See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/262689075
Overview of Facebook scalable architecture
Conference Paper · May 2014
DOI: 10.1145/2618168.2618198
CITATION                                                                                               READS
1                                                                                                      6,079
5 authors, including:
            Daniel Barrigas                                                                                       Melyssa Barata
            University of Coimbra                                                                                 8 PUBLICATIONS   16 CITATIONS   
            2 PUBLICATIONS   5 CITATIONS   
                                                                                                                     SEE PROFILE
               SEE PROFILE
            Pedro Furtado                                                                                         Jorge Bernardino
            University of Coimbra                                                                                 Instituto Politécnico de Coimbra
            224 PUBLICATIONS   806 CITATIONS                                                                      184 PUBLICATIONS   696 CITATIONS   
               SEE PROFILE                                                                                           SEE PROFILE
Some of the authors of this publication are also working on these related projects:
              Medical Imaging and Machine learning View project
              Networked Distributed Data Management View project
 All content following this page was uploaded by Pedro Furtado on 30 May 2014.
 The user has requested enhancement of the downloaded file.
Daniel	
  Barrigas,	
  Hugo	
  Barrigas,	
  Melyssa	
  Barata,	
  Pedro	
  Furtado	
  and	
  Jorge	
  Bernardino,	
   5/30/14	
  
"Overview	
  of	
  Facebook	
  Scalable	
  Architecture",	
  in	
  ACM	
  InternaJonal	
  Conference	
  on	
  
InformaJon	
  Systems	
  and	
  Design	
  of	
  CommunicaJon,	
  ISDOC’14,	
  May	
  16–17,	
  2014,	
  Lisboa,	
  
Portugal.	
  
          ISDOC 2014
           Lisboa, May 16-17, 2014                                     AGENDA
                                                                         Introduction
                                                                         How Facebook works
                            OVERVIEW OF FACEBOOK                         Facebook’sArchitecture
                            SCALABLE ARCHITECTURE                        Conclusions
                            Hugo Barrigas
                            Daniel Barrigas
                            Melyssa Barata
                            Pedro Furtado
                            Jorge Bernardino
           INTRODUCTION                                                HOW FACEBOOK WORKS
             Scalability is essential for web applications with         Facebook  needs to handle large amounts of
              massive numbers of users.                                   information every second.
             The high growth rate of web based systems
                                                                         Facebook continues to be a LAMP (Linux-
              having hundreds of thousands of users accessing             Apache-Memcached-PHP) website, but had to
              it continuously led to response time problems for
                                                                          change and expand its operation to incorporate
              users who are trying to receive information at the
              same time.                                                  other elements and services.
             The current architecture of Facebook is very               Personalized and implemented systems also exist
              large and consists of many technologies and                 inside Facebook, such as Haystack.
              thousands of servers.
             In this presentation Facebook’s architecture and
              how it handles scalability issues is going to be
              described.
           FRONT-END                                                   LINUX & APACHE
             Facebook  uses a variety of services, tools, and           Linux   is a kernel operating system.
              programming languages to make up its core                  It is open source, highly customizable and good
              infrastructure.                                             for safety.
             At the front-end, their servers run a LAMP                 Facebook uses Linux with the Apache HTTP
              (Linux, Apache, MySQL, and PHP) stack with                  server.
              Memcache.
                                                                         Apache is also open source and is the most
                                                                          popular web server.
                                                                                                                              1	
  
Daniel	
  Barrigas,	
  Hugo	
  Barrigas,	
  Melyssa	
  Barata,	
  Pedro	
  Furtado	
  and	
  Jorge	
  Bernardino,	
   5/30/14	
  
"Overview	
  of	
  Facebook	
  Scalable	
  Architecture",	
  in	
  ACM	
  InternaJonal	
  Conference	
  on	
  
InformaJon	
  Systems	
  and	
  Design	
  of	
  CommunicaJon,	
  ISDOC’14,	
  May	
  16–17,	
  2014,	
  Lisboa,	
  
Portugal.	
  
           PHP & BIGPIPE                                                  PHP & BIGPIPE
             PHP
                 fast iterations
                 easy to use
                 Simple to learn
                 good web programming language with extensive
                  community support.
             BigPipeis a dynamic web page system developed
             by Facebook.
                 The general idea is to perform pipelining of sections
                  through the implementation of various stages within
                  web browsers and servers.
           HIP-HOP                                                        BACK-END
             Developed by Facebook                                         At  the heart of the application back end are the
             PHP  compiler.                                                 application servers.
             Converts PHP scripts into optimized C++ code.                 Application servers are responsible for answering
                                                                             all queries and take all the writes into the
             Created to minimize server resources.
                                                                             system.
             Some key features of HipHop are:
                                                                            They also interact with a number of services to
                 Easy to implement extensions
                                                                             achieve this.
                 Reduces CPU and Memory usage significantly
           MYSQL                                                          SCRIBE
             Fastand reliable                                              Scalable distributed logging framework
             Thousands of MySQL servers                                    Useful for logging a wide array of data
                 Users randomly distributed across these servers           Simple data model
             Relational    aspect of DB is not used                        Built on top of Thrift
                 No joins. Logically difficult(Data is distributed
                  randomly)
                 Primarily key-value store
                                                                                                                                 2	
  
Daniel	
  Barrigas,	
  Hugo	
  Barrigas,	
  Melyssa	
  Barata,	
  Pedro	
  Furtado	
  and	
  Jorge	
  Bernardino,	
   5/30/14	
  
"Overview	
  of	
  Facebook	
  Scalable	
  Architecture",	
  in	
  ACM	
  InternaJonal	
  Conference	
  on	
  
InformaJon	
  Systems	
  and	
  Design	
  of	
  CommunicaJon,	
  ISDOC’14,	
  May	
  16–17,	
  2014,	
  Lisboa,	
  
Portugal.	
  
           SCRIBE                                                      THRIFT
                                                                         Lightweight  software framework for cross-
                                                                          language development
                                                                         Very quick
                                                                         Swaps information between various applications
                                                                          developed in different languages.
                                                                         Thrift protocol offers serialization between
                                                                          several languages.
           MEMCACHED                                                   MEMCACHED
             Key-value   memory storage system for arbitrary
              pieces of information that result either from
              database research or page rendering.
             System uses client-server architecture.
             Server maintains a key-value vector association
             Client populates this vector and performs
              research on it.
             The server then stores these values in memory
                  therefore decreases reading time.
             If
               a server consumes the available memory, it
             removes older values.
           HADOOP & HIVE                                               HADOOP & HIVE
             Hadoop   is ideal for scalable systems with an
              enormous need to store and process large
              amounts of information.
             Scales horizontally.
             Instead of replacing existing storage systems,
              Hadoop complements them.
             This process allows existing systems to serve
              data in real time or provide transactional
              interactive business intelligence.
                                                                                                                              3	
  
Daniel	
  Barrigas,	
  Hugo	
  Barrigas,	
  Melyssa	
  Barata,	
  Pedro	
  Furtado	
  and	
  Jorge	
  Bernardino,	
   5/30/14	
  
"Overview	
  of	
  Facebook	
  Scalable	
  Architecture",	
  in	
  ACM	
  InternaJonal	
  Conference	
  on	
  
InformaJon	
  Systems	
  and	
  Design	
  of	
  CommunicaJon,	
  ISDOC’14,	
  May	
  16–17,	
  2014,	
  Lisboa,	
  
Portugal.	
  
                     HAYSTACK                                                HAYSTACK
                       Efficientstorage of billions of photos.
                       High performance.
                       Highly scalable.
                       Uses extensive caching in its main memory.
                     FACEBOOK’S ARCHITECTURE                                 FACEBOOK’S ARCHITECTURE
                                                                                    OVERVIEW OF FACEBOOK SCALABLE
                     CONCLUSIONS                                                    ARCHITECTURE
                       Facebook  has gained a tremendous amount of
                        popularity in the last years and has become the
                        most successful social networking website ever
                        created.
                       If we look past all of the features and
                        innovations, the main idea behind Facebook is
                        really very basic “keeping people connected”.
                       Facebook realizes the power of social networking
                        and is constantly innovating to keep their service
                        the best in the business.                                       Hugo Barrigas
                                                                                        Daniel Barrigas
                                                                                        Melyssa Barata
                                                                                        Pedro Furtado
                                                                                        Jorge Bernardino
                                                                                                                              4	
  
   View publication stats