2014 3rd International Conference on Advanced Computer Science Applications and Technologies

Frontend and Backend Web Technologies in Social Networking Sites: Facebook as

an Example

Hanin M. Abdullah Ahmed M. Zeki

Department of Information Systems
College of Information Technology College of Information Technology
University of Bahrain University of Bahrain
Sakhir, Kingdom of Bahrain Sakhir, Kingdom of Bahrain
han.eenmoh@hotmail.com amzeki@uob.edu.bh

Abstract—Social Networking has been pointed out as an

important evolution in the social and Internet history. The II. THE LEADING SN: FACEBOOK IN BREIF
leader and the pioneer of this evolution was and still Facebook
The origin of the term SNs was for a different purpose than
despite the big and tough rivalry nowadays. Although
Facebook is very well known and most of the people who surf
the one used nowadays. It was originally used to distinguish
the Internet use it and know how to use it and what the between a network used for business purposes and another
different pages are and the modules it has, a few number of one used for socializing [3]. After the evolution of dynamic
technologists know the technology basis upon which this social websites, the term has been used to reflect those
success sustains. This research paper shades a light on the web websites also; websites that have chat rooms, message
programming aspect and other supporting technologies used boards, and the ability to have discussions and leave
for the frontend and backend of social networks. Facebook is comments [4].
taken as an example. The social networking services are those online services,
platforms and sites that utilize the building blocks of SN and
Keywords—Frontend Web Technology; Backend Web relations between people who share different life aspects like
Technology; Web Programming; Social Networks; Facebook interests, activities, backgrounds, and even connections [4].
Such services usually represent each user in a profile.
I. INTRODUCTION The profile facilitates social links and a variety of additional
Web services are becoming the key technology for the services. The social services mentioned here are mostly web-
implementation of integrated and distributed applications for based with different means for the users to interact over the
B2B, cloud computing, and grid computing. The Internet, such as e-mail and instant messaging [4]. However,
performance of mobile applications has critical impact on some of the most recent services utilizes mobile platform
user acceptance and success. Those applications -including applications with and sometimes without web based platform
mobile applications- require the connectivity to backend [4].
services that are hosted in cloud environments [1,2]. Some popular examples of social networking websites
In web context, frontend is the part that the user sees and are: Facebook, Twitter, LinkedIn, MySpace, Pinterest,
interacts with, such as menus, contact forms, etc. In order to Google Plus+, DeviantArt, LiveJournal, Meetup, Tagged,
design and develop such frontend web interface, certain tools Orkut, CafeMom, myLife, and Multiply [5].
and technologies are to be used, which are usually a Facebook is a social networking service and website
combination of HTML, CSS, and JavaScript all being intended to connect friends, family members, and business
controlled by the browser. associates. Facebook is the most used SN with over 1 billion
On the other hand, the backend usually consists of three registered users worldwide [6]. It is the most common of the
parts: a server, an application, and a DB. Backend networking sites, with several runner ups like twitter and
technologies usually consist of languages like PHP, Ruby, MySpace. It has begun as a networking website for college
Python, etc. students and has expanded to include the whole universe [7].
Today’s sophisticated web applications can’t run without A Harvard student who is very well known as the
both the frontend and backend services. This paper looks into youngest billionaire in the world, Mark Zuckerberg, has
the technologies used for the frontend and the backend founded Facebook which was originally named
specifically in Social Networks (SNs) websites. Facebook “theFacebook” in 2004. It has made a quick successful on
will be taken as an example of the SNs. The rest of this paper Harvard campus and later expanded beyond Harvard into
is organized as follows: the next section will briefly other universities. With the exceptional growing in
introduce the SNs including Facebook. Section 3 focusses on popularity, the founder, Zuckerberg, added two other
the architecture of Facebook, while section 4 and 5 will students, Duston Moskovitz and Chris Hughes, to assist.
discuss the Facebook frontend and backend technologies. After few months only, theFacebook became a college
This will be followed by future challenges and conclusion. networking website for the United States of America [7].

978-1-4799-1845-4/15 $31.00 © 2015 IEEE 86

85
DOI 10.1109/ACSAT.2014.22
 Zuckerberg and his partners left Harvard to run for data and connections. For this reason, cache servers,
theFacebook full time just after taking the site in a national running Linux and the open source Memcache software, fill
level. In 2005, theFacebook has been renamed Facebook. At this gap and receives most of the requests. Nearly 95% of
that time, it was only available to schools, universities, data queries are being filled from the cache servers’ 15 TBs
organizations, and companies with English language only, of RAM. Approximately 500,000 queries per second only
but has later expanded to include anyone everywhere [7]. An have to be passed to the MySQL DBs [8].
insight in the mechanisms of Facebook would show a start The bottom tier consists of eight-core Linux servers
into which users create a profile page that shows their friends running MySQL as a DB server application. Facebook is
and networks information [5]. The network of friends in the estimated to have about 800 servers in this tier distributing
profile reflects the ones who can view the profile. The profile nearly 40 TBs of user data. This tier stores all the data about
includes information, friends, status, photos, notes, groups, every object in the DB [5]. Figure 1 shows a general
and the wall. Users can search for friends by e-mail address, overview of Facebook architecture.
school, university, or just by typing in a name or location for
search. Friends are able to see all of each other’s profiles
including contact information. Messages are allowed
between users for free and recently Facebook acquires
money to send messages to people who are not in the friends
list. Facebook makes money also from advertisements. E-
mail notifications are customizable and let users know when
new friends have added them to their list or when someone
has sent a message to them. A popular feature of Facebook is
sharing photographs uploaded from different types of
sources such as cameras and phones.
Users have the option to allow only friends to see their
pictures and information or to allow anyone. Storage is
unlimited, which is a major advantage of Facebook's
photograph sharing capabilities as free service. Figure 1. Facebook architecture [8].
A major redesign after the first website was in 2008,
which was intended to streamline the website and make it The San Francisco facility replicates the Web and cache
easier to see what friends were doing. It was an outstanding tiers, as well as the filers with the DB objects, but it uses the
expenditure and has boosted its dominance in social Santa Clara MySQL DB tier. The Virginia data center is too
networking. Right after that and specifically in 2009, users of far away to share MySQL DBs with 70 milliseconds of
Facebook worldwide were double that of its older competitor Internet delay, give or take, it just won’t work. Thus, it
MySpace [7]. completely duplicates the Santa Clara -facility, using
MySQL replication to keep the DB tiers in sync [8].
III. THE ARCHITECTURE OF FACEBOOK Facebook had more than 30,000 servers in 2012 which may
Facebook data centers are in Santa Clara, San Francisco, and increase overtime.
Northern Virginia. Those centers were built on the backs of
three major tiers of x86 servers which are loaded with open A. How do Facebook Applications Work?
source software [8]. Some of those open source software Facebook applications are a little different from other online
have been developed by Facebook and made publicly open applications in that they live on two different servers [9]. For
source after that. In the coming paragraphs we will be instance you host the application on your own server, which
exploring the three tiers of the architecture. is responsible for the business logic of your application, and
The top tier of the Facebook network which is closer to Facebook’s servers work with the user’s profile. When a user
users is made up of the Web servers that create the Web interacts with your application, Facebook interprets the
pages seen by users. Those servers are mostly with eight request and passes it to your server. Your pages are
cores based on 64-bit Linux and Apache. Most of the SN’s constructed with FBML (Facebook Markup Language), a
pages and features are created using PHP since the start of superset of HTML, to pass back to the Facebook server farm
the Facebook [8]. PHP is a very common well known [9]. Facebook interprets the FBML response from your
computer scripting language specialized for simple and server and generates the resulting page the user sees. The
automated functions. Although Facebook mainly utilizes process is illustrated in Figure 2.
PHP for its powerful capabilities, it has also developed
complex core applications using a variety of different
computer languages, including C++, Java, Ruby, and Python
[8].
The middle tier is mainly for caching servers. Although
more than 800 DB servers are available in the bottom tier,
they can’t serve up all the needed data [8]. To feel the
difficulty, Facebook receives 15 million requests per second

86
87
 B. MYSQL
MySQL is a DB software that has high speed and high
reliability, the reason for which Facebook is using it. It is
used primarily as a key-value store because data is randomly
distributed among a huge set of logical instances. These
instances are spread out across physical nodes, and the load
balancing is done at the physical node level [13].
For customization purposes, Facebook has developed a
custom partitioning scheme in which all data has an assigned
global ID. Another customization of MySQL from Facebook
is the archiving scheme that is based on the frequency of data
for every user [11].
C. PHP
Facebook uses PHP as the main programming language.
PHP is a web programming language with extensive support
and an active development community [14]. PHP is
appreciated for its rapid iterations and dynamic interpretation
as a scripting language. As mentioned earlier, Facebook
doesn't just use PHP of course, but it is heavily invested in
the language [14].
Figure 2. Facebook page processing [9]. A lot of programmers have asked that why Facebook
doesn’t migrate from PHP to other languages and the best
IV. FACEBOOK FRONTEND TECHNOLOGIES answer came from a Facebook engineer named Wong who
To make up its infrastructure, Facebook uses various tools has worked at the company in various roles between 2005
and programming languages. At the frontend, the servers are and 2010. He stated that it is because PHP has incumbent
based on LAMP (Linux, Apache, MySQL, and PHP) stack inertia and Facebook's engineers have managed to work
[10]. The following paragraphs explains what that means. around many of its flaws through a combination of patches at
LAMP is a combination of free and open source all levels of the stack and excellent internal discipline via
software. It refers to the first letters of the OS Linux, the code convention and style - the worst attributes of the
HTTP server Apache, the DB software MySQL, and language are avoided and coding style is rigidly enforced
programming languages PHP, Perl, and Python. Those are through a fairly tight culture of code review. Engineering
the main components used to build a viable general purpose management has never had to take a strong hand here; this
web server in the frontend of any scalable large scale arose largely due to key internal technical leaders just sort of
servicing website [11]. corralling everyone else along [10].
Although LAMP has the standard components which are D. MEMCHACHE
mentioned but variations are possible with respect to the web
scripting languages for instance where the Perl may be Memcache is a memory caching system that is used to speed
excluded. Other variations exist such as variations of the OS up websites that are driven by DBs, such as Facebook, by
like the WAMP which is based on Windows OS and MAMP caching data and objects in RAM in order to reduce the
which is based on Mac OS although this doesn’t apply in the reading time. Facebook uses Memcache as a primary caching
Facebook case in which Linux is the OS [10]. stage which helps in relieving the DB load, if direct access is
The software combination has become popular between granted for all requests. By having a caching system,
developers although the original designers of these programs Facebook can be as fast as it is in recalling the data [11].
did not design them all to work specifically with each other. Facebook has realized that there are downsides to using
The development philosophy and tool sets are common the LAMP stack. For example, PHP is not inherently
among several platforms because it is free of cost and open optimized for large websites and therefore hard to scale. In
source which make them easy to adapt [10]. addition, it is not the fastest executing language and the
extension framework is difficult to use [14]. Keeping in
A. LINUX & APACHE mind that Memcache is considered as the middle tier as has
Linux is an OS which is based on a Unix kernel. In addition been mentioned before although it is described here under
to the fact that it is an open source OS, what makes is the frontend section.
favorable is the good security and customization ability of it.
V. FACEBOOK BACKEND
Facebook runs the Linux OS on Apache HTTP Servers.
Apache is also free, as the rest of the bundle, and is a very Facebook has used different programming languages for its
popular open source web server in use [10,11,12]. Backend. A variety of Facebook services uses languages
such as Java, C++, Erlang and Python. Facebook doesn’t use
different programs for the sake of the variety but they
initially think of a service which they need to implement and

87
88
then they create the framework/toolset for the service and at There are a lot of more programming languages like
the end the right programming language for the function is bagpipe for dynamic pages but the scope of this report is for
chosen [13]. Facebook uses several different languages for the general overview [15].
its different services. PHP is used for the frontend as
explained earlier, and Erlang is used for Chat, while Java and VI. FUTURE CHALLENGES
C++ are used in several places. More scaling challenges will come and the programming
Next subsections will briefly go through some of the languages and technologies Facebook is using will always be
software Facebook is using in its backend. revised and evolved. The pace into which social networking
A. THRIFT is growing leaded by Facebook is incredible. Its user base is
increasing almost exponentially. Facebook is expected to be
Thrift is a cross-language framework which has been running into different performance bottlenecks as it’s
developed in the labs of Facebook. The function of Thrift is challenged by more and more page views, searches,
to tie all of the different languages Facebook uses together to uploaded images, status messages, and all the other ways that
make it possible for them to talk to each other. Facebook has Facebook users interact with the site and each other [15].
made Thrift open source [13]. Here are a few facts to give an idea of the scaling challenge
B. SCRIBE that Facebook has to deal with [15]:
Scribe is a server that is built on top of Thrift. Scribe has • Facebook serves 570 billion page views per month
been designed for the log data streamed from all of the (according to Google Ad Planner).
servers in real time. It aggregates the log data and it is a • There are more photos on Facebook than all other photo
scalable framework useful for logging a wide array of data
[15]. sites combined (including sites like Flickr).
• More than 3 billion photos are uploaded every month.
C. CASSANDRA • Facebook’s systems serve 1.2 million photos per second,
Cassandra is another open source system which is used by (this doesn’t include the images served by Facebook’s
Facebook. It is a distributed storage system which is used by CDN).
Facebook for its inbox search. What makes Cassandra • More than 25 billion pieces of content (status updates,
special is that it has no single point of failure [16]. comments, etc) are shared every month.
D. HIPHOP FOR PHP
VII. CONCLUSION
Although PHP has a lot of benefits which made Facebook
and a lot of other websites to use it, it still has drawbacks Finally, it is good to say that behind of all the well
such as optimization. Hiphop is a transformer which maintained and operated website that the entire world is
transforms the source code of PHP to an optimized C++ code using a very robust programs and systems that are all in
to be ready for a g++ compiler to make the machine code out harmony. Facebook is providing an easy way for friends to
of it [15]. keep in touch and for anyone to have a presence on the Web
without the need to build a website. People have been
E. HAYSTACK “Facebooking” each other for years now making Facebook
Haystack is an object store system which is used by the most used SN worldwide. Another important outcome
Facebook for the storage of photos. It is a high performance out of this report is that Facebook has a lot of self-invented
system that is handling more than 20 billion uploaded photos technologies and languages, some of which have been made
on Facebook, and each one is saved in four different open source and public to be customized. The reason behind
resolutions. Haystack stores photo data inside 10 GB bucket this was the challenges which haven’t been looked into
with 1 MB of metadata for every GB stored [15]. before Facebook in terms of the huge extraordinary traffic
and users.
F. HADOOP AND HIVE
Hadoop is a map-reduce system that performs calculations REFERENCES
on massive amount of data. Facebook uses this open source [1] M. Debusmann, and K. Geihs, “Towards Dependable Web Services”,
system for data analysis. Hive originated from within 10th IEEE Pacific Rim International Symposium on Dependable
Computing, pp. 5-14, 2004.
Facebook, and makes SQL queries with Hadoop, making it
[2] F. Schulz, and W. Theilmann, “Towards Systematic Mobile Cloud
easier for non-programmers to use. Both Hadoop and Hive Performance Analysis”, 6th Joint IFIP Wireless and Mobile
are open source [15]. Networking Conference (WMNC), pp. 1-4, 2013.
G. VARNISH [3] A. Hashmi, F. Zaidi, A. Sallaberry, and T. Mehmood, “Are All Social
Networks Structurally Similar?”, IEEE/ACM International
Varnish is a load balancer and HTTP accelerator. It also Conference on Advances in Social Networks Analysis and Mining
caches the content which can then be served lightning-fast. (ASONAM), pp. 310-314, 2012.
Serving photos and profile pictures and handling billions of [4] L. Brown, “The Social Networking Handbook - Everything you Need
requests every day is done by Varnish [17]. to Know about Social Networking”, Emereo Publishing, 2011.

88
89
[5] Top 15 Most Popular Social Networking Sites,
http://www.ebizmba.com/articles/social-
networking-websites
[6] Z. Whittaker, “Facebook Hits 1 Billion Active User Milestone”,
CNET, 2012, http://news.cnet.com/8301-1023_3-
57525797-93/facebook-hits-1-billion-active-
user-milestone
[7] L. Li, “Social Network Sites Comparison between the United States
and China: Case Study on Facebook and Renren Network”,
International Conference on Business Management and Electronic
Information (BMEI), vol.1, pp. 825-827, 2012.
[8] A. Zeichick, “How Facebook Works?”
http://www.technologyreview.com/featuredstory/
410312/how-facebook-works
[9] W. Graham, “Reaching Users Through Facebook: A Guide to
Implementing Facebook Athenaeum”, Code4lib Journal, Issue 5,
2008.
[10] D. Dougherty, “LAMP: The Open Source Web Platform”, 2001,
http://www.onlamp.com/pub/a/onlamp/2001/01/25/
lamp.html
[11] J. Lee, and B. Ware, “Open Source Web Development with LAMP:
Using Linux, Apache, MySQL, Perl, and PHP”, Addison Wesley,
2002.
[12] Apache Software Foundation, “Axis architecture guide”, 2003,
Version 1.0, http://ws.apache.org/axis
[13] E. Protalinski, “Why Facebook Hasn't Ditched PHP?”
http://www.zdnet.com/blog/facebook/why-
facebook-hasnt-ditched-php/9536
[14] S. Campbell, “How Does Facebook Work? The Nuts and Bolts:
Technology Explained”,
http://www.makeuseof.com/tag/facebook-work-
nuts-bolts-technology-explained
[15] Pingdom, “Exploring the Software Behind Facebook, The World’s
Largest Site,
http://royal.pingdom.com/2010/06/18/the-
software-behind-facebook
[16] G. Wang, and J. Tang, “The NoSQL Principles and Basic Application
of Cassandra Model”, International Conference on Computer Science
& Service System (CSSS), pp.1332-1335, 2012.
[17] Varnish Cache, https://www.varnish-cache.org/

89
90