFSFOEFSJOH େ͖ͳςʔϒϧͷ7JSUVBMJ[FEԽ
7JSUVBMJ[FEMBSHFTDBMFUBCMFDPNQPOFOU Nodefest’ 18ɺHTML5 Conf Ͱհ Introduced Nodefest’ 18 and HTML5 Conf
https://speakerdeck.com/maxmellon/reactzhi-spa-niokeru-pahuomansutiyuningu ࠶ϨϯμϦϯάͷ࠷దԽ 0QUJNJ[BUJPOSFSFOEFSJOH େ͖ͳςʔϒϧͷ7JSUVBMJ[FEԽ 7JSUVBMJ[FEMBSHFTDBMFUBCMFDPNQPOFOU Nodefest’ 18ɺHTML5 Conf Ͱհ Introduced
Nodefest’ 18 and HTML5 Conf ৄࡉ͜ͷεϥΠυʹॻ͍ͯ͋Γ·͢ Details are in these slides https://speakerdeck.com/maxmellon/reactzhi-spa-niokeru-pahuomansutiyuningu
Before 13,529ms After 3,612ms CPU x4 slow Fast 3G Improvement
վળ 374%
͔͠͠… but…
13,529ms 3,612ms 3.6ඵ͍ͬͯͷʁ Is 3.6 second so fast?
PGVTFSTXJMMMFBWFJGQBHFMPBEJOH UJNFJTMPOHFSUIBOTFDPOET News Lab Japanese AMP Office Hour: Introduction to
AMP with Duncan Wright, Strategic Partner Manager ΑΓ https://www.youtube.com/watch?time_continue=150&v=3N6yDLP1WUIa
͞ΒͳΔվળ͕ඞཁ Further performance improvement is needed
ࠓ͞ΒͳΔվળʹ͍ͭͯհ͠·͢ Today I’ll introduce further improvement
ϘτϧωοΫͷௐࠪ Investigation of bottlenecks
ௐࠪ݅ Research condition • /sft/monthlyshift/201701 → /sft/monthlyshift/201702 ͷભҠ •
ը໘αΠζ 1440 x 900 (ϝΠϯλʔήοτ) transition Display size Main target
None
ؔͷ࣮ߦ͚ͩͰ110ms 110ms with function execution only ϘτϧωοΫ ͦͷ1 Bottleneck 1
Կͯ͠Δͷ͔ʁ What is actually happening?
None
None
None
None
None
None
େྔͷmomentͷΠϯελϯεΛੜ Create many instances of moment
ϘτϧωοΫ ͦͷ2 Bottleneck 2 ͯ͢ಉ͡ίϯϙʔωϯτ All same components ΞΫγϣϯ͕dispatch͞ΕΔͱ࠶ϨϯμϦϯά͕ى͖Δ Re-render
happens with each action dispatch
ͳͥ࠶ϨϯμϦϯά͕ൃੜʁ Why re-render?
reselectΈΜͳͬͯΔʁ Reselect ͕ ຖճҧ͏ Object Λฦ٫͍ͯͨ͠ reselect was returning different
object each time
None
ଞʹ͍͔ͭ͘ϘτϧωοΫ͕͋ͬͨ There were more bottlenecks
1ͭ1ͭվળͯ͠Ϩϙʔτʹ·ͱΊͨ All of them were resolved and summarized in reports
Moment Λେྔʹੜ͍ͯ͠Δ • ಉ͡ͷ࠶ੜ͍ͯͨ͠ͷͰͻͱ·ͣmemoize • ͯ͢UnixTimeʹΑΔܭࢉʹॻ͖͍ͨ͠ ˠ ͨͩɼ͓ۚपΓσάϨͬͨͱ͖ͷϦεΫ͕େ͖͘அ೦ Hope to
re-implementation by unix time. Memoized instance for a while Issue 1: many moment instances
Reselect ͕ຖճҧ͏ObjectΛੜ͍ͯ͠Δ • Reselect ͷΛνʔϜͰ࠶֬ೝ • ΞϯνύλʔϯΛհͯ͠࠶ൃࢭ Issue 2: reselect
returning different object each time Introduced how to use reselect and anti-patterns to team
ͦͷ݁Ռ Results
50%ͷϢʔβʔ1.5ඵҎʹ Ӿཡ͢Δ͜ͱ͕Ͱ͖ΔΑ͏ʹ ⚪ ⚪ ⚪ ⚪ ⚪ ⚪ ⚪ ⚪
⚪ ⚪ ⚪ ⚪ 50% of users can load the page within 1.5sec
75%ͷϢʔβʔ3.0ඵҎ ⚪ ⚪ ⚪ ⚪ ⚪ ⚪ ⚪ 75% of
users can load within 3.0 sec
Γ25%3ඵͰඳըग़དྷ͍ͯͳ͍ 25% of users take longer than 3 sec
͜ͷ40%͕ͷՄೳੑ ୯७ʹͯΊΔͱ 40% of them might leave
ࠓɼ͜ΕΛղܾ͢ΔͨΊͷࢪࡦΛݕ౼͍ͯ͠Δ We are examining to solve this
͜ΕΒͷϢʔβʔͷಛ • ଞͷϢʔβʔΑΓCPUͷੑೳ͕͍ʢͱਪଌͰ͖Δʣ CPUΛΘͳ͍Ξϓϩʔν͕ඞཁ The characteristic of these users have
lower CPU spec than other users Will need approach that DON'T use CPU
Prefetch
PrefetchʹΑΔϖʔδදࣔͷߴԽ /monthly /daily 44% 10% ཌͷ ༧ఆ֬ೝ͠Α͏ʂ ࣍ͷߦಈΛ༧ APIͷΞΫηεϩά͔ΒΛਪଌ Research
user action from access logs Rendering speed-up using Prefetch
BFF Client API Request Request Learning Server .PEFM Request
BFF Client API Request SSR Request Response Response Learning Server
.PEFM Response Request ୯७Ϛϧίϑաఔ Markov process
BFF Client API Request SSR Request Response Response Learning Server
.PEFM Response Request ୯७Ϛϧίϑաఔ Markov process ࣍ʹऔಘ͖͢ϦιʔεΛϔομʹೖ #''JOKFDUTJOUPUIFIFBEFSSFTPVSDFTSFRVJSFEOFYU
BFF Client API Learning Server .PEFM Parse Header Pre-fetch Request
BFF Client API Learning Server .PEFM Request Request Response Response
LRU-cache SET
BFF Client API Learning Server .PEFM LRU-cache exist ? click
BFF Client API Learning Server .PEFM LRU-cache exist ? click
ߦಈ༧͕ ”͋ͨͬͨ” ࣌ When the expectation was met
BFF Client API Learning Server .PEFM LRU-cache GET click CSR
BFF Client API Learning Server .PEFM LRU-cache exist ? click
ߦಈ༧͕ ”֎Εͨ” ࣌ When an expectation is missed
BFF Client API Learning Server .PEFM click CSR Request Request
Request Response Response Response
WebWorker × Suspence
None
None
ͦͷ··Έ߹ΘͤΔͷগ͠େม A bit difficult to combine these.
None
worker.js
Hello.js
App.js
https://git.io/fjOCk (maxmellon/react-with-comlink-sample)
Comlink ❤ Suspense
Lazy Rendering
None
༏ઌ͕ߴ͍ High priority
༏ઌ͕͍ Low priority
ॏ͍ܭࢉ͕ඞཁ Required heavy calculation
͜ΕΒͷࢪࡦͰͯ͢ͷϢʔβʔ͕ ετϨεແ͘ར༻Ͱ͖Δͷ͕ཧ I hope these measures will make all users
stress-free to use AirSHIFT
͠ɼհͨ͠ΞϓϩʔνҎ֎Ͱ ༗ޮͦ͏ͳ͕͋Εڭ͑ͯ΄͍͠ Please tell me if you know other effective
approaches
͋Γ͕ͱ͏ Thank you
None