Understanding Web Application Security
Understanding Web Application Security
EXECUTIVE SUMMARY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
THE CHALLENGES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Optional Advantages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Business Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Executive Summary
As enterprises move more of their business transactions online, they face the challenge
of defending a perimeter that grows increasingly porous. Proprietary data and business-
critical operations are being exposed through Web interfaces that are accessible from any
where in the world and highly vulnerable to the Internet’s growing threat environment.
The network firewalls that once locked down the enterprise perimeter are ineffective
against Web-based attacks that are quickly rising in frequency, scale, and severity.
These Layer 7 attacks now account for between 60 and 80% of all reported security
incidents. 1 By exploiting common Web application security flaws, the attacks are able
to cause tremendous business disruption, particularly through the theft of sensitive
enterprise information as well as customer and employee personal data.
This paper examines current trends in Web application security, assessing the present
threat environment as well as limitations in existing approaches to protection. It then
looks at how Akamai’s new distributed Web Application Firewall solution overcomes
these challenges, working as an integral part of a defense-in-depth security architecture
to provide robust and scalable protection that is both practical and cost effective.
Introduction
The Escalating Risk of the Insecure Web
Cyberspace is sometimes called the silent battleground, as both hackers and hacked
want to stay off the public radar screen. However, the harm being done to businesses’
online presence is very real. Every industry is at risk: retail and financial sites are targeted
for credit card and account data; enterprises are targeted for intellectual property and
proprietary data; government organizations are targeted for political or ideological
reasons; and popular Web sites—including social media, online gaming, and entertain-
ment destinations—are targeted for their massive user base. Small businesses are not
safe either, as many attacks are untargeted, with cyber criminals using automated
methods to detect and infect vulnerable sites.
Application layer attacks in particular are one of the biggest threats enterprise IT faces
today. These attacks are proliferating as criminals look to exploit the highly vulnerable
and largely unprotected Web application layer that serves as the new enterprise
perimeter—one that unfortunately gives inadequate protection to the business-critical
data and operations within.
The damage being inflicted is serious. A recent Purdue University study involving more
than 800 CIOs estimates that cyber crime cost businesses more than $1 Trillion in 2008,
through theft of data and intellectual property, as well as damage to customer trust and
brand reputation. Repairing the fallout from data breaches is costly, as businesses can
be subject to reporting and notification requirements as well as lawsuits and fines.
Regulatory compliance is another issue. The credit card industry, for example, has
implemented specific regulatory requirements to ensure that merchants involved in
online credit or debit card transactions secure their Web applications in order to
safeguard customer account data.
Credit card numbers and other personal financial account information are popular targets
among cyber thieves, as card numbers can fetch anywhere from $.10 to $25 per account
in the underground economy, and bank account credentials can sell for $10 to $1000 per
account. Symantec estimates that the underground market for this stolen financial data is
worth hundreds of millions of dollars, while the aggregate monetary value of the accounts
themselves runs in the billions. 3
In addition to targeting customer financial data, cyber criminals are also exploiting
Web application security flaws to go after the enterprise’s crown jewels. A recent McAfee
security report notes that stolen intellectual property, driven by industrial espionage, is a fast
growing target as hackers become increasingly sophisticated and aim for higher-value spoils. 4
These lucrative financial rewards, combined with the low barriers to entry and relatively low
risk involved, have propelled the sharp and continuing rise in cyber crime over recent years.
According to Websense, in the first half of 2009, 61 of the top 100 most popular sites
on the Web were found to have been compromised in this way—either hosting malicious
content or containing a hidden redirect to a malicious site. 5 The malware possibilities are
limitless; for example, they may install keystroke loggers to capture user password data, or
rope the user’s machine into part of a botnet that can be leveraged for other cyber attacks.
Another key trend is the rise of automated attack tools, sites that redirects users to malicious sites where malware is
including the use of zombie armies or botnets, which magnify planted on users’ machines. Such an attack compromised over
and multiply the possible scale of attacks. Unfortunately, 70,000 Web sites in January 2008. 10
an estimated 34 million computers in the United States
may now be part of a botnet, a 50% increase over last year. 7 Cross Site Scripting (XSS) – Cross Site Scripting is another
These armies of infected machines give attackers control over common vulnerability, with one report estimating that 65%
a massive number of computing resources, used to launch of websites are susceptible. 11 This vulnerability allows attackers
large-scale SQL injection attacks, distribute malware or spam, to execute a malicious client-side script by injecting it into
perpetrate password and credit card thefts, execute DDoS the URL of a trusted site. XSS enable theft of sensitive user
attacks, or protect malicious Web sites with fast flux hosting. data, including cookies or login information. It can also tamper
with site content, leveraging social engineering techniques and
Botnets and malware are readily available through the black the trusted site’s reputation to trick users into downloading
market, putting these sophisticated and powerful tools at other malware.
anyone’s disposal. This means it takes very little to carry out
a wide-scale attack. In 2007, for example, a single disgruntled Distributed Denial-of-Service (DDoS) – DDoS attacks are
student was able to launch a crippling, month-long attack another fast-rising category, with both the number and size
launched against the Estonian government and other of attacks escalating quickly. According to one survey, denial
commercial entities. Thus, enterprises must be prepared of service attacks were second only to SQL injection as the
to deal with Web attacks on a massive scale. most commonly reported Web vulnerability in the second half
of 2008. 12 With the help of botnets, attackers are now able to
Major Web Application Threats execute denial of service attacks on a far greater scale than ever
before, making them more difficult and costly to defend against.
Unfortunately, while awareness of Layer 7 threats is growing, Moreover, application-layer DDoS attacks are harder to detect
the vast majority of Web applications still remain largely unpro- than network-layer DDoS attacks, as they use massive amounts
tected. According to the Web Application Security Consortium, of “normal” Web requests that may be difficult to distinguish
today more than 87% of Web applications carry a vulnerability from legitimate traffic.
classified as high risk or worse. 8 Moreover, nearly half of the
tested Web applications contained critical or urgent vulnerabili-
ties detectable with automated scanning. Existing Security Approaches
We highlight some of the leading threats here. Traditional network firewalls and intrusion prevention systems
provide insufficient security against these treacherous Web-
SQL Injection – Numerous vulnerability-tracking databases based attacks. To augment these protections, there two primary
cite SQL Injection as the most commonly exploited Web security approaches used today for securing the Web application layer.
flaw in 2008. One report notes that there were a few thousand Each one has its practical limitations, as we will see, but a
SQL attacks per day at the start of 2008, but several hundred comprehensive security architecture is likely to leverage some
thousand per day by the year’s end. 9 techniques of each.
SQL attacks exploit application vulnerabilities, such as user Secure Coding and Code Review Practices
input fields that are not properly filtered, that allow SQL
In theory, designing and building security directly into Web
code to be maliciously inserted and executed in the database.
application code is an excellent approach to protecting sites.
Once the database is broken into, the possibilities are virtually
Best practices include strict code review at product design,
limitless. Attackers can use the breach to steal or tamper with
development, testing, and deployment stages, using both
data, or install malware across multiple systems.
automated and manual methods, including penetration
testing and code vulnerability assessment tools.
The massive Heartland and Hannaford Brothers data breach,
which led to theft of 130 million credit and debit card numbers,
In practice, however, companies lack the resources to carry
was perpetrated by SQL injection. Extensive, botnet-driven SQL
out these time-intensive tasks consistently. It is also difficult
injection attacks have also been used to plant code on trusted
for companies to maintain up-to-date, in-house expertise,
as the technology and security landscapes evolve quickly.
Understanding Web Application Security 5
Moreover, fixing flaws takes time—and expertise, particularly in today’s complex, applica-
tion environment. A recent WhiteHat study showed that companies took anywhere from
one to four months, on average, to fix known Web site vulnerabilities that were rated
either urgent, critical, or high risk. This number does not even account for vulnerabilities
that did not get fixed; 80% of urgent cross site scripting flaws and 70% of urgent SQL
injection flaws remained unresolved during the year-long study. 13
So, even with the best of intentions, code reviews generally aren’t realistic as the primary
way of securing applications. In the face of market pressures, companies simply don’t
have the time or resources to do them.
While WAF appliances can be very effective, they demand significant resources for
deployment and management, as they often involve changes to the existing network
architecture and they can be difficult to scale across large deployments.
Finally, companies must also purchase and manage failover boxes since the WAF
appliances are deployed inline, producing a single, critical point of failure. If the fire-
wall fails, the entire site fails—or, at the very least, lies completely vulnerable to attack.
Akamai WAF keeps malicious traffic away from the origin Web rule set that provides performance-optimized, out-of-the-box
infrastructure by stopping attacks at their source, at the edges security against major exploitation techniques including SQL
of the Internet. In addition, the service scales automatically, Injection, Cross Site Scripting, Buffer Overflow, HTTP Response
on-demand, offering the capability to defend against today’s Splitting and other Web-based attacks. The core rule set also
massive-scale attacks without worry. defends against malicious bots and security scanners, as well
as attempts to access Trojans and backdoors that may have
How Akamai WAF Works infected the system.
Akamai Web Application Firewall WAF detects attacks by filtering Distributed WAF Architecture
incoming HTTP and HTTPS traffic, based on configurable network
and application layer controls. Performing its inspections at the While WAF protects against the most common threats,
edge of the Internet before Akamai serves each request, WAF not every type of Web application attack is best dealt with
can either block or send alerts for any malicious traffic detected. this way. Some sophisticated types of attacks require more
Akamai then responds to legitimate requests by delivering detailed knowledge of the specific applications and network
offloaded content and applications directly from the edge, infrastructure involved and therefore may be best handled at
communicating with the origin server as needed. origin. Thus, WAF provides a highly flexible and efficient outer
defense layer that works either as a stand-alone firewall or in
By focusing on generic attack payload identification rather than concert with an enterprise’s existing Web application security
attack-specific signature detection, Akamai WAF delivers plug- infrastructure—enhancing the robustness and scalability of
and-play protection against a broad set of attacks, including that infrastructure by offloading the more generic functions
zero-day and unknown vulnerabilities. Its core security parameters to the Akamai platform. The centralized defenses are then
are based on ModSecurity, a trusted and proven industry-standard freed up to focus on more application-specific protections.
Attacker
(China)
Origin Server
(USA)
Understanding Web Application Security 7
WAF is enabled across the 50,000-plus servers in Akamai’s global network that deliver
approximately one-fifth of the world’s Web traffic each day. This means massive distributed
firewall capacity is available on demand, eliminating the planning headaches that are
associated with scaling out centralized infrastructure.
Moreover, the redundancy and resiliency of the distributed EdgePlatform ensure that
WAF protective layer is always available—in contrast to centralized firewalls that create
a single point of failure, and require the purchase and management of backup appliances.
With WAF, failover is automatic and built in, so that the origin remains always protected.
Because every component of the Akamai platform is optimized for speed, WAF delivers
security without a performance hit. In practice, WAF customers have seen no degradation
in response times, even with firewall configurations that require over 100 security filters
applied to each request. WAF protects their applications while continuing to deliver the
full acceleration benefits of the EdgePlatform.
Finally, WAF deals with both legitimate and attack traffic at the edges of the Internet,
where it can be most efficiently handled. By detecting and deflecting malicious requests
near their source, the origin is protected and attack traffic is kept from crossing the Internet.
This unique, distributed approach complements the enterprise’s existing centralized security
infrastructure to provide a robust, defense-in-depth architecture.
Understanding Web Application Security 8
• Quick, easy deployment and simplified management Key Business Benefits of Akamai WAF
• Attacks deflected at the edge of the network, far away • Protect branding and revenue by mitigating risks
from the origin associated with system compromise
Within such an architecture, Akamai WAF delivers a robust and cost-effective outer
defensive ring that complements existing centralized security controls. It offloads and
bolsters these centralized security resources while providing the additional scalability
and reach needed to defend an enterprise perimeter that now extends to the edges
of the Internet. The result is a streamlined origin infrastructure, augmented by a flexible,
on-demand outer layer—together providing vigorous application defenses while reducing
IT planning and maintenance headaches.
Akamai WAF
WAF Appliances, Code Review,
Firewall IDS/IPS, Network Vulnerability Scans
Scanning
Web App DB
Server Server
1
Symantec’s Internet Security Threat Report, April 2009, puts the percentage at 60%, while Cenzic’s Web Application Security Trends Report, Q3-Q4 2008,
puts it at 80%, based on data compiled from multiple third-party threat databases.
2
http://www.sophos.com/sophos/docs/eng/papers/sophos-security-threat-report-jul-2009-na-wpus.pdf
3
http://eval.symantec.com/mktginfo/enterprise/white_papers/b-whitepaper_underground_economy_report_11-2008-14525717.en-us.pdf
4
http://www.mcafee.com/us/about/press/corporate/2009/20090129_063500_j.html
5
http://www.websense.com/site/docs/whitepapers/en/WSL_Q1_Q2_2009_FNL.PDF
6
http://resources.mcafee.com/content/NAUnsecuredEconomiesReport, http://www.fas.org/sgp/crs/terror/RL32114.pdf
7
http://www.gtisc.gatech.edu/pdf/CyberThreatsReport2009.pdf
8
http://projects.webappsec.org/Web-Application-Security-Statistics
9
http://www-935.ibm.com/services/us/iss/xforce/trendreports/xforce-2008-annual-report.pdf
10
http://www.sans.org/newsletters/newsbites/newsbites.php?vol=10&issue=2#sID200
11
http://www.whitehatsec.com/home/assets/WPstats_spring09_7th.pdf
12
http://www.cenzic.com/downloads/Cenzic_AppSecTrends_Q3-Q4-2008.pdf
13
http://www.whitehatsec.com/home/assets/WPstats_spring09_7th.pdf
14
For more information, see the Akamai Information Security Management System Overview, which discusses Akamai’s comprehensive network and operational
security policies in greater detail.
www.akamai.com