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
Scale the Google Way
Search
Terrence Ryan
July 30, 2015
Technology
1
93
Scale the Google Way
A discussion of issues with Google scaling technologies
Terrence Ryan
July 30, 2015
Tweet
Share
More Decks by Terrence Ryan
See All by Terrence Ryan
Vms, Serverless, or Containers
tpryan
0
610
Go for PHP Developers
tpryan
2
1k
Navigating Google Cloud Platform
tpryan
0
410
Which Engine?
tpryan
0
370
Introduction to Containers and Kubernetes
tpryan
2
360
Cloud Next 2017 Roundup
tpryan
2
150
LAMP in Containers
tpryan
1
270
GCP and IoT
tpryan
0
390
30,000,000 Requests in a Hour in the Cloud
tpryan
2
350
Other Decks in Technology
See All in Technology
[MIRU25] NaiLIA: Multimodal Retrieval of Nail Designs Based on Dense Intent Descriptions
keio_smilab
PRO
1
160
金融サービスにおける高速な価値提供とAIの役割 #BetAIDay
layerx
PRO
0
210
経験がないことを言い訳にしない、 AI時代の他領域への染み出し方
parayama0625
0
280
GMOペパボのデータ基盤とデータ活用の現在地 / Current State of GMO Pepabo's Data Infrastructure and Data Utilization
zaimy
2
120
AIエージェントを支える設計
tkikuchi1002
12
2.6k
2025-07-25 NOT A HOTEL TECH TALK ━ スマートホーム開発の最前線 ━ SOFTWARE
wakinchan
0
180
AWS表彰プログラムとキャリアについて
naoki_0531
1
150
猫でもわかるQ_CLI(CDK開発編)+ちょっとだけKiro
kentapapa
0
150
経理出身PdMがAIプロダクト開発を_ハンズオンで学んだ話.pdf
shunsukenarita
1
260
Wasmで社内ツールを作って配布しよう
askua
0
170
風が吹けばWHOISが使えなくなる~なぜWHOIS・RDAPはサーバー証明書のメール認証に使えなくなったのか~
orangemorishita
10
2.6k
With Devin -AIの自律とメンバーの自立
kotanin0
2
960
Featured
See All Featured
Evolution of real-time – Irina Nazarova, EuRuKo, 2024
irinanazarova
8
860
Why You Should Never Use an ORM
jnunemaker
PRO
58
9.5k
Stop Working from a Prison Cell
hatefulcrawdad
271
21k
Optimising Largest Contentful Paint
csswizardry
37
3.4k
Intergalactic Javascript Robots from Outer Space
tanoku
272
27k
4 Signs Your Business is Dying
shpigford
184
22k
Testing 201, or: Great Expectations
jmmastey
43
7.6k
Art, The Web, and Tiny UX
lynnandtonic
301
21k
Writing Fast Ruby
sferik
628
62k
Bash Introduction
62gerente
613
210k
YesSQL, Process and Tooling at Scale
rocio
173
14k
Optimizing for Happiness
mojombo
379
70k
Transcript
@tpryan Scale the Google Way Terry Ryan Developer Advocate
@tpryan Who are you?
None
None
None
None
None
None
None
@tpryan First Cluster History
@tpryan First Cluster •Every machine had a role History
@tpryan First Cluster •Every machine had a role •Every machine
had a name History
@tpryan First Rack History
@tpryan First Rack •Machines belong to groups History
@tpryan First Rack •Machines belong to groups •Slack capacity goes
to waste History
November 1998 December 1998 January 1999 February 1999 March 1999
April 1999
November 1998 December 1998 January 1999 February 1999 March 1999
April 1999 December 1998 50,000
November 1998 December 1998 January 1999 February 1999 March 1999
April 1999 December 1998 50,000 January 1999 150,000
November 1998 December 1998 January 1999 February 1999 March 1999
April 1999 December 1998 50,000 January 1999 150,000 April 1999 500,000
@tpryan Today’s Numbers History
@tpryan Today’s Numbers •Over 3 Billion searches a day History
@tpryan Today’s Numbers •Over 3 Billion searches a day •On
a universe of over 60 Trillion urls. History
@tpryan Today’s Numbers •Over 3 Billion searches a day •On
a universe of over 60 Trillion urls. •We index over 20 billion a day History
@tpryan Today’s Numbers •Over 3 Billion searches a day •On
a universe of over 60 Trillion urls. •We index over 20 billion a day •Average time to deliver under 250ms History
@tpryan Handling this growth took fundamental changes
Networking
70 POP’s 33 Countries B4 WAN
Jupiter Superblock
Jupiter Superblock • 1 Petabit/sec
Jupiter Superblock • 1 Petabit/sec • 100,000 Servers at 10GB/s
Jupiter Superblock • 1 Petabit/sec • 100,000 Servers at 10GB/s
• 10 Libraries of Congress a second
Storage
Colossus Datacenter wide storage
Processing
Productionized Containers
Productionized Containers • Developed as only practical way to manage
Google-scale compute
Productionized Containers • Developed as only practical way to manage
Google-scale compute • Everything at Google runs in a container
Productionized Containers • Developed as only practical way to manage
Google-scale compute • Everything at Google runs in a container • We launch over 2 Billion containers per week.
@tpryan What does this mean for you?
@tpryan
@tpryan Big Table
@tpryan Big Table
@tpryan Big Table MapReduce
@tpryan Big Table MapReduce
@tpryan Big Table Dremel MapReduce
@tpryan Big Table Dremel MapReduce
@tpryan Big Table Borg Dremel MapReduce
@tpryan Big Table Borg Dremel MapReduce
@tpryan
@tpryan Google Cloud Platform
@tpryan Google Cloud Platform Compute
@tpryan Google Cloud Platform Compute Storage
@tpryan Google Cloud Platform Compute Connectivity Storage
@tpryan Google Cloud Platform Compute Connectivity Big Data Storage
@tpryan Google Cloud Platform Compute Connectivity Developer Tools Mobile Big
Data Storage Management
@tpryan Conclusions
None
None
Your App Here
@tpryan Thank You! terrenceryan.com @tpryan This preso: http://bit.ly/tpryan-scale This work
is licensed under a Creative Commons Attribution 2.0 Generic License.