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
Keeping Wikipedia fast [WeLoveSpeed]
Search
Peter Hedenskog
September 20, 2019
Technology
1
470
Keeping Wikipedia fast [WeLoveSpeed]
Peter Hedenskog
September 20, 2019
Tweet
Share
More Decks by Peter Hedenskog
See All by Peter Hedenskog
Measuring Web Performance for Wikipedia using synthetic testing tools
soulislove
0
430
Measuring Web Performance Using Selenium
soulislove
2
840
Monitoring Web Performance using Open Source tools (Stockholm)
soulislove
2
240
Monitoring web performance using Open Source tools (San Francisco & Silicon Valley Web Performance Group)
soulislove
1
350
Monitoring web performance using Open Source tools (South Bay JavaScript Meetup)
soulislove
0
240
Optimise your home page (fast as lightning)
soulislove
1
55
Integrating performance tools into continuous delivery
soulislove
0
270
How to make your boss speed-curious and other webperf tricks - coldfront2014
soulislove
0
180
Sitespeed.io Lightning demo @ Velocity Santa Clara 2014
soulislove
0
120
Other Decks in Technology
See All in Technology
Gaze-LLE: Gaze Target Estimation via Large-Scale Learned Encoders
kzykmyzw
0
310
KiroでGameDay開催してみよう(準備編)
yuuuuuuu168
1
120
20250818_KGX・One Hokkaidoコラボイベント
tohgeyukihiro
0
130
Exadata Database Service on Dedicated Infrastructure セキュリティ、ネットワーク、および管理について
oracle4engineer
PRO
1
360
Evolution on AI Agent and Beyond - AGI への道のりと、シンギュラリティの3つのシナリオ
masayamoriofficial
0
140
Amazon Bedrock AgentCore でプロモーション用動画生成エージェントを開発する
nasuvitz
6
400
AWSの最新サービスでAIエージェント構築に楽しく入門しよう
minorun365
PRO
11
610
広島発!スタートアップ開発の裏側
tsankyo
0
220
[OCI Skill Mapping] AWSユーザーのためのOCI(2025年8月20日開催)
oracle4engineer
PRO
2
130
DeNA での思い出 / Memories at DeNA
orgachem
PRO
2
830
Product Management Conference -AI時代に進化するPdM-
kojima111
0
190
自治体職員がガバクラの AWS 閉域ネットワークを理解するのにやって良かった個人検証環境
takeda_h
2
380
Featured
See All Featured
The MySQL Ecosystem @ GitHub 2015
samlambert
251
13k
Statistics for Hackers
jakevdp
799
220k
Design and Strategy: How to Deal with People Who Don’t "Get" Design
morganepeng
131
19k
個人開発の失敗を避けるイケてる考え方 / tips for indie hackers
panda_program
110
20k
Imperfection Machines: The Place of Print at Facebook
scottboms
268
13k
Fireside Chat
paigeccino
39
3.6k
Done Done
chrislema
185
16k
It's Worth the Effort
3n
187
28k
Agile that works and the tools we love
rasmusluckow
329
21k
How to train your dragon (web standard)
notwaldorf
96
6.2k
Building Better People: How to give real-time feedback that sticks.
wjessup
367
19k
Building a Scalable Design System with Sketch
lauravandoore
462
33k
Transcript
Keeping Wikipedia fast Peter Hedenskog - @soulislove
Keeping Wikipedia fast Peter Hedenskog - @soulislove
@soulislove
@soulislove Sweden France
@soulislove
@soulislove
What to do? @soulislove
@soulislove
@soulislove
@soulislove NO!!!!!
@soulislove Jean Bernadotte?
@soulislove
@soulislove Sweden France
@soulislove
@soulislove
@soulislove
Lets talk about performance @soulislove
@soulislove Today Our setup RUM & Synthetic Learnings I’ve got
the last four years Case study one regression
@soulislove https://news.ycombinator.com/item?id=20903868
@soulislove https://phabricator.wikimedia.org
@soulislove https://grafana.wikimedia.org
Why is performance important? @soulislove
We want to bring free knowledge to the world independently
of where you live and your economic status. @soulislove
@soulislove
Engineers/dev cares too! @soulislove
Why is performance hard (for us)? @soulislove
Keeping Wikipedia fast @soulislove https://news.ycombinator.com/item?id=20903868 Keeping Wikipedia fast is easy
right?
@soulislove
The Wikipedia Performance Team Challenge All Wikis are different (JS/CSS)
All pages are different (JS/CSS) All users are different (JS/CSS) @soulislove
Our history of performance testing @soulislove
PHP -> RUM-> Synthetic @soulislove
RUM @soulislove
@soulislove Metrics from real users Sampled (1/100) Buckets: platform, browser,
location https://github.com/wikimedia/mediawiki-extensions-NavigationTiming https://grafana.wikimedia.org/d/000000143/navigation-timing?refresh=5m&orgId=1
@soulislove
@soulislove
@soulislove
How we use RUM @soulislove Metrics from “all” users/scenarios Median,
75, 95, 99 - percentiles Alert on regressions First Paint / LoadEventEnd BFF with synthetic
Synthetic testing @soulislove
@soulislove
@soulislove Browsertime + WebPageReplay
That flat line @soulislove
Deviation @soulislove
How we use synthetic @soulislove Fixing the chaos (or creating
more?) Wayback machine Three URLs per alert First Visual Change BFF with RUM
Learnings: Synthetics @soulislove
Validate metrics! @soulislove
@soulislove
https://phabricator.wikimedia.org/T187981 @soulislove
@soulislove
page1 != page2 @soulislove
@soulislove Deviation
@soulislove User journey: second view
@soulislove User journey: second view
Server matters! @soulislove
1. AWS vs GCS vs other cloud providers 2. Servers
change over time (what runs on the same physical server?) 3. C4.xlarge != C4.xlarge @soulislove
@soulislove https://phabricator.wikimedia.org/T192138 https://phabricator.wikimedia.org/T192138 https://phabricator.wikimedia.org/T192138 https://youtu.be/pYbgcDfM2Ts?t=1575
Testing multiple steps are hard @soulislove
How long time do your user stay on each page?
How long do browsers keep HTTP connections open? @soulislove
Browser versions matter @soulislove
None
Know when browsers are updated!!! @soulislove
Learnings: RUM @soulislove
RUM can be good for finding regressions @soulislove
None
User Timing API != what shows on screen @soulislove
@soulislove
New element timings are better! @soulislove
@soulislove
Browser versions are important @soulislove
@soulislove
@soulislove
The hidden tabs incident @soulislove
@soulislove The idea: async all the things … use setTimeout
to run things later.
@soulislove https://phabricator.wikimedia.org/T146510 setTimeout
@soulislove https://phabricator.wikimedia.org/T146510
@soulislove 10% of the traffics opens in another tab!
What are we missing? @soulislove
@soulislove
@soulislove RUM Higher sample rate Buckets per page type Which
metrics are important?
@soulislove Synthetic Real mobile phones (T197847) Easier for devs to
add tests (T225416)
Case study: 3/9-2019 incident @soulislove https://phabricator.wikimedia.org/T231929
@soulislove Firefox
@soulislove Firefox
@soulislove Chrome
@soulislove Firefox
@soulislove WebPageTest
@soulislove
@soulislove TTFB? No: because visible with WebPageReplay
@soulislove ttfb?
@soulislove ttfb?
@soulislove ttfb?
@soulislove Screenshots/video Diff the HAR using https://compare.sitespeed.io or size per
content type in Graphite
@soulislove ttfb?
@soulislove We got 311 span class=“cs1-visible-error”!!! Citation errors: not shown
to readers https://en.wikipedia.org/wiki/Help:CS1_errors
@soulislove Credits Pippi and father - SVT Quick et Flupke
- Hergé The king shouting - Expressen The scream - Edward Munch Napoleon - Horace Vernet Engineers India Space Shuttle - Expressen Various pictures of Carl Gustaf - Swedish tax payers through the apanage
@soulislove Questions?? @soulislove
[email protected]