social networks The technology stack of Philipp Weber HDE MTS 2015.01.23
Jul 17, 2015
social networksThe technology stack of
Philipp Weber HDE MTS 2015.01.23
Table of Contents
● Overview
● Included networks
● The stacks
● Comments / Takeaway
*Some of the displayed numbers and
technologies reach back to the year of 2011.
**Rights to the registered Names / Logos /
Trademarks featured are property of and
belong to their respective owners.
Disclaimer
Facebook - 1.35 billion monthly active
users
Tumblr - 420 million users
Twitter - 284 million monthly active
users
LinkedIn - 300 million users
Google+ - 300 million monthly active
users
Included Networks
Frontend Backend
The stack
Databases Cache(Hosting)
Meta
Templating /
Javascript
Intermediate-
Layer
Programming
Languages
Server
specifics
Offline
processing
The stack - Facebook
● 864 million daily active
users
● 700 million mobile daily
active users
Haystack
(Object-store)
Memcached
markup
Language
PhpCassandra
Hadoop +
Hive
BigPipe
→ pagelets
Thrift
JavaVarnish
HTTP-accel.
HipHopVM -
Php-compiler
Frontend Backend Meta
The stack - Tumblr
● Microblog / social
networking website
● ~40k requests per second
● 15 billion page views per
month
● belongs to Yahoo
The stack - Tumblr
HBase
Redis
(Hosting)
Gearman and
Kestrel
PHP
HA-Proxy
Scala
Ruby
MySQL
Apache Kafka
Finagle
Frontend Backend Meta
The stack - Twitter
● 80% of Twitter active users
are on mobile
● Average of ~5750 Tweets/s
● Record 143.199 Tweets/s
Flock Memcached(Hosting)Bootstrap
JVM
Finagle
Java
Netty
JavaScriptRedis
Hadoop+
Scalding
Scala
Frontend Backend Meta
User-Store
The stack - LinkedIn
● Leader in Professional
networking
● Opened their Network data
to busines-research in 2008
The stack - LinkedIn
Oracle Voldemort
Node.js
Java
GRails =
Java-Rails
Hadoop
Espresso Scala
JRuby
JavaScript
MySQLJQuery
Backbone.js
Frontend Backend Meta
The stack - Google+
● Build on Circle API
● Usually use special Google
technology and libraries
● Close interaction with
android planned
The stack - Google+
BigTable
Closure
Java
Custom JVM
GoogleFile
System
Closure C++
PythonJQuery
Frontend Backend Meta
Takeaway
So what can we learn from these examples?
Takeaways
#1 Embrace Open Source
Open Source systems benefit from university
research and a big community. Facebook
and Twitter release or found many of their
software in OpenSource projects.
Takeaways
#2 Scalability:
Prepare for scalability starting at the
foundation. Monolithic systems make rapid
growth impossible or error-proned.
Takeaways
#3 Discover new solutions:
Highscalability.com writes a very in-depth
technology-blog with interviews from tech-
giants and their experiences. Definitely worth
a read.
Thank you for
your Attention
Sources
https://github.com/facebook
http://www.slideshare.net/meet.hak/facebook-technology-stack
http://newsroom.fb.com/company-info/
www.infoq.com/presentations/Scale-at-Facebook
https://about.twitter.com/company
https://blog.twitter.com/2013/new-tweets-per-second-record-and-how
http://www.slideshare.net/caniszczyk/twitter-opensourcestacklinuxcon2013
Tumblr
http://www.theguardian.com/technology/2014/oct/23/tumblr-yahoo-revenues-2015-ads-nsfw
http://highscalability.com/blog/2012/2/13/tumblr-architecture-15-billion-page-views-a-month-and-harder.html
https://press.linkedin.com/news-releases
https://engineering.linkedin.com/technology
https://www.youtube.com/watch?x-yt-ts=1421914688&x-yt-cl=84503534&v=OXN3wuHUBP0
Google+
http://www.usatoday.com/story/tech/2013/10/29/google-plus/3296017/
http://highscalability.com/blog/2011/7/12/google-is-built-using-tools-you-can-use-too-closure-java-ser.html