2012 Global Security Report

Dear Reader,
Organizations, regardless of industry and size, continue to face similar information security risks. Old systems with known flaws can
take time to decommission and new systems are implemented with little or no thought to security. In its third year, the Trustwave
2012 Global Security Report will help you understand today’s information security threat landscape, as well as how to better
protect your organization from cyber attacks in the years ahead.

The Trustwave 2012 Global Security Report is a reflection and analysis of investigations, research and other client engagements
conducted throughout 2011. During the past year, Trustwave SpiderLabs investigated more than 300 breaches and performed
more than 2,000 penetration tests around the world.

Research featured in the report is collected from the many data sources maintained by Trustwave, such as our managed security
service and SSL offerings, allowing us to bring new perspectives to the global state of information security.

We’re excited to share the Trustwave 2012 Global Security Report with our customers and the industry at large. By understanding
how breaches happen, and sharing that knowledge with you, we work to eliminate information security threats for all businesses.

Regards,

Nicholas J. Percoco
Senior Vice President & Head of SpiderLabs
Contributors
Authors
Ryan Barnett
Sol Bhala
Marc Bown
Jonathan Claudius
Josh Grunzweig
Rob Havelt
Charles Henderson
Jibran Ilyas
Ryan Jones (UK)
Ryan Jones (U.S.)
Paul Kehrer
Mike Kelly
Ryan Merritt
John Miller
Steve Ocepek
Nicholas J. Percoco (lead)
Garret Picchioni
Christopher E. Pogue
Michael Ryan
Luiz Eduardo Dos Santos
Sean Schulte
Colin Sheppard
Barrett Weisshaar
Chris Woodbury
John Yeo

Editor
Sarah B. Brown

Art Direction and Design

Nathan Glick

Organization Contributors
CERT.br
United States Secret Service

Contact Us
For comments or questions regarding this report, please contact Trustwave SpiderLabs at the information listed below.

To request information about our services for environments or applications, we at Trustwave SpiderLabs are available to discuss
any organization’s needs.
+1 312 873-7500
[email protected]
https://www.trustwave.com/spiderlabs
Twitter: @SpiderLabs / @Trustwave
 1 Executive Summary

Table of Contents 3 2011 Incident Response Investigations

3
3
4
5
6
6
6
Unique Data Sources, Countries and Methodologies
Types of Data Targeted
Industries
Target Assets
System Administration Responsibility
Detection
Attack Timeline
7 The Breach Triad
7 Infiltration
9 Aggregation
9 Exfiltration
10 International Perspectives
10 Attacker Source Geography
11 Europe Middle East and Africa
13 Asia-Pacific
15 Latin America and Caribbean
17 Malware Statistics
17 Common versus Targeted Malware
18 Data Export Functionality
18 Malware Types
18 Data Export

19 United States Secret Service:

Protecting the Nation’s Leaders and Financial Infrastructure
21 Security Weaknesses under the Microscope
22 In the Workplace: Four Vulnerable Resources
22 The Network – Legacy Issues Still At Large
26 What’s in Our Inbox? 2011 Email Trends
29 The Web – Multi-Vector Analysis of Modern Attack Techniques
34 Blind Faith in Mobile
35 Our Defenses: Four Basic Controls
35 Business Password Analysis
41 A Study of SSL
46 Anti-Virus: The Elephant in the Room
47 Walking through Firewalls

51 Information Security Strategy Pyramid for 2012

51 Education of Employees
52 Identification of Users
52 Homogenization of Hardware and Software
53 Registration of Assets
53 Unification of Activity Logs
54 Visualization of Events

55 Global Conclusions
56 Appendix: What is a Penetration Test?
1 Executive Summary

Nearly every week in 2011 brought reports of data breaches

in the media, ranging from the theft of personally identifiable
information to sensitive government documents to credit card
data. Cyber criminals targeted many diverse organizations. Those
most affected represent a broad spectrum of organizations that
have one thing in common: valuable data.

2012 Key Findings

Each year we strive to issue an informative and educational
report on the latest security issues and trends, as well as provide
insight into unaddressed legacy issues.

• Customer records remained a valuable target for attackers,

making up 89% of breached data investigated.

• For the second year, the food and beverage industry made
up the highest percentage of investigations at nearly 44%.

• Industries with franchise models are the new cyber

targets: more than a third of 2011 investigations occurred
in a franchise business.

Executive
• In 76% of incident response investigations, a third party
responsible for system support, development and/or
maintenance of business environments introduced the

Summary
security deficiencies.

• Law enforcement detected more breaches in 2011 – up

from 7% in 2010 to 33% in 2011.

• Data harvesting techniques continued to target data “in-

transit” within victim environments showing up in 62.5%
of 2011 investigations.

• Anti-virus detected less than 12% of the targeted malware

samples collected during 2011 investigations.

• For Web-based attacks, SQL injection remains the number

one attack method for the fourth year in a row.

• The most common password used by global businesses

is “Password1” because it satisfies the default Microsoft
Active Directory complexity setting.

The Trustwave 2012 Global Security Report highlights these risk

areas and more, offering predictions on future targets based on
our analysis and perceived trends.

Real-World Data, Expert Analysis

The Trustwave 2012 Global Security Report is founded on data from
real-world investigations and research performed by Trustwave
SpiderLabs in 2011. Standardized tools were used to record data
and other relevant details for each case or test. Trustwave is strongly
committed to protecting the privacy of our clients, and the statistics
within this report are presented in an aggregate form only.
 Executive Summary 2

The report follows four distinct sections:

2011 Incident Response Investigations Information Security Strategy

This section analyzes the results of more than 300 incident Pyramid for 2012
response investigations performed due to a suspected security
breach identified by either the target organization or a third party, To improve any organization’s security posture, Trustwave
such as a regulatory body, law enforcement or other group. SpiderLabs recommends six areas to focus on in 2012:

• Education of Employees — The best intrusion detection

Security Weaknesses under systems are neither security experts nor expensive
technology, but employees. Security awareness education
the Microscope for employees can often be the first line of defense.

This section features data correlation and analysis from many • Identification of Users — Focus on achieving a state
sources, including: where every user-initiated action in your environment is
identifiable and tagged to a specific person.
• Analysis of more than 2,000 penetration tests performed
on 300,000 devices. • Homogenization of Hardware and Software —
Fragmentation of enterprises computing platforms
• Review of 25 different anti-virus vendors against the is an enemy to security. Reducing fragmentation
various malicious files Trustwave SpiderLabs encountered through standardization of hardware and software,
in 2011. and decommissioning old systems, will create a more
• Data from more than 2 million network and application homogenous environment that is easier to manage,
vulnerability scans. maintain and secure.

• Analysis and trends from 16 billion emails collected from • Registration of Assets — A complete inventory or registry
2008 to 2011. of valid assets can provide the insight needed to identify
malware or a malicious attack.
• Review of approximately 300 Web-based breaches
publicly disclosed by organizations in 2011. • Unification of Activity Logs — Combining the physical
world with the digital affords organization new ways to
• Usage and weakness trends of more than 2 million real- combine activities and logs to identify security events
world passwords used within corporate information systems. more quickly.
• Analysis of almost 300,000 unique digital certificates • Visualization of Events — Log reviews alone are no longer
(SSL) from scans of more than 17 million Internet-facing sufficient. Visualizing methods to identify security events
devices, including Online Certificate Status Protocol within the organization better narrow security gaps.
(OCSP) usage data from Trustwave.
• A review of 250,000 public devices from 132 different
countries for Broken Network Address Translation Global Conclusions
(BNAT) instances that could expose internal services to
external attackers. Any business can be a target; those most susceptible will be
businesses that maintain customer records or that consumers
frequent most, such as restaurants, retail stores and hotels. The
risk is even greater for brand name chains. Areas of focus for
2012 include employee security awareness, anti-virus software
and legacy firewalls.

By learning from others’ misfortunes or vulnerabilities, and

applying tactical and strategic change outlined in this report,
any organization will be better able to reduce the likelihood of
incidents and resultant data loss.
3 2011 Incident Response Investigations

Types of Data Targeted

2011 Incident Continuing the trend of previous years, 89% of investigations

involved the theft of customer records, including payment card
data, personally identifiable information and other records,

Response
such as email addresses. Active email addresses of consumers
are valuable to attackers as they can lead to further attacks like
traditional phishing or sophisticated, targeted attacks. Cyber

Investigations
criminals continue to focus their efforts in this area due to the
large number of available targets and well-established black
markets where criminals are quickly able to turn items such as
payment card data into cash with minimal effort.

Trustwave incident response engagements are undertaken

in response to a security issue, either identified by the victim Trustwave SpiderLabs is one of a
organization or a third party, such as law enforcement or a
regulatory body. Data from these investigations are analyzed few firms authorized to conduct
and findings and trends are presented in an aggregated form.
It is important to note that the data presented in this report are
payment card data breaches on
not survey data — all data within this section are from actual behalf of all five major card brands
Trustwave SpiderLabs investigations.
and, as a result, payment card
Unique Data Sources, data breach investigations remain
Countries and Methodologies prevalent within the data set.
Several engagements in 2011 found that criminals explicitly
In 2011, Trustwave SpiderLabs performed more than 300 data
targeted business financial account numbers (e.g., account routing
breach investigations in 18 countries. More investigations were
codes, merchant identification numbers) to perpetrate payment
conducted in the Asia-Pacific (APAC) region than in the previous
card fraud. When merchant identification numbers from legitimate
year, primarily the result of maturing data disclosure laws and
businesses are obtained, criminals utilize this information to
compliance mandates. For example, more countries in the APAC
configure their own fraudulent payment systems and perform
region are adopting and adhering to the Payment Card Industry
card testing with stolen payment card accounts. These fraudulent
Data Security Standard (PCI DSS). With this adoption more
transactions then appear to originate from a legitimate business.
organizations are made aware of their obligation to report data
breaches when they occur. Similarly, the Latin America–Caribbean
This process is also used to launder money through an
(LAC) region had increased data breach disclosure procedures
unsuspecting merchant. For instance, an attacker can use a batch
and adoption of compliance mandates, such as PCI DSS.
of payment cards to make purchases and then perform credits
(or charge-backs) to a small set of payment cards. The result is
the consolidation of value from stolen cards to payment cards
that are in the control of the attacker. The business unknowingly
facilitating the transactions does not lose or gain anything except a
small transaction processing fee during the process, as the money
received is equal to the amount transferred out of their accounts.

By far, the theft of trade secrets were the most advanced breaches
in terms of attacker technical skill level and persistence. Trade
secrets are unique to a given entity and, unlike payment card
data, an attacker cannot simply move on to another target

>300 18
organization to obtain this information. Therefore, efforts to gain
trade secret data are far more focused.

Data Breaches Countries

 2011 Incident Response Investigations 4

New this year, electronic protected health information (ePHI)

Industry Breakdown and Data Targeted theft investigations accounted for 3% of the caseload. We
attribute this addition to the continued adoption of breach
notification laws, and a maturing of information security policies
within the health care industry.

For the theft of authentication credentials, the motive is not

Food & one of immediate financial gain, but information gathering for
Beverage a subsequent attack. In many cases such data, particularly from

43.6% Retail
a consumer-focused organization, can be utilized in a targeted
attack against a commercial or government organization.

33.7%

Hospitality
8% Industries
Consistent from the prior year, the food and beverage, retail
and hospitality industries accounted for about 85% of data
Financial
3.4%
Entertainment breach investigations. In these industries, the primary target
3.4%
was payment card data. While such businesses typically
Energy
2.7% represented a smaller reward for attackers in comparison
Education
Health
Care
to large banks or payment processors, they continue to be
1.1% 1.1% a target due to well-known payment system vulnerabilities
Marketing Technology and poor security practices on behalf of those responsible
.8% .8%
for the upkeep of these systems. Organized crime groups in
Professional
Services
Nonprofit Travel
Agency particular continued to focus on these industries.
.4% .4% .4%

Personal
Services
More than one-third of breached entities in food and
.2% beverage, retail, and hospitality represented franchised
businesses. Standardization of computer systems among
the franchise models is common and, in the event a security
deficiency exists within a specific system, deficiencies
will be duplicated among the entire franchise base. Cyber
criminals took full advantage of this vulnerability, targeting
specific franchised businesses and exploiting common
points of failure across franchisee properties.

20 40 60 80 100

Customer Records
(Cardholder Data, PII,
Email Addresses)
Trade Secrets 6% 89%
Electronic Protected Health Information (ePHI) 3%
Business Financial Account Numbers 1%
Authentication Credentials 1%
5 2011 Incident Response Investigations

Target Assets
Information systems involved with payment processing continue
to be the Achilles’ heel of the payment industry and represent the
easiest way for criminals to obtain payment card magnetic stripe
data en masse. Once magnetic stripe data is obtained, attackers are Investigation Basics
able to perform fraud by encoding stolen data onto legitimate or
counterfeit cards, subsequently purchasing goods and services. When a security event occurs, incident response
investigations are undertaken to identify if and what
Point-to-point encryption (P2PE) solutions, while not bulletproof,
sensitive information was extracted from the target
have the potential to lower the risk of POS system breaches. When
properly configured to protect data in transit, P2PE technology can organization. In the event that sensitive information has
dramatically reduce the currently broad attack surface of payment been exposed, Trustwave SpiderLabs performs a thorough
systems, whether data is sent between merchants and their payment analysis to quantify the specific information at risk. Various
processing banks, or via the merchant’s own internal systems.
public breach disclosure laws and compliance mandates
E-commerce targets increased from 9% to 20% over the previous typically require timely reporting of this information. To
year, largely due to additional engagements in the APAC region, meet the demands of accuracy and timeliness, we employ
where e-commerce compromises are more common than software a robust methodology called “sniper forensics” that allows
POS system compromise.
us to quickly focus on the most important aspects of an
ATMs were infrequently targeted. However, if payment investigation by understanding and following the data flows.
card magnetic stripe data and PIN are successfully obtained by
an attacker this results in direct access to cash. The most common Once an in-depth understanding of the incident is reached,
method to obtain this information is hardware tampering (i.e.,
containment and remediation plans are implemented to
keyboard overlays, cameras and skimming devices). But in a trend
consistent with our investigations over the last two years, cyber remove the threat and reduce the risk of re-occurrence.
criminals obtained this information via system intrusions and the As other prominent leaders in the industry have stated, an
subsequent installation of ATM-specific malware instead.1 understanding of the threat factors responsible for the breach
is of upmost importance, given that this intelligence can
Employee workstations and servers were the primary targets
for the theft of trade secrets and credentials. In these cases, determine the response. Involvement of law enforcement in
email with malicious intent was sent to targeted and specific these investigations often plays a critical role in augmenting
employees. This email contained an attachment, such as a PDF, our own intelligence in this respect.
an executable file or a URL. Users accessed the file or link and
malware was then deployed to their systems. Once installed, it
established an initial foothold that ultimately allowed additional
propagation within the internal network by establishing a tunnel
for the attackers for further attacks.

Employee Work Station 1%

Assets Targeted by System Type ATMs 1%
Business System 3%

Software POS E-Commerce

75% 20%

0 20 40 60 80 100

1
ATM Malware Analysis https://www.trustwave.com/downloads/spiderlabs/Trustwave-Security-Alert-ATM-Malware-Analysis-Briefing.pdf
 2011 Incident Response Investigations 6

1.5% 5.5%
58% 35%
35.8% 2011
Q3 2010 Investigations

2008 2009 2010 2011

Initial Attacker Entry

Attack Timeline
Many times compromises are detected at greatly varying intervals and the time from initial breach date to incident investigation may
be six to 12 months or more. The graph above represents investigations that took place in 2011, but demonstrates that initial entry by
the attacker could have taken place up to three years before detection and investigation.

System Administration The remaining 84% of organizations relied on information reported

to them by an external entity: regulatory, law enforcement, third

Responsibility party or public. This reliance has serious drawbacks; in those

cases in which an external entity was necessary for detection,
analysis found that attackers had an average of 173.5 days
The majority of our analysis of data breach investigations – within the victim’s environment before detection occurred.
76% – revealed that the third party responsible for system support, Conversely, organizations that relied on self-detection were
development and/or maintenance introduced the security able to identify attackers within their systems an average of 43
deficiencies exploited by attackers. Small businesses within the days after initial compromise.
food and beverage and retail industries were most often impacted
by these attacks, as they typically outsource all development and The most common method of identification was regulatory
support of their systems. Anecdotally, merchants were unaware of detection. It should be noted though, that law enforcement
the security best practices or compliance mandates by which their notifications increased almost five-fold to 33%. This increase
partners were required to abide. In other instances, victims were can be attributed to work performed by the United States Secret
unaware that this third party was only responsible for a subset of Service and Electronic Crime Task Force members. Due to the
security controls – thus still leaving these systems open to attack. efforts by these and other law enforcement agencies worldwide,
the number of our investigations that resulted from law
Self enforcement detection increased from 7% in 2010 to 33% in 2011.
24% The involvement of law enforcement can minimize the damage
inflicted upon compromised organizations. Law enforcement is
often privy to additional intelligence, which can result in victim
notification prior to actual fraud.
SYSTEM Third Party
ADMINISTRATION
RESPONSIBILITY 76% Third Party 2%
Public Detection 3%

Self-Detection
16%

Regulatory
Detection Detection
46%
The number of self-detected compromises decreased in 2011;
only 16% self-detected compared to 20% in 2010. This may Law
indicate a decline in resources for properly detecting incidents. Enforcement
33%
7 2011 Incident Response Investigations

The Breach Triad

At its most basic form, a data breach consists of three elements:
infiltration, aggregation and exfiltration.

Infiltration
Remote access solutions are still the most widely used method of Although method of entry was unknown in 19.9% of cases, many
infiltration into target networks. Organizations without dedicated possessed a common indicator of compromise (IOC), specifically
information technology (IT) staff often hire third-party vendors to weak and/or default administrative credentials.
maintain their systems and networks. These vendors use remote
access applications or a virtual private network (VPN) to access System logins require a username and a password, and often
the customer systems. When these services are left enabled, an these combinations are pitifully simple: administrator:password,
attacker can access them as easily as an approved administrator. guest:guest, and admin:admin were commonly found in
our investigations. Many third-party IT service providers
With the number of IP addresses in the world, how are attackers use standard passwords across their client base.
able to identify remote access applications open to attack? To In one 2011 case, more than 90 locations were compromised due
illustrate, picture an international airport, with many airlines and to shared authentication credentials.
planes arriving from locations around the world. Each plane
is sent to a predetermined “port” based on a variety of factors, Another IOC is often client-side attacks, which are difficult to
such as airline or arrival and departure information. A plane from detect as the date of the initial compromise may occur months
“Airline A” will always dock in the terminal designated for Airline A. before an investigation when log files needed to identify the
attack are no longer available. During a client-side attack,
Computers communicate similarly; there are 65,535 ports and attackers implant malicious code on victim systems via a file,
each is used for different types of communication. Ports used Web page or other document viewed in a client application such
by remote access applications, unless altered from their default as a Web browser or document viewer. Systems administrators
configuration, will always be the same. utilized production environments for personal use (frequently
accessing personal email accounts, social networking sites and
An attacker can scan the Internet for hosts that respond to queries even online Flash or Java-based gaming sites) in about 60%
on one of these ports. The results of the scan will produce a list of these cases, demonstrating the effectiveness of these types
of hosts (along with system information suggesting the host’s of attacks. In many cases, the breach was also extraordinarily
function) that are potential targets. Once they have a focused difficult to detect.
target list of IP addresses that have open remote access or VPN
ports, they move to the next part of the attack: weak credentials. Structured Query Language (SQL) injection continues to be a
common infiltration mechanism for a wide variety of applications,
most often for Web pages. Web pages today consist of dynamic
Sharing credentials from one location to components to improve the user experience, and many pages
ask for additional information, ranging from bank account
another potentially puts every customer numbers to geographical location to shopping preferences, to
using the same username:password improve speed and efficiency. Such pages make SQL queries to a
database where user information is stored, sending and receiving
combination in a position to be information that impacts performance and drive business
functionality to Web applications. In a SQL injection attack, the
compromised. Web pages that use this dynamic content are not doing proper
input validation.
 2011 Incident Response Investigations 8

Attackers used SQL injection to infiltrate environments 6.9% of The use of weak and/or default credentials continues to be one
the time. Attackers use SQL injection to execute code on the of the primary weaknesses exploited by attackers for internal
target systems, which often results in a compromise of the propagation. This is true for both large and small organizations,
system running the database. and largely due to poor administration. In one instance, attackers
were able to compromise as many as 250 unique critical systems
After achieving an initial point of compromise, commonly referred at a single target location by exploiting duplicate credentials.
to as a “foothold” or a “beachhead,” attackers work to identify
additional targets on the compromised network, and propagate Overall, the propagation methods most commonly used in 2011
the intrusion. were similar to those being used last year and several years prior.
Most target networks are Windows-based and use the NetBIOS
In 2011 the top three methods of propagation were: protocol for file and print sharing. Attackers need only scan
the network from the foothold for devices sharing file and print
services to identify additional targets (specifically for ports 135,

80%
Use of weak 137, 139 and 445). They can also use a technique called Address
Resolution Protocol (ARP) cache poisoning, a complicated attack
administrative
that allows an attacker to view network traffic and intercept clear
credentials text credentials and other sensitive data in real time.

Attacks such as these, however, were not needed in many of

15%
the networks investigated in 2011. Instead, systems using shared
Default hidden administrative username and password combinations, as well as
administrative shares mapped drives and open-by-default Windows hidden shares,
enabled attackers to quickly identify additional targets, gain
credentials and administrative access and then subsequently
deploy their malware. These types of attacks can propagate

5%
across an entire small network (between one and 20 devices) in
Remote access solution less than 10 minutes.
credential caching
The third most used method of propagation is remote access
caching. Many remote access programs have the option to
“cache” or remember login credentials. While convenient for
Insecure X.25 Interface .4% the end user, it is not secure; best security practices dictate that
caching be disabled.
Malicious Insider .4%
Directory Traversal .4%
Physical Access 1.1%
Authorization Flaw 2.3%
Remote File Inclusion 2.7%
Admin Interference 4.2%

SQL Injection
6.9%

Remote
Unknown Access
(Weak Credentials or
Client-side Attacks) Application
19.9% 61.7%
9 2011 Incident Response Investigations

Aggregation Hybrid 4.3% In Transit

Like 2010, attackers in 2011 were more successful at
harvesting data in transit than they were attacking
Data Redirection 5.2%
62.5%
stored data. Further, these attackers were more adept
at hiding malware (e.g., memory dumpers, keystroke
loggers and network sniffers) in plain sight, with
processes appearing as subtle variants of legitimate
process names, or as legitimate process names running Stored Data
from non-standard directories. Data exposure volumes
are difficult to track and/or estimate, primarily due to
28% DATA
the data harvesting methods used, but in cases where HARVEST
memory dumpers and/or key loggers were used, METHOD
malware lived on a target system undetected for an
average of six months before discovery.

Exfiltration
Exfiltration, the third component of the Breach Triad, is
the act of actually removing the data from the targeted
systems. For 2011, the number one method is the removal
of data via the same method in which the system was
entered. Because the majority of breaches go unnoticed
for long periods of time, attackers often do not need to
establish an alternative method of data exfiltration. Physical Access .4%
Same as
In cases where SQL injection is used as an infiltration Native Internet Entry Method
method, it can also be used as a method of exfiltration.
By this method, attackers can dump database tables
Protocols
15.3% 45%
with hundreds of thousands of customer records
containing names, addresses, phone numbers and
credit card numbers.

Attackers continue to exploit the lack of a firewall, or

firewalls without egress filters to enable data exfiltration;
88.4% of cases involved firewall deficiencies, with 78%
Built-In
Malware
of organizations lacking firewalls completely. Functionality
Of the breach investigations involving firewall 39.3%
misconfigurations, 99% of the organizations’ firewalls
did not include proper egress filtering. Egress filtering
employs rules to ensure data is being sent to the proper
location, over the proper port, using an authorized
protocol. In interviews conducted during investigations,
the pervasive rationale behind the lack of egress filters
is the belief that the internal network is “trusted” and
any traffic originating from the trusted network must
likewise be trusted. This rationale would only be accurate
if a breach were not possible. Assuming a breach is
not possible is an unrealistic view; data breaches are
affecting organizations daily and globally. Practical,
preemptive measures should be taken to ensure that,
if a compromise occurs, the attacker has to circumvent
an additional layer of technical controls to successfully
extract data from a compromised environment.
 2011 Incident Response Investigations 10

Origin of Attack

Russian Federation 29.6%

United States 10.5%

Romania 4.1%
United Kingdom 3.5%
Vietnam 3.2%

Canada .6%

Japan 1%
Netherlands 1%
Germany 1%
Spain .6%
France .6% Malaysia 1%
Hong Kong .6%
Mexico .3% Portugal .3% China .3%
Luxembourg .3%
Sweden .3%
Brazil .3% Austria .3%
Colombia .3% Ukraine 1%
Italy .3% Turkey .6% Taiwan .3%
Chile .3% Czech Republic .3%
South Korea .3%
Poland .3%
Slovakia .3% Maldives .3%
Belarus .3% Kuwait .3%
Estonia .3% Georgia .3%
Egypt .3%

* 32.5% Unknown Origin

International Perspectives
Attacker Source Geography
Based on our investigations, attacks in 2011 originated from 40 Based on our investigations and analysis of the source IP
different countries, although the largest percentage shows origin addresses, attackers are using networks of compromised
to be unknown. Source IP addresses do not necessarily establish systems to mask their actual locations. For some regions, such as
where attackers are physically located and maintaining online Asia-Pacific, the increase is likely to be a reflection of abundant,
anonymity is very easy for attackers today. Therefore, the unknown and rising, broadband coverage combined with a still-maturing
points of origin simply represent anonymous service endpoints. information security industry.

Both public anonymity services, such as Tor, and private alternatives

available for small fees exist for dedicated criminals. Even when the
point of origin is anonymous, this information can frequently assist
law enforcement. Therefore, sharing intelligence among victim
organizations, law enforcement and private security companies,
such as Trustwave, is essential in combating cyber crime.
11 2011 Incident Response Investigations

Europe, Middle East and Africa

In contrast to data compromise trends in the Americas, very 2010. The change stated that investigations only proceed when
few data compromises occurred in POS networks in Europe, a minimum of 10,000 Visa cards are suspected to be at risk, and
the Middle East and Africa (EMEA). Rather, as a result of it was often these smaller merchants who had been associated
higher adoption of “chip & pin” (EMV) and deprecation of with SQL injection-based data compromises.
magnetic stripe (mag-stripe) transactions within Europe, fewer
opportunities exist in EMEA for the theft of track data used in Visa Europe introduced the PFI2 Lite program in November 2011 to
mag-stripe transactions. establish guidelines for performing investigations for merchants
with less than 10,000 Visa cards at risk. Next year may see an
However, across the region many mag-stripe enabled POS increased number of investigations of smaller compromised
systems remain in use to support mag-stripe only cards or entities as a result.
transactions that fall back to mag-stripe when EMV fails. As
such, card-present compromises do still occur in small numbers. A pervasive problem with e-commerce compromises is highly
inadequate logging and monitoring. Small and medium-
Overwhelmingly, e-commerce merchants in EMEA were the sized e-commerce merchants typically do not have logging
targets for cyber criminals. E-commerce businesses allow configured to identify possible security events. Further
attackers to be geographically indiscriminate and concerned exacerbating investigations, merchants will sometimes erase
only with identifying targets that pose little technical complexity everything as part of the containment process, including logs,
in compromising. following a compromise.

The typical vulnerabilities exploited in EMEA investigations were Unlike previous years, investigators from Trustwave SpiderLabs
insecure, but legitimate file upload mechanisms or exploitable found no cases in EMEA where compromised resources were re-
remote file inclusion vectors. used for activities outside of data theft. In other words, attackers
did not utilize the compromised infrastructure for file-sharing,
Very few SQL injection-based data compromises were hosting illegal content, hacking tools or other activities. The
investigated over the last year in EMEA. This may in part have attackers appeared to be solely focused on obtaining data from
been due to a regulatory change introduced by Visa Europe in target systems.

Search engine to Scanner to

identify websites identify websites
with vulnerable with vulnerable
components components

Exploit upload
mechanism or
remote file inclusion
vulnerability
Card data not stored?
Modify payment page
to siphon off CHD

Browse file system to

Upload
identify other targets
multi-function
in shared hosting
webshell
environment

Card data stored?

Access backend
database containing
CHD
Typical Attack Methodology in Europe
Cycle Repeats

2
The PCI Security Standards Council’s PCI Forensic Investigator (PFI) Program
https://www.pcisecuritystandards.org/approved_companies_providers/pci_forensic_investigator.php
 2011 Incident Response Investigations 12

to the environment via the X.25 node, the attacker identified an

internal development system and proceeded to re-rewrite a well-
known rootkit to function on the HP-UX operating system. The
rootkit was then installed across a number of cardholder data
processing servers to mask the presence of other malicious
programs introduced by the attacker.

During the operation, the malicious scripts harvested cardholder

data by terminating the legitimate instances of payment-
processing software and then restarting the software with a
Trojanized-debugger attached. The debugger captured all inter-
process communications including unencrypted payment card
data from within the system memory, which was otherwise
encrypted when at rest on the disk and in transit on the network.

The attacker went unidentified within the environment for almost

18 months. Of note, the attacker was only identified when a
subtle flaw within their own customized malware alerted the
payment service provider’s operational staff to suspicious activity.

Notable Events: EMEA 2011 It is worth noting that the payment service provider’s environment
was not PCI DSS compliant. Without mandates that strictly regulate
In one of the most significant EMEA compromises payment processors, individual merchants that take steps towards
of 2011, in which a payment service provider was PCI compliance still remain at risk of compromise on third-party
systems that store and process their data. Appreciation that such
hacked, multiple servers and a wide area network
a breach necessarily affects many merchants at once highlights
with more than a thousand hosts were attacked. the risk of partnering with small hosting/service providers with
Trustwave SpiderLabs identified the single point of limited security expertise.
weakness as a legacy X.25 node. X.25 is a protocol
Finally there is continued traction toward data privacy legislation
suite which was widely used in the 1980s to build
across the European Union. Proposals have been drafted, but
Wide Area Networks. Today it remains commonly still need to be approved by national governments. This effort
utilized by financial institutions for inter-bank data signals a movement towards mandatory data breach disclosure
exchange. laws across the region, as well as potential fines for organizations
that do not adequately safeguard customer data. As such we
Unlike the relatively low-skilled e-commerce compromises, the expect to see continued growth in demand for proactive security.
attacker in this case demonstrated persistence and novelty in the
technical aspects of the compromise. Having gained initial access
13 2011 Incident Response Investigations

Asia-Pacific
In 2011, APAC investigations made up 19% of investigations Attackers are not concerned with
overall. A significant vulnerability was discovered in Australian
“integrated point of sale” products (i.e., point of sale software that the victim’s nature of business, and
communicates with payment card terminals). Attackers remotely
collected card details from these systems for use in counterfeit indiscriminately choose targets that
cards operations around the world.
offer little resistance to attack.
In APAC, as witnessed in other parts of the world, attackers
The reality is that the cost of finding vulnerable sites is close
are increasingly automating the process of finding victims and
to zero, and attackers increasingly use software that constantly
extracting valuable data. This lowers the cost of performing
searches the Internet for potential victims. As a result an attacker
attacks, which in turn lowers the minimum yield for a victim to
stands to profit from a site accepting just a handful of payment
be of interest.
cards per year. As with the EMEA e-commerce compromises
it is a volume game for the attackers; given the relatively low
Approximately 90% of APAC investigations were undertaken as a
overhead costs, a conveyor-belt-like process for finding and
result of payment card data compromises.
exploiting targets provides a satisfactory yield for the criminals.

In addition to payment card compromises, Trustwave investigated

Many of our investigations—55%—took place due to compromises
cases in APAC involving denial of service, loss of intellectual
in cardholder-present, or brick-and-mortar, environments.
property, internal fraud, computer misuse and a variety of other
Almost all of the cardholder-present cases occurred in Australia
computer-based incidents. Prior to 2011, all investigations related
and involved integrated point of sales environments.
to payment card data compromise in APAC involved e-commerce
breaches. While attackers are now migrating to POS systems,
In almost every brick and mortar case in APAC, attackers gained
e-commerce attacks are still common.
access to the victim environment via remote access software
intended for use by a legitimate third-party provider. Alarmingly,
A relatively small number of publicly disclosed vulnerabilities
many of these support vendors were using the same or similar
accounted for the majority of e-commerce compromises. These
passwords for all of their clients. Worse still, passwords were
vulnerabilities appeared in popular shopping cart software. In
often the name of the vendor that provided remote support.
most cases, patches had been released to resolve the issues,
but had not been applied. Attackers used pre-packaged toolsets
Preventing these attacks again relies on the implementation
to exploit these vulnerabilities to dump data, gain access to an
of security fundamentals. Ensuring that appropriate password
administrative interface or to upload malicious software to the
security controls are in place for internal staff and that external
Web server.
service providers are subjected to the same level of adherence
is key. Similarly, ensuring that cardholder data is truncated,
As in EMEA, remote e-commerce attacks designed to capture
tokenized or properly encrypted as soon as possible in the
payment card data in real time increased in 2011, however,
transaction flow minimizes the chance of compromise.
approximately two-thirds of e-commerce attacks continued to rely
upon stored data, indicating these merchants continue to store
Following security basics like strong passwords, secure remote
payment card data on their systems. Many of these compromised
access, least privilege and patch management would have
entities reported that a third-party was responsible for the
prevented almost all of the compromises investigated in APAC in
administration of their systems. They often did not know that
2011. In particular, organizations should ensure that their third-
payment card data was being stored, and that their service provider
party service providers leverage appropriate information security
had not been applying software patches in a timely manner.
controls when dealing with their data.
Similarly, most merchants did not believe their site was a target
for cyber attackers. Some merchants believed, wrongly, that
attackers leveraged sophisticated techniques that would be
difficult to protect against or that victims were chosen carefully
by a cost/benefit equation.
 2011 Incident Response Investigations 14

In the majority of the stored data attacks, a PIN entry device

that routinely shared payment card data with the POS device
was in use or had been used in the past. Additionally, a piece
of software used to interface the POS device with the PIN entry
device was misconfigured to log this cardholder data onto
the hard disk of the POS system. As a result, all payment card
details processed by that POS system would also be stored in log
files on the disk of the POS system. In several cases Trustwave
SpiderLabs investigated, this amounted to more than three years
of transactions.

In-transit attacks were first seen in Australian-based investigations

towards the end of 2011 and are thought to be an evolution of the
stored data attacks. The in-transit attack relies on the presence of
a PIN entry device that shares clear-text cardholder data with the
POS. Attackers then place memory-dumping malware on the POS,
and collect this data in real time as it is processed.

POS Attacks in Australia This memory dumping malware is no different from the malware
samples observed in the U.S. and EMEA. The malware succeeds
if any device transmits clear-text payment card data through the
In a common integrated point of sale (POS)
POS regardless of the version and type of software being used
environment, a PIN entry device, connected to the POS on the POS. This type of attack is not unique to Australia and
device, is used to read cards and collect PINs. The similar compromises have occurred in other countries in the
PIN entry device conducts the financial transaction APAC region.

and informs the point of sale device whether the

Most of the newly deployed PIN entry devices used by Australian
transaction was a success or not. By design, the point integrated point of sale merchants today are no longer vulnerable
of sale device should not be able to access cardholder to either of these attacks. As with e-commerce, though, the
account details. In fact, this is one of the key reasons cost of performing an attack is relatively low and attackers will
continue to have a viable business even if a small proportion of
a separate PIN entry device is used, as it reduces the
the entire integrated POS merchant base still has vulnerable PIN
risk of a compromise affecting cardholder data should entry devices.
a merchant’s system be breached.

Attackers discovered that some PIN entry devices do not properly

protect payment card data and that a compromised POS device
can, in some situations, result in access to payment card data. ACQUIRER
These compromises fell into two main categories: stored data BANK
attacks and in-transit attacks.

PIN 4
Authorization
entry device POS message
“Charge cusomer $x” 1
Authorization message -
2 important fields encrypted 3 Authorization
response

5
Informs POS if transaction
was successful or not
15 2011 Incident Response Investigations

Latin America and Caribbean

Companies in LAC have been targets for cyber criminals for New for 2011 in LAC was the use of information technology, social
many years, especially those companies in countries that networks and other methods to publicize confidential documents
have implemented online banking services. Economic growth, and recruit people with the intent to disrupt services through
particularly in places such as Brazil, has been driving more denial of service and other types of attacks. Some attackers
people and business online, opening up avenues of attack for also used denial of service attacks to distract the target while
cyber criminals. performing additional attacks to steal confidential information.

Early Days
The birth of commercial Internet in Brazil comes from the
establishment of the Brazilian Internet Steering Committee (CGI.
br) in May 31, 1995. CGI.br is a multi-stakeholder organization,
composed of members from government, private sector, non-
governmental organizations and the academic community, and
was created with the purpose of coordinating and integrating all
Internet service initiatives in Brazil, as well as promoting technical
quality, innovation and the dissemination of the available services.

One of the CGI.br attributions is promoting studies and technical

standards for network and service security in the country. The
development of incident response capabilities in Brazil originated
from discussions inside the CGI.br Security Working Committee
and culminated with the creation of the Brazilian National
Computer Emergency Response Team - CERT.br (initially called
NIC BR Security Office - NBSO) in June of 1997. Since 2006
CERT.br has been maintained by NIC.br, which is the executive
branch of the Brazilian Internet Steering Committee.

CERT.br Observations: Activities and Initiatives

Incident Response and Cyber CERT.br is responsible for handling any incidents that involve
Brazilian networks connected to the Internet, providing
coordination and support to organizations involved in incidents,
Security Efforts in Brazil establishing collaborative relationships with other entities, such
as other CSIRTs, Universities, ISPs and telecommunication
companies, and maintaining public statistics of incidents handled
Brazil, like many other countries, has observed a and spam complaints received.
significant increase in computer security incidents
and online crimes in the past few years. As criminals As a Software Engineering Institute Partner, CERT.br delivers the
develop new techniques and evolve their skills, there CERTÆ Program Incident Handling courses in Brazil, helping
is a growing need for cooperation, coordination and new Computer Security Incident Response Teams (CSIRTs) to
establish their activities and prepare their staff. Currently there
awareness to deal with the threats. With that in mind, are more than 35 CSIRT’s in Brazil.
several initiatives have been put in place in order to
raise awareness and prepare the country to manage In the awareness field, CERT.br produces videos, guidelines and
incidents accordingly. other literature targeting different audiences. For end-users there
are educational videos and an Internet Security best practices
guide, covering basic security concepts, information about virus,
worms, fraud, and vulnerabilities. For network administrators
there are guidelines with best practices on network security and
technical white papers about specific threats.
 2011 Incident Response Investigations 16

Aiming for the improvement of network monitoring and the

proactive detection of incidents in the country, CERT.br coordinates
Current Statistics and Trends
the “honeyTARG Honeynet Project,” a chapter of the Global From January to September 2011, CERT.br handled about 318,000
Honeynet Project, which uses low-interaction honeypots to gather incident notifications. This number represents a growth of 215%
information about the Internet infrastructure abuse by attackers when compared to the same period during 2010, and 123% when
and spammers. The initiative encompasses two sub-projects: these nine months are compared with the whole year of 2010.
These incidents are split in categories such as fraud, worms
The Distributed Honeypots Project- a network of distributed (which includes bots spreading), and attacks to Web servers,
honeypots hosted at partner organizations with the goal of scans, DoS, intrusions and “others.”
increasing the capacity of incident detection, event correlation and
trend analysis in the Brazilian Internet space. For the international Some trends observed since 2010 are the rise in attacks to Web
community, publicly available statistics and anonymized data is servers and fraud attempts. The Web server attacks are, for the most
donated to other National CERTs and research organizations that part, to host phishing, Trojans, malicious scripts and tools to attack
provide information about detected network security incidents to other Web servers. Regarding fraud attempts, notifications related to
affected parties. For the Brazilian community, there is a service phishing are now greater in number than Trojan notifications.
that notifies CSIRTs and network administrators about the
attacks originated from their networks, along with the relevant We have also noticed an increase in reports of scans for SIP
information for detection and recovery. service (5060/UDP - used for VoIP connections). Although scans
for SIP have been seen on the Internet for quite some time –
The SpamPots Project- comprised of 10 sensors deployed in and in the CERT.br honeypots top scanned ports for about two
nine countries to measure the abuse of network infrastructure for years – it was only by the third quarter of 2011 that it made the
sending spam. It also helps to develop better ways of identifying list of top 10 scanned ports. Further information about statistics
phishing and malware, as well as botnets abusing open proxies on incident notifications is available at http://www.cert.br/stats/.
and relays.

CERT.br is also part of the CGI.br Anti-Spam Working Group (CT-

Sources
Spam), which developed several national initiatives against spam, • About CGI.br. http://www.cgi.br/english/
including an awareness campaign for end-users, the evaluation
and proposal of anti-spam legislation and the definition of a • About CERT.br. http://www.cert.br/en/
Code of Practice for Email Marketing.
• Antispam. http://antispam.br/
However, the most significant initiative to reduce the abuse
• honeyTARG Honeynet Project. http://honeytarg.cert.br/
of the Brazilian broadband networks by spammers is the
adoption of “Port 25 Management” in all domestic broadband • Cartilha de Segurança para Internet 3.1. http://cartilha.cert.br/
networks. Because of the regulatory environment in Brazil, the
adoption of this best practice required coordination among the • Estatísticas Mantidas pelo CERT.br. http://www.cert.br/stats/
Internet Industry, regulatory authorities and consumer rights
organizations. Finally, on November 23, 2011, an agreement
defining the steps for implementation was signed by CGI.br,
NIC.br, the Brazilian National Telecommunication Agency
(ANATEL), the Associations of Telecommunication Providers and
the Associations of ISPs. The expected benefits include reducing
the abuse of Brazilian networks by spammers, including the
abuse performed by spambots.
17 2011 Incident Response Investigations

Malware Statistics
Malware comes in all shapes and sizes, and is often purposefully engagements during 2011. DLL hooking is an example of
designed to capture and exfiltrate data, provide remote access, or legitimate code techniques that can be used by malware authors
automate compromised systems into a botnet — or to just cause to perform malicious actions.
general mayhem. Historically, Trustwave SpiderLabs analyzed
malware specific to incident response investigations, yielding
interesting samples not previously publicly available. In 2011, Packers
Trustwave SpiderLabs began building a database of malware
samples, gathering samples from a SpiderLabs-maintained Packers are utility applications that can reduce the size of an
honeypot network and from underground malware repositories. executable and often include encryption or reverse engineering
protections. Packers can be used by legitimate applications to
The database is used to identify trends in malware development, reduce their memory footprint and protect intellectual property
and to see how advancements in mass-deployed malware and and trade secrets present in the application code. Malware
targeted malware influence each other. By establishing a broad authors have long used packers to obfuscate their malicious
collection, specific malware capabilities can be correlated not just binaries in order to avoid detection by anti-virus and confound
between malware variants, but also across families and categories. researchers attempting to understand their code.
The collection is based on publicly identifiable malware samples.

Common versus
Targeted Malware
Common, mass-distributed malware usually seeks to self-replicate
through security vulnerabilities. Targeted malware doesn’t self-
replicate and may not exploit common vulnerabilities. Without these
traits, it is more difficult for anti-virus software to detect targeted
malware as malicious. While anti-virus products detected at least
60% of all malware samples in our database, when we focused
only on samples found during our compromise investigations,
anti-virus detected less than 12% as malicious.

Common malware usually contains components for infection,

privilege escalation, and command and control. While these
components can be switched out, doing so requires packaging
a new variant of the malware. Trustwave SpiderLabs found
targeted malware to be much more modular, allowing for a per-
attack workflow to be established. In approximately 89% of these
database samples, malware had direct exfiltration mechanisms
built-in, sending the stolen data automatically to the attacker.

Scheduling a system-wide service is a fairly common technique

for both mass-distributed and targeted malware. Running as Common
versus
Targeted
a service allows malware to recover from removal attempts,
maintain a high level of access and read the memory of other
processes. Both common and targeted malware use this
technique, especially in the case of memory scrapers, accounting
for approximately 42% of our database of public samples.

Targeted malware is becoming more advanced; approximately

13% of our database samples used inside knowledge or an in-
depth understanding of how the target business application
worked to directly hook into the target applications. Techniques
such as DLL registration, the AppInit_DLLs registry setting3,
and DLL Hijacking4 were all observed in Trustwave SpiderLabs

3
Working with the AppInit_DLLs registry value. http://support.microsoft.com/kb/197571
4
Dynamic-link library. https://en.wikipedia.org/wiki/Dynamic-link_library#DLL_hijacking
 2011 Incident Response Investigations 18

Packers are much more common in public malware samples, versions of the affected application software were released,
appearing in more than 36% of our database samples, than in confirming the sophistication and dedication of the organizations
Trustwave SpiderLabs’ case-specific malware samples, which developing and deploying this malware.
at approximately 16%, likely due to the different needs of the
malware authors. Targeted malware, lacking self-propagation Classic, high-level languages such as C++, Delphi, Perl, and
functionality, generally flies under the radar of anti-virus .NET have remained the favorite for malware authors. Old build
software. For such malware, using an identifiable packer can dates for the compilers continue to be observed, suggesting high
actually increase the chance of detection. Malware destined for degrees of code reuse and minimal modification.
widespread distribution must work harder to disguise itself, and
its authors need to protect sensitive information, such as domains Reverse engineering of malware samples often uncovers
for command and control, in each variant to avoid detection by plagiarism from online examples or re-purposing of existing open
law enforcement and other Internet security organizations. source code. A noticeable trend in samples collected during 2011
is an increase in the use of Perl2Exe in order to embed a portable
Known packers, like UPX, are being used by more than 56% of Perl environment with the malware. Because of its ability to parse
packed common malware. Armadillo and PECompact were used large batches of language for text, Perl is attractive to malware
about 8.5% and 5.2%, respectively. For targeted samples, however, developers needing to parse through data in search of credit
Armadillo was used 34.3% of the time and UPX only 27.1%. card or other personally identifiable information.

Samples from both the database and customer engagements

contained malware packed with custom packers. While the Data Export
purpose and functionality of custom packers is essentially the
same as out-of-the-box versions, the malware methods and An emerging trend in 2010, HTTP is now the most likely protocol
resulting samples did not match any of the known packer utilities to be used for data exfiltration in 2011. In analyzed samples,
on the marketplace. 41.2% of malware used HTTP, or TCP traffic over ports 80 and
443, to exfiltrate data. HTTP and HTTPS are regularly chosen
for data exfiltration and control as Web traffic filtering is not as
Malware Types widespread as other egress filtering protections. The growth of
malware using HTTP(S) should motivate enterprises to improve
Memory-parsing malware accounted for 42.1% of investigations. filtering for this common protocol.
Keystroke loggers5 and application-specific malware tied for
RDP 5.9%
second place at 13.2% each.
Other
Application-specific malware is an emerging trend, it requires 11.7%
a detailed knowledge of the targeted platform, for instance,
in the case of POS, ATM or other bespoke business system.
Application-specific malware directly targets sensitive data in
SMTP HTTPS
41.2%
memory, storage or by tricking the application to pass the data
directly to the malware during processing. Investigations in 2011 11.8%
revealed attackers returning to upgrade their malware as new
Serial Sniffer 2.6%
Memory
Rootkit 2.6% Parser
Network Sniffer 7.9%
CC Data Interceptor 7.9%
42.1% 29.4%
FTP

File Transfer Protocol (FTP), historically a favorite exfiltration

method, was utilized by only 29.4% of malware. And only 11.8%
used Simple Mail Transfer Protocol (SMTP, the standard email
Remote Access protocol) to export data.
Trojan
10.5%
Malware samples that did not include any type of direct exfiltration,
requiring an attacker to return to compromised hosts to recover
captured data, was also observed in 2011. Some attackers may
Keystroke Logger be moving away from automation, which can indicate a pattern
13.2% Application Specific
of activity and trigger alerts, to increase the duration between
compromise and detection. By staying “quiet” in an environment,
13.2% the attacker will likely have more time to achieve their objectives.

5
A keystroke logger intercepts data as it is being entered at a computer terminal via the keyboard, touch screen or external data
entry device (e.g., card reader).
19 2011 Incident Response Investigations

United States
Secret Service:
Protecting the
Nation’s Leaders
and Financial
Infrastructure
Hugh Dunleavy
Special Agent in Charge, Criminal Investigative Division

In the spring of 2010, undercover agents of the United States McCulloch, discussed the creation of the Secret Service to combat
Secret Service New York Field Office discovered some postings the counterfeiting of U.S. currency. At the time, nearly one-third
on an Internet forum from a member using the online nickname to one-half of all U.S. currency in circulation was counterfeit,
“f1ex.” In these messages, “f1ex” proudly boasted of his ability to which threatened to destroy an already fragile wartime economy.
compromise the networks of financial institutions and discussed Ironically, that evening after meeting with McCulloch, Abraham
his global network for the distribution of stolen financial data. In Lincoln was shot at Ford’s Theatre and died the next morning.
the early stages of the investigation, these agents, assigned to Today, the Secret Service has a dual mission: to safeguard the
the New York Electronic Crimes Task Force, learned that “f1ex” nation’s financial infrastructure and to protect national leaders.
had been a fixture in the criminal underground since 2003, with
associations to cyber criminal organizations such as Shadowcrew, Over the years, the Secret Service has maintained a long history
dismantled by the U.S. Secret Service in 2004. Agents classified of protecting American consumers, industries and financial
“f1ex” as an overseas hacker involved in selling illegally obtained institutions from fraud. With the evolution of payment systems
credit card account numbers through online forums and various and modernization of commerce, the Secret Service has also
other means. evolved to ensure the protection of the economy. The passage
of new legislation in the 1980s gave the Secret Service authority
Why is the Secret Service, an agency renowned for protecting the for investigating credit card and debit card fraud and parallel
President of the United States, investigating an Internet hacker? authority with other federal law enforcement agencies in
The answer goes back to April 14, 1865, and the creation of the identity theft cases. In 1984, Congress gave the Secret Service
U.S. Secret Service. As the nation’s Civil War neared its end, concurrent jurisdiction to investigate financial crimes as they
President Abraham Lincoln and his Treasury Secretary, Hugh relate to computers.
 2011 Incident Response Investigations 20

The Secret Service has long recognized that partnerships and of stolen credit cards. During the meeting, Poo was taken into
cooperation act as force multipliers in conducting investigative custody. A subsequent analysis of Poo’s laptop computer revealed
and protection operations. In 2001, Congress recognized the more than 100 GB of data, including approximately 413,000
value of the Secret Service Electronic Crimes Task Force (ECTF) credit card account numbers with an estimated value of $206
model established in the New York Field Office, where law- million. This analysis also revealed evidence of multiple network
enforcement, the private sector and academia collaborated in intrusions into government and banking sector systems.
detecting and suppressing computer-based crime. Through
2001’s USA PATRIOT Act, Congress directed the Secret Service On April 13, 2011, in the Eastern District of New York, Lin Mun
to establish a network of ECTFs to combat the rise of cybercrime. Poo plead guilty to violating Title 18, United States Code, Section
Currently there are 31 ECTFs: 29 domestic task forces and two 1029 (Access Device Fraud). On November 4, 2011, Poo was
located overseas in London and Rome. These ECTFs and their sentenced to serve 10 years in a federal prison.
associated partnerships allow the Secret Service to employ
both proactive and responsive investigative tactics centered on The investigative mission of the Secret Service has evolved to
exploiting vulnerabilities identified in the cybercrime underworld. keep pace with the information revolution and rapid globalization
Agents and ECTF partners have achieved success investigating of commerce. The combination of advanced technology with the
financial and cybercrimes that range from bank and wire fraud worldwide Internet has created the venue for transnational cyber
to network intrusions, from botnets to credit card offenses and criminals to operate with nearly complete anonymity. The Secret
many cybercrimes in between. This explains why Secret Service Service and their law enforcement partners are committed to
undercover agents were looking into cybercrimes when they disrupting and dismantling these criminal networks. The arrest
identified “f1ex” as an investigative target of interest. and successful prosecution of Lin Mun Poo is just one instance
that demonstrates the proactive approach and cooperation that
Through the spring and into the summer of 2010, undercover exemplifies the collaborative efforts of the Secret Service’s ECTFs.
Secret Service ECTF agents monitored and engaged “f1ex” The Secret Service will aggressively continue its mission to
attempting to identify the hacker who now had been traced safeguard U.S. financial infrastructure and payment systems and
back to Malaysia. As this investigation progressed, Secret preserve the integrity of the U.S. economy. The Secret Service is
Service agents learned that “f1ex” was planning on traveling to proud to partner with law enforcement, the private sector and
the United States. Agents arranged to meet “f1ex” in New York academia to accomplish this mission.
City to purchase stolen credit card account numbers. During the
course of the investigation, agents identified “f1ex” was Lin Mun Please visit the Secret Service website at
Poo, a Malaysian citizen. On October 21, 2010, an undercover http://www.secretservice.gov for more details and a complete list
agent met with Poo at a Queens, New York, diner and purchased of resources.
$1,000 worth of compromised credit card numbers. The New York
ECTF later identified the account numbers were issued from a
bank in Nepal.

In a second meeting with undercover agents, arrangements

were discussed for a continued long term relationship for the
distribution of compromised data, further illustrating Poo’s
access to stolen data. Agents set up in a hotel room in Brooklyn,
New York waited for the deal to be finalized. Poo arrived with
other associates and negotiations began to purchase thousands
21 Security Weaknesses under the Microscope

Email
Mobile

The Web

Network

Common Uses Security

Weaknesses
under the
The Digital
Microcosm
Microscope
Firewalls Businesses are continually looking to protect
their assets, from employees to customer
records to intellectual property and beyond.

Passwords This section reviews client trends to identify

four fundamental resources that are vital to
business operations. We also identified four
defenses that may require significant budget,
whether as capital expenditures or operating
costs, due to its iterative process.
Common Defenses
Time and time again Trustwave clients ask:
“Are the attackers getting better or are we

Encryption getting weaker?” The answer isn’t simple, but

the following sections aim to put weaknesses
Anti-Virus under the microscope to find ways to solve
security problems.
 Security Weaknesses under the Microscope 22

In the Workplace: Device/Service Issues

This category refers to a well-documented yet still pervasive

Four Vulnerable Resources issue of devices and services configured with default and blank
passwords, or weak and easily guessable passwords such as
“password.” Device and service examples include:
Every single day, employees access networks, send and receive
email, access the Web, and use mobile devices. Some employees
also manage such services for their companies. A cyber criminal • Routers, network switches, firewalls and security devices
sees the workplace as an opportunity, and they use these same with blank, weak, or default passwords
services, in part or combined, to execute a targeted attack. • Database services such as Oracle or Microsoft SQL
administrative accounts
Trustwave SpiderLabs performed more than 2,000 tests on • Web application framework administrative accounts
targeted attack vectors in 2011. While tests were conducted on
• Administrative interfaces for VoIP and other PBX/telcom
areas of physical, social, wireless and devices like ATMs and
systems
kiosks, this section will analyze the four most vulnerable: network,
email, the Web and mobile devices. The security community
The impact of this category varies by device type and, with certain
continues to focus on new attack vectors, while older threats are
devices such as routers or databases, there is often an easy path
often overlooked, ineffectual security controls are implemented,
for an attacker to escalate their privileges or access data directly.
and problems that have existed for years persist.

Workstation / Remote Access Issues

The Network – Legacy Issues Blank or easily guessable local system accounts for end-user
workstations or workstations with ad-hoc services such as
Still At Large VNC, PCAnywhere, or other remote access software can be a
weak point for many organizations. Like the previous issue, the
impact of this category varies by device type and content, and
Issues that have been pervasive for years include password
vulnerabilities here can allow for an escalation of privileges,
security, legacy devices, protocols and attacks, and ineffectual
especially if 1) the system in question stores cached domain
security controls, continue to affect the security of networks.
credentials, 2) there is password reuse between local and domain
accounts, or 3) the same password is used for local accounts
Network Authentication across multiple systems.

One of the most pervasive vulnerabilities of 2011 is network Network/Transmission Issues

authentication. This vulnerability generally fits into one of four Authentication credentials transmitted over the network in clear
broad categories: text or weak or legacy authentication schemes are another issue
of which to be aware. These vulnerabilities can be exploited
Network/Domain Issues by passive or active man-in-the-middle techniques to harvest
This category generally refers to issues within a Microsoft Active passwords as they are transmitted over the network, either
Directory (AD) domain, Netware Domain, or any other centralized directly or by gathering data that can easily be cracked (such as
network file or print-sharing authentication. Vulnerabilities may the legacy Microsoft LM Half Challenge6 ).
be lack of password policy enacted at the domain or, more often,
exceptions to domain password policy, such as weak passwords
for service accounts. Others include temporary administrative
accounts that are never revoked or administrators exempting
Legacy Attacks
their accounts from policy enforcement. This allows an attacker
An abundance of networks and systems were still found
or a malicious insider, once they gain entry to the network
vulnerable to legacy attack vectors; many of these vectors date
environment, the ability to access moderately privileged accounts.
back 10 years or more. Organizations are implementing new
This can often lead to a compromise of the entire domain. Since
technology without decommissioning older, flawed infrastructure.
domain authentication is used as a central authority for many
Attack vectors found include:
different purposes, accessing sensitive data becomes a trivial
exercise when an attacker can operate as a domain administrator.
Layer 2
Attacks that allow for passive and active man in the middle, such
as ARP spoofing / ARP cache poisoning and other vectors at the
lower layers, remain high impact for many organizations, allowing
everything from credential and session theft, to direct data theft.

6
How to disable LM authentication on Windows NT. http://support.microsoft.com/kb/147706
23 Security Weaknesses under the Microscope

Unencrypted Protocols
Protocols that transmit sensitive information in the clear
Vulnerability Scan Statistics
remain an issue for many organizations even though more
The next section analyzes more than two million scan results
secure replacements exist. Such protocols are widely known
from 2011.7
to be vulnerable to passive and active attacks from simple
eavesdropping to session theft.
Default Credentials
Many applications and devices are shipped or installed with
Legacy Protocols default usernames and passwords, often with full access rights.
Almost unbelievably, protocols such as Unix “r” services are still
These default passwords are frequently not changed, which can
found in abundance in many environments. Documentation of
allow an attacker to use them to gain access.8 Leaving default
authentication bypass and other attack vectors for these protocols
passwords unchanged is particularly dangerous for applications
have existed for years. They are often overlooked, however, as
accessible from the Internet.
the systems were implemented before the risks associated with
these protocols were widely known. Organizations running these
systems work on maintaining functionality, but never assess the

28%
system security.
of Apache Tomcat installations
Misconfigured Network Access Rules with an accessible administrative
Network access control devices such as packet filtering routers interface have default credentials
and firewalls are often implemented and configured incorrectly.
Organizations are not only implementing the wrong type of device
as a cost savings (opening themselves up to straightforward

10%
denial of service attacks) they also often implement these devices of JBoss installations with an
without using best practices that have been established for 15 or accessible administrative interface
more years. Pervasive issues such as access control rules that have default credentials
essentially render the device useless were common, as well as
things like the non-implementation of egress filtering, which can
allow for virus or worm propagation, and provide an attacker with

9%
an easy method of creating an exfiltration channel.
of phpMyAdmin installations have
default credentials, and a further 2%
do not require authentication at all
Paper Tigers
Organizations frequently implemented security controls with little

2%
or no efficacy against the threat it was intended to mitigate. The
generic term for this is a “paper tiger,” or “security theater” to use of Cisco devices with an
a term coined by security strategist Bruce Schneier. accessible administrative interface
have default credentials
Many paper tigers were found in 2011; one example was the use
of host-based firewalls in place of actual network segmentation.
Many organizations architect large flat networks. While not good For many common applications and devices, Trustwave
network architecture, it was implemented at one point, likely TrustKeeper® vulnerability scans show which are left with default
because it was simple and inexpensive at the time and today credentials. These include applications that could allow an
re-architecting would be a large undertaking. Organizations attacker to compromise other applications or servers, or gain
addressed segmentation by simply adding host-based firewalls direct access to sensitive data stored on internal databases.
to their otherwise flat network rather than undergoing a re- phpMyAdmin, in particular, has been linked with several notable
architecting exercise. This solution does not provide the same breaches, including the 2011 breach of Dutch certificate
level of security as proper segmentation and, for a malicious authority Gemnet, in which the attackers gained access through
insider, it is barely a speed bump for layer 2 and man-in-the- a phpMyAdmin server that did not require authentication.
middle attacks.

7
Delivered through the Trustwave TrustKeeper® platform, Trustwave’s vulnerability scanning service scanned more than 2,000,000 customers in 2011. These customers elect to have network and
application vulnerability scans perform at various intervals throughout the year. Trustwave SpiderLabs developed the proprietary scanning technology and maintains the vulnerability signatures for
TrustKeeper by providing weekly (or more frequent, if critical) updates to our cloud-based scanning engines.
8
Default credentials to nearly every commercial product can be found online easily. For example, http://cirt.net/passwords contains a database of more than 450 vendors representing nearly 2000 passwords.
 Security Weaknesses under the Microscope 24

Unencrypted Data Transfers

Although mainstream encrypted protocols for transferring Web Exposing these services provides attackers an avenue of access
pages, email, and other files and data have existed for more than to sensitive information, allowing them to directly attack a
a decade, their insecure predecessors continue to predominate. database server, which may have default passwords. Or they
While legitimate applications may exist for the use of unencrypted may be able to uncover missing security updates, rather than
protocols across the Internet (e.g., websites with no sensitive discovering flaws in a Web application, allowing an indirect
content or functionality), in many cases the insecure protocols attack against the server.
are used to transfer sensitive data. More than a quarter of all
HTTP services scanned by TrustKeeper had login pages that Insecure Remote Access
transmitted credentials unencrypted. Despite the wide availability of secure VPN solutions, 22%
of organizations continue to use insecure remote access
42.54% applications. Without robust authentication and data encryption,
Insecure Protocols these applications do not provide adequate security for remote
HTTP w/ Insecure Logins access, potentially exposing usernames, passwords and other
7.6% FTP sensitive data. Additionally, the applications provide direct access
4.43%
HTTP POP3 to a computer or device, giving attackers more areas to attack,
22.14% 4.31%
increasing the risk of compromise for those hosts.
IMAP
4.31%
Other 0.1%

22%
Insecure Remote
Access Solutions

VNC 2%
PCanywhere 3%
57.11%
Telnet 3%
Secure Protocols
Remote Desktop14%

Overly Permissive Network Access

of all organizations scanned by

10%
TrustKeeper allowed connections
from the Internet to internal
database servers; 85% of these
were MySQL database servers

of all organizations scanned by

3%
TrustKeeper had results suggesting
that one or more of their systems
were essentially not protected by
a firewall
78%
Secure Remote
Access Solution
TrustKeeper scans reveal that a significant number of organizations or No Remote
do not adequately protect network services that should not be Access Used
exposed to the Internet, such as database servers and Windows
networking services. Whether due to misguided policies, firewall
misconfiguration or lack of firewalls in the first place, these services
end up accessible to the Internet. Database servers, particularly
MySQL, are the most frequent victims, and a significant number of
these appear to come from shared hosting providers.
25 Security Weaknesses under the Microscope

Top 10 Network Risks

Below is a top ten list of the issues found during the more than
6 Continued Use of Wired Equivalent
Privacy (WEP) Encryption
2,000 penetration tests conducted in 2011.10 WEP is a protocol for encrypting
transmissions over IEE802.11 wireless
networks. Packets are encrypted using

1
the stream cipher RC4 under a root key CVSSv2 Score
Weak or Blank Password for an
Administrative System Account shared by all radio stations. Security 8
analyses of WEP show that it is inherently
Windows or Unix Systems may have an CVSSv2 Score flawed; an exploit tool exists for almost
easily guessed or null password for an every step in the encryption process.
6.7
administrative level account.

7 Client Sends LAN Manager (LM)

2 Sensitive Information Transmitted Response for NTLM Authentication

Unencrypted on the Wire Any number of mechanisms can “trick”
a client into attempting to authenticate
Sensitive information such as CHD, PII or CVSSv2 Score
to a malicious server/service (e.g., MITM,
SSN is not encrypted while traversing in-
6.7 DNS or DHCP attacks, embedded links
ternal networks.
in Web pages) making this vector easy to
implement. If a user is an administrator

3 MS-SQL Server with Weak or No of his or her own system (very common),
compromise of the host is easier to
Credentials for Administrative Account CVSSv2 Score
accomplish and an attacker will have
Microsoft (MS) SQL server may have access to the local system, domain or 4.7
an easily guessed or null password for CVSSv2 Score domain administrator credentials. By
administrative accounts such as the 4.7 implementing a server with a known
system administrator account. NTLM 8-byte challenge, it is possible to
perform cryptographic attacks against
a captured LM client hash using a

4 Address Resolution Protocol (ARP)

Cache Poisoning
combination of pre-computed hash tables
(rainbow tables) and brute force to reveal
the plaintext password.
ARP cache poisoning, or ARP spoofing,
is an OSI Layer 2 attack. A gratuitous
ARP message is sent to one or more
machines on the subnet stating that the
MAC address has changed; the message
CVSSv2 Score 8 Misconfigured Firewall Rules Permit
Access to Internal Resources
usually contains the attacker’s MAC as a 10 Depending on the complexity of the
substitute. When the attacker turns on IP firewall access control list, mistakes can CVSSv2 Score
forwarding, sent packets will be routed cause data to be forwarded to hosts 4.7
through the attacker’s machine. inside the network.

5 Wireless Clients Probe for ESSID’s from

Stored Profiles When Not Connected 9 Storage of Sensitive Information Outside
the Designated Secured Zone
A Karma attack occurs when an attacker Sensitive information is stored in CVSSv2 Score
starts up a bogus wireless AP that will unencrypted files on local workstations or
allow association and access for any client 3.3
network file shares.
probe from a stored profile. In this way the
CVSSv2 Score
client connects to the Karma AP instead
of the intended AP. If the attacker’s AP has
Internet connectivity and is configured to
route traffic, the victim can perform tasks
4.7
10 Sensitive Information Transmitted
Over Bluetooth
normally but not know they are connected 2011 has seen developments in tools that
to an attacker. can be used to sniff sensitive information CVSSv2 Score
if it is transmitted over Bluetooth. Because
4.5
of this an eavesdropping attacker can
sniff this information.

10
Note: For each risk, we documented the Common Vulnerability Scorning System Version 2.0 (CVSSv2) score as documented and maintained by the
Forum of Incident Response and Security Teams (FIRST) in which Trustwave SpiderLabs is a member.
For more information, visit http://www.first.org/cvss/cvss-guide.html.
 Security Weaknesses under the Microscope 26

New threats and emerging attack vectors continue to receive

the most attention from security departments and budgets,

2008
2009
2010
2011
while older problems remain unaddressed. Many issues found
in network penetration tests and vulnerability scans are well-
known, some more than 10 years old, and others date back
to the very beginning of shared and networked computing.
These vulnerabilities are actively exploited by attackers and
often represent the path of least resistance. Older, widely
known vulnerabilities make exploitation simpler and the tools
to exploit these issues are mature, several revisions deep, and
even feature rich.

As the saying goes, those that fail to learn from history are
doomed to repeat it. Organizations must look at these old issues
and address them.

What’s in Our Inbox?

11
2011 Email Trends
Spam and junk mail peaked in 2008; the percentage of “good”
emails has slowly increased each year since. Although spam and
junk mail have declined, mail containing viruses and malicious
executables have nearly doubled year over year (although
they still represent less than 1% of all email being processed).
Attackers were more likely to send malicious emails during the
early hours of the morning, peaking between 8 a.m. and 9 a.m.
Eastern Standard Time, slowly tapering off over the course of the
day. Similar to the trend observed in 2010, attackers are moving
away from mass quantity email (although it still exists). Rather,
attackers are becoming more focused on targeting users with
emails containing malware and malicious Web links.

Email Statistics
Trustwave mailMAX processes more than four billion email
messages every year. To make sure that every message sent is
clean before entering or leaving the network the messages are
thoroughly scanned for viruses, spam and unauthorized content.
Spam messages and other unsolicited emails waste time and
cost businesses money.

The percentage of email messages processed as spam dropped

noticeably from the fifty-percentile range (where it had maintained
the previous three years) to 36.7% during 2011. Part of the drop-
off of messages categorized as spam may be due to the improved
use of additional real-time blacklists (RBLs), implemented in late
2010. RBLs block known bad email servers by default, limiting
51.2% 52.7% 52.9% 36.7%
the need for further computationally expensive analysis.
Trustwave saw a 7% increase in email stopped by RBLs in 2011
ANNUAL SPAM TOTALS
compared to the previous year. Further aiding the decrease of
spam, a number of large botnets were taken down during the A review of the percentages of spam messages received for
year, including Rustock, believed to be responsible for sending all of 2011 and comparing them to the three previous years.
40% of all Internet spam.12

11
Data in this section is based on Trustwave mailMAX. mailMAX is a solution for spam filtering, email encryption and archiving. Between
the years 2008 and 2011 the system processed and analyzed more than 16 billion emails for our clients.
12
“Rustock botnet responsible for 40 percent of spam.”
http://www.goodgearguide.com.au/article/358165/rustock_botnet_responsible_40_percent_spam/
27 Security Weaknesses under the Microscope

Spam Subject Lines

The majority of all spam identified–83%–consists of two categories:
pharmaceutical pills and pornography. The remaining categories
cover a range of topics, such as imitation wristwatch spam (4%).
While Nigerian advance-fee fraud scams are often talked about,
they represented less than 1% of all spam processed in 2011.

Pharmaceutical Pills 54%

Pornography 29% Misc. 7%

Spam Subject Lines

Fake Watch 4%
Dating 1%
Learn Languages 1%
Categorization of Spam Subject Lines Loan 1%
Credit 1%
Phish/Virus 1%
Insurance 1%
Work at Home <1%
Nigerian Scams <1%

Dangerous Files
Interception of executable files via email has almost doubled every Known viruses detected in emails dropped to three quarters of a
year since 2008. Blocking dangerous files such as executables percentage point in 2011 after three years of steadily increasing. This
sent through email helps restrict the spread of malicious worms decline further attests to the fact attackers are moving away from
and Trojans that may be attempting to spread to new hosts. broad-based attacks and becoming more targeted in their approach.

0.008% 1.5%

1.2%
0.006%

0.9%

0.004%

0.6%

0.002%
0.3%

0.000% 0.0

2008 2009 2010 2011 2008 2009 2010 2011

Percentage of Percentage of
Executable Attachments Viruses Detected
 Security Weaknesses under the Microscope 28

Temporal Analysis
Analyzing the maximum peak spikes by time-of-day, month and year,
some interesting conclusions based on the results can be drawn. 7000

The number of executables and viruses sent in the early morning 6000
hours increased, eventually hitting a maximum between 8 a.m.
and 9 a.m. Eastern Standard Time before tapering off throughout 5000
the rest of the day. The spike is likely an attempt to catch people
as they check emails at the beginning of the day. 4000

Executables and viruses accounted for almost 3% of all 3000

electronic mail in August and September 2011. Based on this
data, an individual was most likely to be emailed a virus between 2000
8am and 9am during the month of September 2011. The time
from compromise to detection in most environments is about 1000
six months; therefore, if these methods were successful, March Executables by Hour
2012 should be a busy month for incident responders and breach

t
12 m
m
m
m
m
m
m
m
m
m
am

m
m
m
m
m
m
m
m
m
pm
m
pm

gh
disclosures.

1a
2a
3a
4a
5a
6a
7a
8a
9a

1p
2p
3p
4p
5p
6p
7p
8p
9p

p
10
11

ni
10
11
id
M
Email remains one of the most prevalent and pervasive targets
for cyber attack today. According to a 2011 Pew Internet
600000
Survey13, using email is tied at 92% with using search engines
as the most popular activities performed by online adults. Since
many attackers tend to be opportunistic, the popularity of email 500000

and its ability for dynamic action makes it a natural choice as an

attack vector. Attackers go where the victims are, and more users 400000
on a particular platform translate into more potential victims for
the attacker.
300000

In a year of headlines constantly warning of new cyber threats

it is important to remember the basics. A healthy amount of 200000

skepticism and adherence to security best practices, such as

not opening unexpected attachments, will help prevent the 100000
initial vulnerability attackers look for. Awareness, education
and communication should be the security mantra as attacks
Viruses by Hour
continue to evolve and test our defenses.
pm

t
m
m
m
m
m
m
m
m
m
am

12 m

m
m
m
m
m
m
m
m
m
pm
m
gh
1a
2a
3a
4a
5a
6a
7a
8a
9a

1p
2p
3p
4p
5p
6p
7p
8p
9p

p
10
11

10
11

ni
id
M
3.0%

2.5%

2.0%

1.5%

1.0%

0.5%

Viruses by Month
y

ch

ril

ay

ne

ly

st

r
be

be

be

be
ar

ar

Ju

gu
Ap

M
ar

Ju
nu

ru

em

to

em

m
Au
M
b

ce
Oc
Ja

pt

ov
Fe

De
Se

13
“Search and email still top the list of most popular online activities.”
http://www.pewinternet.org/Reports/2011/Search-and-email/Report.aspx
29 Security Weaknesses under the Microscope

The Web – Multi-Vector Analysis Session Hijacking 2%

Leakage of
of Modern Attack Techniques
Link Spam 3%
Information
What motivates attackers to hack Web applications? What
methods are used? What vulnerabilities are exploited?
Account Takeover 4%
34%
Disinformation
Organizations are struggling to find answers to these critical
7%
questions. Numerous community security projects exist to
track Web application vulnerabilities, such as CVE and Bugtraq,
Monetary
however, they only provide data for one dimension of the Loss
standard risk equation: 7%
TOP WHID
RISK = THREAT x VULNERABILITY x IMPACT OUTCOMES

Real-world, Web application breaches, on the other hand, provide

additional information, such as exploit likelihood, to enable
Planting of
Malware
9%
2011
research into actual cyber threat trends. This information helps
to identify the types of organizations attacked, the motivation
behind the attacks and the sources of the attacks. The Web
Hacking Incident Database (WHID)14 is a project dedicated to
Defacement
maintaining a list of publicly disclosed Web application-related 10%
security incidents. The WHID first serves as a tool for raising
awareness of Web application security problems, and second, Downtime
aids risk-rating methodology processes by providing statistics
of real-world Web application security incidents. Unlike other 24%
resources covering website security, which focus on the
technical aspect of the incident, the WHID focuses on the impact
of the attack. To be included in the WHID, an incident must be Monetary loss, occurring in 7% of incidents, is largely the result
publicly reported, be associated with Web application security of criminals utilizing various methods of fraudulently transferring
vulnerabilities and have an identified outcome. funds out of victim’s online bank accounts. They leverage client-
side banking Trojans (such as Zeus and SpyEye), which monitor a
user’s Web activity and, when a user interacts with online banking
WHID Statistics for 2011 sites, it either steals their login credentials or alters the transfer
to request data.
The criteria for the WHID are restrictive by definition, and
the number of incidents that are included is not very large — Planting of malware results in a related outcome: by adding
approximately 300 incidents were included in the database for malicious code to the attacked websites the attackers convert
2011. This is a sample of the overall Web application compromises hacked websites into a primary method of using client-
that occurred but are not publicly disclosed and/or reported on side attacks to further the propagation of malware, such as
by media outlets. Therefore, the analysis in this document is banking Trojans.
based on relative percentage rather than absolute numbers.15
Ideological Hacking
There are two main motivations driving the bulk of Web Hacktivists are ideologists who use the Internet to convey their
application attacks that we see today: hacking for profit and message. Their goals are most often: downtime (24%) and
ideological hacking. defacement (10%).

Hacking for Profit Similar to real-world civil disobedience demonstrations such as

Professional criminals are increasingly developing new ways to “Occupy Wall Street,” online hacktivist groups aim to bring down
generate revenue from compromising Web applications. The websites in order to disrupt normal business operations. While
top outcome in 2011, leakage of information, is largely due to any amount of downtime for a website is undesirable, there are
criminals extracting sensitive customer data from e-commerce often critical time windows where being offline can cause major
websites. This data can then be sold on the black-market for damage. Sample scenarios of critical timeframes include: fund
identify theft and fraud purposes. raising efforts and seasonal shopping, such as Cyber Monday.

14
Trustwave SpiderLabs is the WHID project sponsor. For further information about the WHID, refer to http://projects. webappsec.org/Web-Hacking- Incident-Database
For a list of all active projects, visit Trustwave’s website at https://www.trustwave.com/spiderLabs-projects.php.
15
The WHID should not be seen an exhaustive source of data to demonstrate real-world threats, but it does provide evidence that Web application
attacks happen frequently.
 Security Weaknesses under the Microscope 30

Besides downtime, another hacktivist goal is website defacement. Attack Method Analysis
Web defacements are a serious problem and a critical barometer The top attack category is “unreported.” This means that 37% of
for estimating exploitable vulnerabilities in websites. Defacement the incidents reported did not specify a specific attack method,
statistics are valuable as they are one of the few incidents that are likely attributed to:
publicly facing and thus cannot easily be swept under the rug.
Insufficient Logging
Traditionally, defacements are labeled as a low severity issue Organizations may not have properly configured their Web
as the focus is on the impact or outcome of these attacks (the application infrastructure in a way to provide adequate monitoring
defacement) rather than the fact that the Web applications and logging mechanisms. If proper monitoring mechanisms are
are vulnerable to this level of exploitation. What should not not in place, attacks and successful compromises may go by
be overlooked, however, is that the threat and vulnerability unnoticed for extended periods of time. The longer the intrusion
components of the equation still exist. What happens if the lasts, the more severe the aftermath. Visibility into HTTP traffic is
defacers decided to not simply alter some homepage content one of the major reasons why organizations often deploy a Web
and instead placed malicious content within the site? Web application firewall.
defacement attacks should not be underestimated.

Public Disclosure Resistance

The majority of Web defacements were of a political nature, targeting
Most organizations are reluctant to publicly disclose the details
political parties, candidates and government departments, often
of the compromise for fear of public perception and possible
with a very specific message related to a campaign.
impact to customer confidence or competitive advantage.

Clickjacking 1%
Stolen Credentials 2%
Predictable Resource Location 2% Unreported
Cross-Site Request Forgery 2%
Brute Force 3% 37%
Banking Trojan 3%

Denial of Service
23%

SQL Injection
27%
31 Security Weaknesses under the Microscope

In many cases this lack of disclosure, apart from skewing

Top Attack Method per Vertical statistics, prevents the fixing of the root cause of the problem.
This is most noticeable in malware-planting incidents, in which
the focus of the remediation process is removing the malware
from the site rather than fixing the vulnerabilities that enabled
Government attackers to gain access in the first place.
Denial of Service 41%
For the other top known attack methods, they correspond to the
outcomes covered previously. SQL injection is number one and
it most often results in leakage of information outcomes. Denial
of service, at number two for known attack methods, results in
Entertainment
downtime for the target websites. Specifically, application-layer
SQL Injection 43% denial of service attacks is a huge concern for two main reasons:

Bypass Network Security

There are many methods for rendering a Web application
www
Web 2.0 inaccessible rather than network bandwidth saturation. Web
Cross-Site Request Forgery 14% applications are relatively fragile and attackers are able to
send precise requests, which target Web application resources
that require large processing power, and thus may more
easily consume the site’s available resources. These types of
Finance application layer attacks are not normally flagged by networking
Banking Trojan 36% infrastructure security devices.

Often Excluded From Application Penetration Testing

Due to restricted rules of engagement, most organizations do
Retail not actively test application layer denial of service attacks when
$ SQL Injection 27% conducting penetration tests. This is the main reason why this
attack type is not listed in the Top 10 Web Application Risks
found on page 32.

Technology Attack Method per Vertical Market Analysis

A few interesting conclusion can be drawn from attack methods
SQL Injection 37% by vertical. First, attack methods may be cross vertical — both
SQL injection and denial of service attacks are vertical market
agnostic. They may essentially be used against any website
regardless of what type of market it is in.
Hosting Providers
Cross-Site Request Forgery 14% However, some attacks are used more depending on the vertical
market. For example, banking Trojan software is very specific and
targeted at not only the banking industry but also to work against
specific banking websites themselves. Additionally, cross-site
Media request forgery (CSRF) attacks can theoretically be used on
SQL Injection 17% any Web application, although they are most commonly used by
attackers on Web 2.0 social media websites such as Facebook
and Twitter.

The takeaway for organizations is that this data should be

Education
correlated in a threat modeling process to ensure that proper
SQL Injection 40% prioritization is applied to these attack vectors.

Politics
Denial of Service 78%
 Security Weaknesses under the Microscope 32

Top 10 Web Application Risks

The vulnerabilities and attacks listed below are ranked by 3. Cross-Site Scripting (XSS)
collective risk, based on the frequency of vulnerability findings,
Risk Ranking Analysis:
difficulty in launching the attack, exploit likelihood and the
While XSS flaws are the prevalent finding within Web
potential impact when exploited by criminals. For example,
applications, the resulting risk level is lower than SQL injection
while SQL injection flaws are not the most common vulnerability
as attackers are not leveraging them as much in profit-driven
encountered during application assessments, it is the number
attack scenarios.
one attack vector found in both the Web Hacking Incident
Database and the number one Web-based method of entry in Application Weakness Reference:
incident response investigations. Combined with the potential CWE-79: Improper input neutralization during Web page
impact of bulk extraction of sensitive data makes SQL injection generation
the number one Web application risk of 2011. Conversely, CSRF http://cwe.mitre.org/data/definitions/79.html
is one of the most common application vulnerabilities found in
application assessments, but requires a more complicated attack Attack Method Reference:
scheme, relegating it to eighth on the list. CAPEC-63: Simple Script Injection
http://capec.mitre.org/data/definitions/63.html

1. SQL Injection WHID Incidents:

http://www.google.com/fusiontables/DataSource?snapid=S329845bfBR

Risk Ranking Analysis:

SQL Injection is the number one risk for Web applications in all 4. Authorization Bypass
three of our data sources: internal application assessments,
internal incident response/forensic investigations Web-based Risk Ranking Analysis:
method of entry and the WHID attack method. Authorization bypass is the result of unenforced access
control profiles (i.e., users should not be able to access other
Application Weakness Reference:
users’ data). Authorization and access controls are often not
CWE-89: Improper neutralization of special elements used in
consistently applied to all resources.
a SQL command
http://cwe.mitre.org/data/definitions/89.html Application Weakness Reference:
CWE-862: Missing authorization
Attack Method Reference:
http://cwe.mitre.org/data/definitions/862.html
CAPEC-66: SQL injection
http://capec.mitre.org/data/definitions/66.html Attack Method Reference:
CAPEC-87: Forceful browsing
WHID Incidents:
http://www.google.com/fusiontables/DataSource?snapid=S329834qI6g http://capec.mitre.org/data/definitions/87.html

WHID Incidents:
2. Logic Flaw http://www.google.com/fusiontables/DataSource?snapid=S3298656i9X

Risk Ranking Analysis: 5. Session Handling Flaws

Logic flaws are tricky as they are not easily integrated into
automated dynamic application scanning tools (DAST) and Risk Ranking Analysis:
are difficult to spot within standard Web application logging Session handling flaws allow attackers to impersonate a valid
mechanisms. The results of these two issues are that most and authenticated user. Attackers may manipulate Session
organizations cannot identify logic flaws and then are not able IDs (credential prediction), trick end users into authenticating
to spot if or when criminals exploit them. Only through manual a Session ID (session fixation) or use XSS attacks to steal a
application assessment can business logic flaws be identified. Session ID (session hijacking).

Application Weakness Reference: Application Weakness Reference:

CWE-841: Improper enforcement of behavioral workflow CWE-642: External control of critical state data
http://cwe.mitre.org/data/definitions/841.html http://cwe.mitre.org/data/definitions/642.html

Attack Method Reference: Attack Method Reference:

CAPEC-77: Manipulating user-controlled variables CAPEC-196: Session credential falsification through forging
http://capec.mitre.org/data/definitions/77.html http://capec.mitre.org/data/definitions/196.html

WHID Incidents: WHID Incidents:

http://www.google.com/fusiontables/DataSource?snapid=S329845bfBR http://www.google.com/fusiontables/DataSource?snapid=S329956MqHr
33 Security Weaknesses under the Microscope

6. Authentication Bypass 8. Source Code Disclosure

Risk Ranking Analysis: Risk Ranking Analysis:
To protect sensitive data or functions, applications rely on Proprietary application source code can be disclosed through
authentication controls as a first defense. Attackers can a number of methods, such as code left by developers
sometimes bypass these controls to access the application in browsable directories or misconfiguration in the Web
without credentials. This is a common vulnerability in Rich servers file handlers. Web application firewalls are often used
Internet Applications (RIA) and thick-client architectures. to identify code leakages and can block pages from being
Web services are another culprit as they do not prevent served to the client.
attackers from directly accessing them and instead assume
that authentication controls will be handled by the main Application Weakness Reference:
user interface. CWE-540: Information exposure through source code
http://cwe.mitre.org/data/definitions/540.html
Application Weakness Reference:
CWE-306: Missing authentication for critical function Attack Method Reference:
http://cwe.mitre.org/data/definitions/306.html CAPEC-116: Data excavation attacks
http://capec.mitre.org/data/definitions/116.html
Attack Method Reference:
CAPEC-36: Using unpublished Web service APIs
http://capec.mitre.org/data/definitions/36.html
9. Detailed Error Messages

WHID Incidents: Risk Ranking Analysis:

http://www.google.com/fusiontables/DataSource?snapid=S329897Ft92 Verbose error messages can provide significant aid to an
attacker. The error messages can provide configuration data,
source code or other useful information for fine-tuning attack
7. Cross-Site Request Forgery (CSRF) payloads. Error pages are also often used as the conduit for
data exfiltration when using SQL Injection attacks.
Risk Ranking Analysis:
CSRF allows a malicious website to force a legitimate user Application Weakness Reference:
to execute commands on the targeted Web application, CWE-209: Information exposure through an error message
possible when the command is formatted in a predictable http://cwe.mitre.org/data/definitions/209.html
manner known by the attacker. Unless the Web application
uses request validation tokens, it is most likely vulnerable to Attack Method Reference:
CSRF attacks. CAPEC-54: Probing an application through targeting its error
reporting
Application Weakness Reference: http://capec.mitre.org/data/definitions/54.htm
CWE-345: Insufficient verification of data authenticity
http://cwe.mitre.org/data/definitions/345.html
10. Vulnerable Third-Party Software
Attack Method Reference:
CAPEC-62: Cross-site request forgery Risk Ranking Analysis:
http://capec.mitre.org/data/definitions/62.html An application can only be as secure as the infrastructure it runs
on (i.e., application frameworks or servers). PHP applications
WHID Incidents: have a number of Remote File Inclusion (RFI) issues where an
http://www.google.com/fusiontables/DataSource?snapid=S329890UMOT
attacker can trick the Web application into downloading code
from a third party site and executing it. The main goals of these
attacks are either botnet recruitment or installing a Trojan/
backdoor interface for executing commands on the server.

Application Weakness Reference:

CWE-830: Inclusion of Web functionality from an untrusted
source
http://cwe.mitre.org/data/definitions/830.html

Attack Method Reference:

CAPEC-175: Code inclusion
http://capec.mitre.org/data/definitions/175.html

WHID Incidents:
http://www.google.com/fusiontables/DataSource?snapid=S331015Y6ZO
 Security Weaknesses under the Microscope 34

Blind Faith in Mobile

A mobile device such as a smartphone or tablet computer is often In the past several months both iOS and Android have come
mistaken for a miniature PC. Unfortunately this confusion has led under scrutiny for excessive and persistent collection of location
to many implementation mistakes and trust assumptions over the data. Additionally, several carriers have been criticized for not
past few years by organizations of all sizes and industries. disclosing the nature of the built-in diagnostic utilities on their
devices. Malware authors are certain to find ways to monetize
Most, if not all, mobile devices are consumer-grade technology. such a rich data source.
Consumer-grade mobile devices are designed to 1) attract the
widest spectrum of users, 2) be simple to use, and 3) drive sales Android Focus for New Malware
of media, games and other content. Until recently, mobile device The trend toward Android as the target platform of choice for
security has been slow to develop. Attack trends are starting to malware authors should come as no surprise. Android dominates
appear, though, and taking a proactive approach today can help the worldwide smartphone market, powering 56% of devices.16
mitigate risk tomorrow. Android’s ability to install applications from third-party stores
provides a direct route to consumers for malware authors.
In the past year, existing PC-based malware proliferated in Foreign marketplaces and those catering to pirated applications
the mobile space. Mobile devices can provide malware with are breeding grounds for counterfeit applications or legitimate
information such as location tracking and access to photos, applications modified to include malware.
video, and even audio. The Android platform is a major focus for
malware developers due to the availability of third-party market While Apple has had some short-term success at stymieing
places for applications and the ease of gaining root access. attempts to jailbreak their current iOS 5, strong desire from
the user base to install non-approved applications eventually
Mobile Integration of Banking Trojans resulted in a jailbreak in that platform. One can now expect to
2011 saw an increase in mobile versions of the largely PC-based see an increase in the release of new iOS malware. Although
banking Trojan malware market. The release of Zeus source a highly visible segment of the market, iOS still represents a
code, and its eventual merge with SpyEye malware, included minority of the market with only 18% of devices. Much like
Android and iPhone components used to capture Mobile Apple has experienced in the world of traditional, computer-
Transaction Authentication Numbers (mTAN) and mobile one- targeting malware, their somewhat limited market share may
time passwords. By capturing mTANs, bot authors are able to assist in reducing the platform’s attractiveness as a target.
access the banking information of those infected with SpyEye.
With mobile payment systems gaining traction, including virtual Mobile security faces challenges on several fronts. As mobile
wallets and near-field communication payment mechanisms device adoption increases, malware developers will think up new
built into phone hardware, malware targeting financial and ways to penetrate this insecure market. At the same time, traditional
payment card data continues to be a pressing concern. By malware is being updated to include mobile components,
abusing the mTAN authentication process, malware authors are furthering its reach. Increasing amounts of sensitive data are
demonstrating their ability to keep up with security mechanisms stored on our mobile devices, in turn increasing the relative value
meant to secure transactions. of each device to an attacker. Evolving mobile platforms must not
only fight with each other for market share, but must compete on
In 2011, the propagation of mobile banking malware has mainly features, including security and transparency. With such a young
been limited to social engineering rather than forced downloads yet ubiquitous technology, it is hard to pinpoint exactly where
or worm-type replication. With the strong trend toward developing the next security concern will arise, but it is fair to say that 2012
malware for Android (discussed below), it seems only a matter will bring its own share of interesting threat developments in
of time before a systemic vulnerability leads to greater infection this space.
rates through direct attacks and replication.

Location-Aware Malware
Carriers aren’t the only ones tracking the location of mobile
devices. Malware targeted at mobile platforms often collects GPS
and other location information, reporting back with other stolen
data. While it isn’t clear why attackers are collecting location
information, it is not difficult to imagine the ways to generate
value from it. For instance, stolen payment card information used
in a region local to the legitimate user is less likely to activate
fraud detection.

16
“Android market share reaches 56 percent; RIM’s, Microsoft’s cut in half.”
http://www.dailytech.com/Android+Market+Share+Reaches+56+Percent+RIMs+Microsofts+Cut+in+Half/article22852.htm
35 Security Weaknesses under the Microscope

Our Defenses: Networks susceptible to man–in-the-middle attacks leave a

user’s account vulnerable to compromise regardless of the

Four Basic Controls user’s password complexity. An attacker that captures a user’s
LM or NTLMv1 hash as they’re authenticating against an Active
Directory Service can simply pass the hash without even having
A perfect system does not exist. Under enough scrutiny, everything
to worry about cracking the password. NTLMv2 is not susceptible
has its flaws. Through identification and analysis of those flaws
to passing the hash and requires offline cracking where the
and sharing this analysis with industry, marked improvements in
strength of a user’s password would be tested.
security postures can be made.

In this section, four different types of defenses are reviewed by

Weaknesses in Cryptographic Methods
Another factor that can contribute to an account compromise
looking at the weakness that exists within the implementation of
is the cryptographic algorithm used to encrypt a password. If a
some of the most common security controls: business passwords,
weakness exists in the algorithm, an attacker will take advantage
data transmission encryption, anti-virus and the firewall.
of that weakness rather than resorting to attacks on the password.
No organization can do without these four basic controls.
An example is the use of LAN Manager (LM) hashes to store
Unfortunately, when controls are not implemented correctly or
Microsoft Windows passwords. LM hashing is considered to be
flawed from the start, there is a false sense of security imparted
a legacy algorithm but is still in use in many environments today.
upon the adopting organization, impacting both the security
It was used as the primary hashing algorithm for pre-Windows
posture and the operating budget.
NT systems, and was carried over and enabled by default in
later versions of Windows in order to maintain legacy support.
Business Password Analysis LM hashes can be cracked with rainbow tables in a matter of
minutes because of the way the hash is designed. When an LM
Passwords continue to be a pertinent topic of discussion and hash is created the user’s plain-text password is converted to all
study within both the security community and the world of upper case characters followed by null-padding the password
technology at large. However, few studies have had the advantage up to 14-bytes. This “fixed-length” password is two 7-byte
of large amounts of real-world data. In this section, passwords DES encrypted hashes. Instead of needing to crack the entire
from Trustwave’s client businesses are analyzed. password, an attacker can crack each half individually then
merge the two results together.
Password Risks Unrelated to Password Choices
The strongest password choice may not matter if the underlying Microsoft finally disabled LM hashing by default starting with
system is weak, whether due to a cryptographic weakness, exploit Windows Vista and Server 2008, but it is still commonly seen
or external factors. Even with solid technological foundations, a in Windows XP/2003 implementations. An LM hash has a
variety of human fallibilities can undermine the security of the 14-character limitation. If a user’s password is over 14 characters,
system as a whole. Windows will not hash the password with LM and only hash
using NTLM, thereby mitigating the issue. Other examples where
Even users proactive in ensuring account safety can experience attacking the cryptographic weakness is better than attacking
a system compromise by attack vectors unrelated to their the actual password itself is for WEP enabled wireless networks.
password selection. An ever-present example is the MS08-067
SMB vulnerability for Microsoft Windows 2000, XP/Server 2003, Old-Fashioned Methods
and Vista/Server 2008.17 This four-year-old vulnerability is trivial Writing down passwords is still prevalent within the workplace,
to exploit and enables an attacker to compromise a system in especially in organizations that implement complexity
seconds without requiring a single user password. requirements, frequent password expiration and password
histories to prevent password recycling. The effect of increasing
Patched systems can also become compromised as a result of password complexity policies is often reduced memorability, a
third-party services installed on a user’s system. This ranges from key requirement for a password. In approximately 15% of physical
unpatched services such as an Apache Web Server that enables security tests performed at client sites in 2011, written passwords
an attacker to exploit a buffer overflow, or more commonly, were found on and around user workstations.
remote access tools that bypass a user’s login credentials or
that only require a secondary password, such as free editions A tried and true method of attacks is the installation of a
of VNC. Installed by an end-user to enable remote access, VNC keystroke logger. To successfully install a keystroke logger, an
utilizes secondary passwords that bypass a user’s system login, attacker needs a user to leave their machine for only a matter
even if a secondary password is set. By default, VNC does not of seconds. Keystroke loggers can also be installed remotely if
encrypt communications between the client and server making a user’s machine is vulnerable to remotely executable exploits.
a user’s machine susceptible to man-in-the-middle attacks that
can capture a VNC password, if one is even set.

17
“Microsoft Security Bulletin MS08-067 – Critical. Vulnerability in Server Service Could Allow Remote Code Execution (958644).”
http://technet.microsoft.com/en-us/security/bulletin/ms08-067
 Security Weaknesses under the Microscope 36

Social engineering is increasingly being used by attackers to Shared accounts leave an environment open to a complete
obtain user account information. In this method, attackers take compromise if just one machine is compromised. It also enables
advantage of the trusting nature of people in an attempt to an attacker to focus on a couple of machines with exploitable
convince them to divulge sensitive information. Phishing attacks vulnerabilities; with this access, the attacker can obtain and
focused on social networking websites are now commonplace, crack passwords for shared accounts. Readily available tools
and pose a danger to the corporate environment because of the such as Medusa quickly allow an attacker to determine whether
associated risk of users using similar passwords for everything the account they’ve just obtained is in fact used elsewhere on
ranging from Twitter and Facebook to their Active Directory other machines.
account at work. Social engineering attacks can also involve
an attacker interacting with the user whose account they are Poor Password Selection
attempting to steal. Whether the attacker poses as the new IT Users are not creative when it comes to passwords, and creativity
administrator or uses bribery, extortion or intimidation, this non- was found to decrease with each successive password in our
technical attack vector has proven to be useful for attackers. study. User passwords featured local sports teams or activities
near their location. Users also created passwords based off their
Password Pitfalls company, whether a variation of the company’s name or products
In response to strong password policies, users are finding creative in their password.
ways to override these policies. Common examples include:
• Setting usernames as the password in cases where Passwords were also correlated to the time period in which the
complexity requirements aren’t enforced password was set. Users established a base password and
modified it with the time period in which they needed to reset
• Adding simple variations to fit usernames within complexity
their password, whether as a specific month, year or season. With
requirements, such as capitalizing the first letter and adding
this kind of predictability, an attacker need only crack a historical
an exclamation point to the end
password then make intelligent guesses on what the user’s current
• Using dictionary words verbatim or applying simple password might be.
modifications to adhere to complexity requirements
Incremental passwords were also a common practice. Users
Companies are also assigning poor default passwords for new set a base password then simply added one number to the end,
employees with examples such as “changeme” and “Welcome.” incrementing it from there when time came to change their
At times, users are not required to change such default passwords password, leaving the password open to attack by predictability.
upon login.
IT administrators should be aware of what passes as a complex
Service accounts, especially ones that are automatically password, especially in Active Directory environments. Users can
generated (e.g., accounts used between applications and back- create passwords that meet complexity requirements because they
end databases), were discovered to also include poor default contain the minimum amount of characters and include a couple
passwords, and IT administrators forgot to change them. A of character variations. The Active Directory password complexity
frequently occurring example was Microsoft SQL Server’s common policy states that a password is required to have a minimum of
system administrator (sa) username and password combination. eight characters and three of the five character types (Lower Case/
Domain administrators in Active Directory environments can also Upper Case/Numbers/Special/Unicode.) With that, “Password1”
ignore password policies if setting a user’s password through completely adheres to these policies, as does “Password2” and
Active Directory Users and Computers. “Password3.” Users are creating passwords that meet the bare
minimum requirements for length and character types, to aid with
Shared Passwords the memorability of the password.
Shared passwords can cripple an IT environment if compromised.
Shared passwords among services and machines are a common
tradeoff for manageability over security. Shared local administrator
passwords make administration of large numbers of machines by
IT staff possible, but do pose an inherent but generally accepted
risk. Another example includes accounts that are utilized by
common services across multiple machines that typically require
administrative access. A prevalent example would be accounts used
to initiate backup software on machines within an environment.
Accounts for the Backupexec suite are commonly discovered in
Active Directory environments to have domain admin privileges.
Accounts that are used for automatically logging into a machine
typically utilize a shared password, specifically on POS machines
in retail locations.
37 Security Weaknesses under the Microscope

Password Complexity versus Length Below are examples of the total possible combinations for a
Users and IT administrators often believe that simply adding password at a given length, utilizing a brute force attack. These
complexity to a password will make it inherently more secure. figures assume a standard 95 character U.S.-English keyboard.
While this might be the case when it comes to someone guessing
individual passwords, it’s not the case when it comes to utilizing
password-cracking tools. Character substitution or “L33tsp33k” Password Possibilities
is often used in passwords to (in theory) increase the difficulty of

Password Length
cracking a password.
10 5.98737x1019
Some character substitution examples are:

9 6.30249x1017
Original Character Replacement Character
8 6.6342x1015

A @ or 4 7 69,833,729,609,375

6 735,091,890,625
E 3
5 7,737,809,375

I ! 4 81,450,625

3 857,375
S 5
2 9025
Character substitution defends against a dictionary attack; if there
isn’t an exact match for that password in the dictionary list, it is
missed. However, when a password-cracking tool is utilized in a 1 95
brute force attack, the tool is already using all possible character
combinations and is not going to increase the difficulty or time
Total Number of IPv4 Addresses for Size Reference: 4,228,250,625
required to crack the password. However, simply increasing
Total Number of IPv6 Addresses for Size Reference: 3.40282x1038
the number of characters has a dramatic effect on the difficulty
of cracking a password. Every character added to a password
increases the possible combinations exponentially for a password,
Trustwave Password Study Methodology
Trustwave obtained the source for passwords from Trustwave
making brute force attacks on longer passwords impractical.
SpiderLabs investigations, most often from Windows Active
Directory servers, and a system to recover (“crack”) them.

Windows Active Directory was chosen as a password repository

for several reasons. First, it was the largest source of password
examples; nearly every organization has an Active Directory domain
for user account storage. Second, unlike Internet forum passwords,
Active Directory accounts are generally subject to higher scrutiny
and more restrictions in regard to complexity. This makes them
better examples of “honest” password choices by users. Third,
Windows passwords are recoverable by a wide variety of tools,
such as John the Ripper, Cain and others. Finally, the continued use
of weak Windows LAN Manager for storage of password hashes
makes recovery an easy task.
 Security Weaknesses under the Microscope 38

Password cracking was performed on a custom system built using

off-the-shelf parts totaling less than $1,500. The system was built with
an EVGA Motherboard, a quad core AMD processor, 16 gigabytes of
RAM, and utilized three NVIDIA 460GTX graphics cards (GPUs) as
the primary medium for the password cracking process.
Password1
A number of password cracking methods were utilized:

1. The password cracking process began with an attack

against all LM hashes. Using Cryptohaze Multiforcer18, a
free publicly available brute forcing tool, Trustwave launched
a brute force attack to recover the plaintext passwords
for all LM hashes containing standard ASCII characters.

2. Using the recovered plaintext passwords from the Cryptohaze

Multiforcer attack as a wordlist, Trustwave then used John
the Ripper, another free publicly available tool, coupled with a
set of word mangling rules against all NT hashes.

These attacks took roughly 10 hours and recovered more than

200,000 of the 2,521,248 passwords analyzed.

3. Using a third free and publicly available tool called

oclHashcat-plus from the Hashcat suite19 Trustwave
used publicly available wordlists coupled with the
Kore-Logic20 word mangling rules set in an attempt
to recover the remaining unrecovered NT hashes.

4. A fourth attack, known as a masking attack, was coupled

with oclHashcat and the same publicly available wordlists
to recover additional passwords for a total of 2,521,248
passwords analyzed.

To ensure client confidentiality, Trustwave filtered and

anonymized all passwords by removing identifying data (such
as usernames and domains) as well as randomizing password
hashes before recovery.

Top 25 Passwords
This list was created through accumulated data
combined with limited filtering (i.e., statistical
bias, industry-specific passwords, etc.).

Variations of “password” made up

about 5% of passwords and 1.3%
used “welcome” in some form.
0 20000 40000 60000 80000

18
Cryptohaze Multiforcer. http://www.cryptohaze.com/multiforcer.php
19
Hashcat. http://hashcat.net/oclhashcat-plus/
20
“Crack Me If You Can” – DEFCON 2010. http://contest-2010.korelogic.com/rules.html
39 Security Weaknesses under the Microscope

Top 20 Sequences Password Complexity

Password complexity can be affected by character sequencing. In this category, user choices are broken down by use of one or
In the following examples, the top user choices of the sequence more of each type of character in passwords. As before, users
of letter (l), number (n), and special character (s) can be seen. appear to lean towards letters and numbers, most specifically
lower case.
llllllnn (passwo12)
100000 Password Frequency by Complexity
l = Letter
n = Number All Lower

u = Upper Case All Upper

s = Special Character
All Number
80000
All Special

Lower/Upper

llllllln (passwor1) Lower/Number

lllllnnn (passw123)
Upper/Number
60000 llllnnnn (pass1234)
lllnnnnn (pas12345) Upper/Special

ulllllnn (Passwo12) Lower/Special

nnnnllll (1234pass)
ulllllln (Passwor1) Number/Special

ullllllln (Password1)
Lower/Upper/Number
40000 ulllllnnn (Passwo123)
ulllllnnn (Passwo123) Lower/Upper/Special
ullllllnn (Passwor12)
ulllllllnn (Password12) Upper/Number/Special

llllllll (password)
Lower/Number/Special
ulllnnnn (Pass1234)
ullllnnn (Passw123) Lower/Upper/Number/Special
20000 ulllsnnnn (Pass!1234)
ulllllnnnn (Passwo1234)
ulllllsnn (Passwo!12)
Password Length
ullllsnn (Passwo!1) In this category, the overall length of analyzed user passwords
can be seen:
60%

The most common sequence for passwords appears to utilize six 40%
letters and two numbers, followed closely by seven letters and
one number. This is a typical result for many Active Directory
installations as it correlates with easy to type and remember user
choices. However, this also results in easy to guess user choices:
20%
nearly all the combinations of these two sequences could be
guessed by basic rule set changes in password recovery tools.

Of additional note is that passwords containing special

characters do not appear until the 16th most popular choice. This
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
would indicate that users appear to avoid special characters, and
not subscribe to using them as substitutions for popular letters
(“L33tSp33k”). The most common password length appears to be eight characters.
The most likely reason for this is that eight characters tend to be
the accepted length for many Active Directory installations.
 Security Weaknesses under the Microscope 40

Keyword Usage As a result, “Password1,” while a terrible password choice,

The following statistics showcase some common user keywords: meets the same complexity requirements that “X$nc*(24” does.
Most users opt for the easier to remember password, and as
such they conform to the least required options that satisfy the
27,191 passwords requirements.

Months used English

spelling of months
Also of note is that within the default set of Windows AD rules,
there is no protection from similar password choices between
password histories. So while passwords cannot be the same
(January – December)
for incremental passwords (depending on the policy for the
domain), there is no native rule to prevent users from numerically
incrementing passwords. Thus, “Password2” could follow
U.S. 72,389 passwords “Password1,” and so on.

used U.S. States

States (Illinois, California)
Recommendations
The solution to password security starts with eliminating weaker,
older and insecure technologies. In the case of Windows AD, the
use of LAN Manager for password storage simply needs to go.
NT Hash-based storage, while not without issues of its own, at
74,368 passwords least allows for a larger key space (128 characters, Unicode) than

Seasons used seasons LAN Manager. On top of this, assistive cryptographic techniques
to slow down dictionary attacks would be a welcome addition.
(spring, fall) Many of these items are already available in Unix-based systems
but require third-party additions to work with Windows products.

170,013 passwords No solution is complete without some component of user

Baby
education and awareness. Users need to be instructed and
used names in the encouraged to avoid policy overrides – especially in the case of
“top 100 male and those users with an administrative capacity. Whether it’s setting

Names female baby names of an artificially weak password for a domain service, or eliminating
user password incrementing, these changes will enhance the
2011” list. baseline robustness of user password choices.

What should users consider in their password choices? For one, it’s
These categories were sometimes expanded to local sports time to stop thinking of passwords as words, and more as phrases.
teams, city nicknames or any information pertinent to the
organization’s location. “ThisIsMyPasswordNoReallyItIs” is, all things considered, a far
harder to guess passphrase than the previously mentioned
Implications “X$nc*(24”. Given that many rainbow tables have reached eight
When ground rules are set for security, users will often conform to nine or more characters for recovering NT passwords, length
to the lowest level of complexity that satisfies the requirements. is one of the few effective constraints left. Standing in the way
For example, the default “use complexity” setting in Windows of this, of course, is the ease of remembering the password. A
Active Directory requires: passphrase allows for the benefit of length and memory without
• The password is at least six characters long. overt complexity.
• The password contains characters from at least three of
the following five categories: A combination of a properly designed password storage method
and a properly designed methodology/policy for user password
• English uppercase characters (A - Z)
choice goes a long way. If either of these is weakened, the
• English lowercase characters (a - z) entire system is weakened; therefore some organizations may
• Base 10 digits (0 - 9) opt to explore alternatives to passwords where possible in their
• Non-alphanumeric (For example: !, $, #, or %) environment. Many areas of an organization can use two-factor
authentication to eliminate the reliance on user choice in the
• Unicode characters
security equation for particular authentication schemes.
• The password does not contain three or more characters
from the user’s account name.
41 Security Weaknesses under the Microscope

A Study of SSL
The story of Secure Sockets Layer (SSL) is intertwined in the
origins of e-commerce. Originally released by Netscape in
1995 in an effort to assure customers of the safety of Internet
transactions, it has been adopted and extended considerably
over the years. The first public release, version 2.0, contained a
number of security flaws that were later addressed as part of a
redesign, which resulted in version 3.0. This version is the basis
for the SSL implementation we use today.

SSL Certificate
The fully qualified domain
Subject name of the server

A unique identifier used to

Serial number identify the certificate

The Certificate Authority

Issuer that created the certificate

Specification of the
Signature algorithm used, SSL
certificates almost
algorithm exclusively use SHA-1 /
RSA

Start and end date during

Validity which the certificate is valid

Public key The actual key

Specifies what the

Purposes certificate may be used for

A cryptographic hash of
Signature the key

From a security perspective, the makeup of SSL certificates Security researchers and
found on active Internet systems provides an interesting view into even possibly criminals have
ecommerce security. In order to gather a large sample set to pull had recent success factoring
Certificate Authorities
data from, Trustwave’s SSL team scanned more than 17 million 512-bit RSA keys, due to the
(CAs) are trusted
internet-facing systems for SSL certificates and processed the computational power available
organizations that
results. This process yielded 289,926 unique certificates, which via cloud services. In turn,
issue certificates
were categorized according to a number of attributes. this has caused Certificate
used in the secure
Authorities (CAs) to refrain
identification and
Bit Strength from offering this type of
encryption of network
The strength of a key is generally associated with its bit strength, certificate, again raising the
transactions. Trustwave
referring to how long the key is, considering that each additional bar as part of the ongoing
is a Certificate
bit increases the amount of possible values that an attacker battle between attackers and
Authority and a top 10
would need to test. This type of attack is referred to as brute defenders.21 Cloud services
global issuer of SSL
force, where an attacker tries every possible permutation of will likely accelerate this
certificates.
values until the key is found. process due to the large-
scale, cost-effective amount
Most modern cryptography relies on computational security, a of processing that the model
methodology that increases bit strength as computational power offers.
increases in order to stay ahead of the attacker’s capabilities.

21
“RSA-512 certificates abused in the wild.”
http://blog.fox-it.com/2011/11/21/rsa-512-certificates-abused-in-the-wild/
 Security Weaknesses under the Microscope 42

Total Certificates by Bit Strength

A number of valid 512-bit certificates still exist on active Internet
systems, although adoption of 1024 and 2048-bit keys has been
Total Certificates by Bit Strength
predominant. As there have been no published instances of a Other 1%
1024-bit key being factored (the term used for discovering the 4096 1%
key when discussing asymmetric algorithms such as RSA), it is
512 1%
encouraging that this strength of key is being used in 41% of
certificates. 2048-bit certificates also offer good security and are
a strong choice for the near future.

Expired and Self-Signed Certificates

Every SSL certificate contains a field that denotes its validity
1024 2048
period. These fields often read as “not before” and “not after,” and 41% 56%
allow a client to ensure that the certificate is in good standing.
The use of a certificate after its valid period expires suggests
that the organization is not in good standing with the Certificate
Authority, and should be taken as an indication that the session
is not secure.

Certificates that are not issued by a trusted CA are known as

“self-signed” and represent another security concern in regard
to SSL. Self-signed certificates are only vouched for by the entity
Expired, Self-Signed and Valid Certificates
itself, with no third-party validation whatsoever. There are valid
Expired
use cases for these certificates in specific situations, such as
10%
internal resources in cases where clients’ Trusted Root is pre-
populated with the organization’s self-signed certificates, but
they should not be relied on in general. Self-Signed
18%
Hosts using Expired, Self-Signed
and Valid Certificates Valid
Self-signed certificates expired without renewal in 20% of
cases. Many of these invalid certificates came from instances 52%
of Web hosting control panels, such as CPanel and Plesk.
Several Hewlett-Packard printers were discovered as well, Self-Signed/
hosting expired and self-signed certificates. These platforms Expired
simply do not receive the same level of attention or scrutiny that 20%
e-commerce servers do, and often get overlooked in terms of
SSL certificate security. Self-signed certificates offer the client
no assurances to the validity of the server to which they are
communicating. In addition, organizations relying on self-signed Self-Signed Certificates by Bit Strength
certificates cannot revoke those certificates as they are not listed
on the two primary mechanisms that exist to perform revocation: Other 1%
Certificate Revocation Lists (CRLs) and Online Certificate Status
512 2%
Protocol (OCSP).

2048
Self-Signed Certificates by Bit Strength 28%
Self-signed certificates trend more towards 1024-bit encryption
than their CA-issued counterparts. The relatively small number of
512-bit certificates found here is encouraging. 1024
69%
43 Security Weaknesses under the Microscope

Long Validation Periods

Limiting a certificate’s lifetime ensures that an entity revalidate Extended Key Usage Common Key Usage Values
their ownership of the domain in question on a regular basis.
Trustwave imposes a limit of three years on any certificate TLS Web Server Digital Signature, Key
issued, which is a generally accepted standard among Certificate Authentication Encipherment
Authorities. Certificates that are valid beyond three years are
considered unusual, and as the time limit increases the reliability
of the certificate’s validity decreases. TLS Web Client
Digital Signature
Authentication

Others RapidSSL CA
42% 25%
Sign Executable Code Digital Signature

Digital Signature, Non-Repudia-

Email Protection
tion, Key Encipherment

CAs offer certificates with the certificate-signing attribute,

among other key usage values. Certificate signing is the value

CERT .
that allows a CA to validate and sign certificates belonging to
other entities; this value separates a CA from other entities in the
VALIDITY SSL trust model.
>3 YEARS
The Web browser, according to what is called a certificate
chain, enforces the certificate signing key usage value. A valid
GoDaddy certificate chain may have a number of entities, each performing
15% validation for the entity below. Each of these entities that signs
a certificate must also possess a certificate that includes the
www.psoft.net 1% Certificate Signing attribute. If not, the Web browser should
DigiCert High-Assurance CA-3 1%
generate an error that the certificate presented is invalid.
Equifax Secure Global eBus. CA 2%
VeriSign Class 3 Secure Server CA 2%
COMODO High-Assurance CA 2%
Starfield Secure CA 3% Certificate 1:
UTN-USERFirst-Hardware 3%
Network Soultions CA 4% Entrust.net Secure Server Certification Authority

Key Usage: Certificate Signing

Surprisingly most of the issuers are trusted CAs. The top two
CAs listed make up 40% of this segment, illustrating the disparity
between different CAs’ approaches to revalidation requirements.

Key Usage Certificate 2:

Certificates can be used for a range of functions, and each
certificate provides information about its own authorized uses. SecureTrust CA
In addition to Web server authentication, key uses also include
certificate signing, code signing, client authentication and other Key Usage: Certificate Signing
security roles. Roles are defined in the key usage and extended
key usage values of a certificate. These two fields have some
common elements. If the elements are not in agreement on
the certificate it is technically considered invalid. For example,
a certificate that specifies the “TLS Web server authentication”
Certificate 3:
role in the extended key usage field would also normally call out
www.trustwave.com
“digital signature” and “key encipherment” in the key usage field
to be consistent. This is because in order for a Web server to
Key Usage: TLS Web Server Authentication
establish its authenticity, it needs to provide a certificate that can
authenticate itself (digital signature) and allows encryption of
data (key encipherment).
 Security Weaknesses under the Microscope 44

Key Usage Security Concerns Number of Hosts Using Certificates Without Key
Certificate implementations naturally vary across software vendors. Usage (KU or EKU) by Issuer
To allow compatibility for any variance, Web browsers often accept
certificates even if certain fields are missing or incomplete.
x.509
Hosts Issuer
Data gathered by Trustwave has uncovered an interesting Version
statistic in this area: a small but significant number of hosts
that offer certificates lack any type of key usage extension. In
294 www.psoft.net 1
some cases, this is due to the use of the older X.509 version 1
standard, which does not support the key usage extensions. In
258 lifesize.com 1
other instances, the certificate in question uses version 3, but
omits these extensions altogether.
167 UM Web CA 3
The full ramifications of the version 3 certificates are unclear,
considering that each Web browser could handle them differently. 156 PCoIP Root CA 3
This scenario introduces uncertainty – it is a condition that should
not occur according to the specifications. Although by virtue of 123 TAA ROOT CA 3
being “extensions,” key usage can be omitted completely. The
result depends on how a browser programmer deals with this
111 localhost CA 3
unexpected case.

84 Foo Bar, Inc 1

OCSP Findings
As a Certificate Authority, Trustwave maintains an Online
66 Infrastructure Certificate Authority 66
Certificate Status Protocol (OSCP) server that allows Web
browsers to confirm the validity of an issued certificate. OCSP
was created as a more flexible alternative to the Certificate 58 dummy_ca.thecus.com 3
Revocation List (CRL) method to provide controls for certificates
after they have been issued. From a statistical perspective, OCSP 52 Spiceworks Desktop Install CA 1
data offers an interesting view into the client-side adoption of
this relatively new protocol. 50 Google Internet Authority 3

Apple’s mobile platform iOS is among the top requesters

47 Snake Oil CA 3
of OCSP data, even more so than Apple’s desktop operating
system. This suggests a tightly coupled integration of OCSP into
43 Dev CA 3
the iOS platform.

Also of interest is the predominant position of Windows in these 38 ca.three12.com 1

statistics, especially considering that Firefox represents a larger
share of the requests when compared to Internet Explorer (IE) and 35 Lebshama CA 1
Google Chrome combined. (The latter two statistics are combined
due to the fact that IE and Chrome share an OCSP library.) The Siemens Com ESY HD
strong showing of Windows XP hosts confirms that Microsoft’s 35 3
Security Office
10-year-old operating system is still maintaining a significant
market share. Windows 98 made an appearance, using a Mozilla 34 DMSBU CA (1024 bit RSA) 1
browser to perform OCSP on an unsupported platform.

34 ImageScape CA 1
The ever-expanding use of SSL, coupled with recent news
of successful attacks against Certificate Authorities and SSL
technologies themselves, are bringing more interest to this field University of Connecticut
34 3
Certificate Authority 2017
than ever before. As the sampling of data provided here has
shown, behavior can vary to a surprising degree across SSL
implementations. Whether talking about key usage values, bit 32 mmca.merunetworks.com 3
strength, certificate status checking or validation periods, every
issuer and browser developer has a unique take on implementation. 1699 Other hosts Various
45 Security Weaknesses under the Microscope

The SSL industry has matured considerably in the last decade.

OCSP greatly improves the ability to revoke certificates over CRL,
and Extended Validation offers clients another level of assurance
about the validity of a certificate. It is important that, as an
industry, we continue to convert security findings into security
improvements. The SSL system is a cornerstone of Internet trust,
and like any stone, should not remain unturned.

OCSP Requests by Browser / Library OCSP Requests by Operating System

Misc. <1% Android <1%
Opera 1% Undetermined <1%
ocspd (Safari & Chrome) Linux 1.49%
4.67% Macintosh
MS Crypto API
18%
(IE & Chrome)
17.18%
Apple iOS
Firefox 17.52% Windows
59.32% 61.99%
securityd
(Apple iOS)
17.52%

OCSP Requests by Operating System and Browser

Mozilla on Windows 98 <1%

Undetermined <1%
Opera on Windows <1%
Firefox on Linux 1.41% Other <1%
Safari/Chrome on Macintosh 4.67%
Firefox on Macintosh 5.54% Firefox on Windows NT 6.1(7) 20.88%

Firefox on Windows NT 6.0 6.08%

(Vista)

Firefox on Windows 11.52%

(Unknown version)
Apple iOS 17.52%

Firefox on Windows NT 5.1 (XP) 13.48%

IE/Chrome on Windows 17%
 Security Weaknesses under the Microscope 46

Anti-Virus:
The Elephant in the Room
“We have anti-virus, shouldn’t we be protected?” is often heard Malware targeting specific companies or products will often go
during Trustwave investigations. The historical perception of anti- undetected by anti-virus products simply because these vendors
virus and the sometimes blind faith in its ability to detect and never encounter this malware.
stop malware is one of the reasons attackers are so successful in
what they do. The accepted industry approach when a malicious New methods of detection, such as heuristics, have been
sample is discovered is to create a signature that can then be developed to detect previously undiscovered malicious files as
added to the anti-virus signatures for future detection. they appear. However, this technology is still in development and
has not reached a state of maturity that allows it to be used as the
The process of signature creation starts with identifying new primary method of detection in most anti-virus solutions.
malware, whether during a forensics investigation, witnessed on
a honeypot system, or received via a submission to an online
service. Once detected, a signature must be created for the
sample. It is during this phase that the battle between accuracy
and speed is fought. Detections are constantly performed and
New
signatures created and there is customer demand for immediate
protection from anti-virus companies.
Malware
Signatures need to maintain a level of quality in order to properly
Created
detect malicious samples. If signatures are created using a
method that is too generic, there is a possibility that false
positives may occur, leading to benign samples being detected
as malicious. Conversely, if signatures are too specific, there is a
possibility that a slight variant to a malicious sample will not be
detected at all, leading to false negatives. All of these factors are
taken into consideration when a signature is created.

After creation, signatures are pushed into the quality assurance New
(QA) phase in order to ensure no harm is caused to client Malware
Discovered
systems when deployed. Historically, there have been a few
instances where signatures have been pushed to customers that
flag critical files on the host operating system as malicious. This
led to those systems crashing, as the critical files were either
deleted or quarantined by the anti-virus solution.22 In one of the
stranger instances, Microsoft Security Essentials flagged and
removed Google Chrome as a banking Trojan.23 The QA phase is
often coupled with the signature development cycle.
Signature
Created
After the signature development phase, a final phase is enacted,
where updates containing the new signatures are pushed to
clients, requiring them to download the update. After client
systems have been updated, protections against this specific
malware sample or family of samples have been put into place.

When malicious samples first appeared and anti-virus was in its

infancy, this signature creation process made sense. However, Signature
as the number of malicious files increased, problems using Pushed in
the above method quickly surfaced. The delay in time between Update
when a malicious file is created and when the average user has
protection on their system leaves end users unprotected for long
periods. Even when detections are put in place, end users are only
protected against previously encountered, known malicious files. Signature Creation Process

22
“Horror AVG update ballsup bricks Windows.” http://www.theregister.co.uk/2010/12/02/avg_auto_immune_update/
23
“MSE false positive detection forces Google to update Chrome.”
http://www.theinquirer.net/inquirer/news/2113892/mse-false-positive-detection-forces-google-update-chrome
47 Security Weaknesses under the Microscope

100

Undetected
Percent Detection Rate

80 Malware

60

40

20
Vendor 1
Vendor 2
Vendor 3
Vendor 4
Vendor 5
Vendor 6
Vendor 7
Vendor 8
Vendor 9
Vendor 10
Vendor 11
Vendor 12
Vendor 13
Vendor 14
Vendor 15
Vendor 16
Vendor 17
Vendor 18
Vendor 19
Vendor 20
Vendor 21
Vendor 22
Vendor 23
Vendor 24
Vendor 25
The Results
Trustwave SpiderLabs aggregated more than 70,000 malicious Modern day firewalls have a variety of new features, including
samples in 2011 utilizing a propriety correlation and analysis application-level intelligence, onboard intrusion prevention, anti-
database. Samples were used to perform an analysis on anti- virus modules, load balancing, reputation intelligence and others.
virus coverage, overall and among vendors. On average, anti-virus Modern firewalls have very specific and contextual knowledge of
identified 81% of all samples analyzed (four out of five malicious a given traffic stream to enforce a much more granular level of
samples). The highest rate of detection came in at around 83%, control than their predecessors.
while the lowest rate of detection was found to be 70%.
Network Address Translation (NAT) is one firewall technology
Though 81% may be a passing grade for a student, when it concerns that surfaced in the early 1990s. NAT was proposed as a stopgap
the security of an organization, it demonstrates that relying on anti- solution to interconnect devices with the public Internet without
virus as a core component of the information security program still consuming as much public address space. It also made the
exposes the organization to malware threats. internal addresses un-routable on the external Internet by using
private address space (defined in RFC1918), familiar to most
Anti-virus will almost always be a key component in any information network users as:
security budget, but it should not be relied upon with the level of
confidence that many instill in it. Instead, it should be treated and RFC1918 Private Address Space
viewed for what it is: a single layer of defense against attackers,
Start IP Destination IP
but one that will be often and easily bypassed. Anti-virus should Prefix
always be used in conjunction with other techniques in order to Address Address
detect a threat against the host and/or network. Such techniques
10.0.0.0 10.255.255.255 10/8
include, but are not limited to, intrusion prevention systems, log
analysis, proper network segmentation, and properly configured
172.16.0.0 172.31.255.255 172.16/12
firewalls between segments.
192.168.0.0 192.168.255.255 192.168/16

Walking through Firewalls

When hosts from a private address range wish to communicate
A firewall, at its core, is a simple technology. It takes a set of with a public address, they need to go through NAT. There are
instructions given by an administrator and implements those two basic forms of NAT in use today: Source Network Address
instructions to influence whether a set of traffic can pass through Translation (SNAT) and Destination Network Address Translation
the device. Nearly 25 years after its introduction, and despite the (DNAT). SNAT performs modification of source addresses to a
plethora of competing technologies that have been introduced in public IP as traffic traverses from private to public address space
recent years, it is still critical for IT security. to ensure a return path. DNAT compliments this by performing
modification of destination addresses as traffic traverses from
 Security Weaknesses under the Microscope 48

Outbound Traffic (SNAT) Inbound Traffic (DNAT)

Source IP Destination IP Source IP Destination IP

Security Zone Security Zone
Address Address Address Address

Private 192.168.1.1 1.1.1.1 Private 1.1.1.1 2.2.2.2

——MODIFIES SRC IP ONLY—— ——MODIFIES DEST IP ONLY——

Public 2.2.2.2 1.1.1.1 Public 1.1.1.1 192.168.1.1

public to private address space to ensure the traffic hits the true When looking at a simple BNAT scenario, we see a slightly different
initiator. SNAT and DNAT are able to accomplish these tasks by result, which results in a broken communication channel.
maintaining a state table of which devices are communicating to
and from private and publicly addressed networks. 1. 192.168.1.1 --------> SYN --------> 192.168.2.1
2. 192.168.1.1 <------ SYN/ACK <------ 192.168.2.2
Despite the maturity of firewalls, little progress has been made 3. 192.168.1.1 --------> RST --------> 192.168.2.2
to improve security of the underlying components. Vulnerabilities
present in lower-level functions, such as NAT, can confuse In this case, because 192.168.2.2 responded to our request instead
higher-level functions resulting in a degraded security state. of 192.168.2.1, our client terminates the connection with a TCP RST,
It is also likely, that as organizations and product companies as we were trying to talk to 192.168.2.1 and not 192.168.2.2.
begin to focus on advanced functionality they lose sight of the
underlying core importance of foundational firewall and border BNAT scenarios are usually a result of a device misconfiguration
device components of such as stateful inspection, traffic flow and or device subsystem malfunction. They are more likely to occur in
network address translation. complex networks, such as when an organization deploys multiple
infrastructure vendors without a consistent vision of the overall
Increased network complexity is slowly creating gaps in defenses. network traffic flow. BNAT commonly exists in environments
These gaps may not introduce substantial risk by themselves where asymmetric routing is present. Asymmetric routing is
but, taken as a whole, they can have a significant adverse effect IP communication that takes different paths from source to
on the overall security posture of a given network. In response, destination and destination to source.
organizations are asking professionals to specialize more,
creating an experience gap at the ground level where networking Trustwave’s Managed Security Services team frequently identifies
and security meet. We often see this during our incident response BNAT conditions and helps organizations correct these scenarios
investigations; the network and security administrators may when installing unified threat management (UTM) and other
sit across from each other, but assume certain aspects of the stateful enforcement devices into customer environments. When
environment are being managed by the other. In actuality, no one BNAT scenarios go unidentified and uncorrected, the traffic flow
is managing those aspects. through a network can cause improper NAT actions, resulting
in a broken communications channel similar to the initial BNAT
Due to these growing gaps, Trustwave SpiderLabs performed scenario example noted previously.
research to determine ways in which this core function could
be exploited. During this process we identified a new attack Trustwave SpiderLabs recently identified “BNAT hijacking”: a
vector, dubbed “Broken NAT” (BNAT), which could be exploited malicious user successfully makes use of broken communications
by malicious users to gain access to internal devices previously channels and converts them into valid TCP sessions with little
thought inaccessible. effort. BNAT hijacking is achieved by making the local TCP stack of
the malicious user more forgiving when receiving responses from
BNAT in its most basic form is observed during TCP session initiation. an uninitiated target by “Reflectively ACKing”(rather than RSTing)
When a client wants to initiate a normal TCP session with a server and then pivoting to the SYN/ACK responder for the remainder of
they need to perform a TCP 3-way handshake as follows24 : the communication session.

1. 192.168.1.1 --------> SYN --------> 192.168.2.1

2. 192.168.1.1 <------ SYN/ACK <------ 192.168.2.1
3. 192.168.1.1 --------> ACK --------> 192.168.2.1

24
TCP 3-way handshake defined in RFC 793, Figure 7. http://www.ietf.org/rfc/rfc793.txt
49 Security Weaknesses under the Microscope

1. 192.168.1.1 --------> SYN ----------> 192.168.2.1

2. 192.168.1.1 <------ SYN/ACK<------ 192.168.2.2 Port Service Percent
3. 192.168.1.1 --------> ACK --------> 192.168.2.2
4. 192.168.1.1 ------> PSH/ACK ------> 192.168.2.2 21 FTP 9%
5. 192.168.1.1 <------- ACK <-------- 192.168.2.2
22 SSH 9%
Note: 192.168.2.1 only sees the first SYN packet; the remainder
of the connection traverses through 192.168.2.2 25 SMTP 10%

80 HTTP 34%
This new process accomplishes a number of things that could be
useful to a malicious individual trying to exploit a BNAT service:
443 HTTPS 34%

1. Allows completion of the TCP 3-way handshake with a 445 Microsoft-DS 1%

service that was previously unreachable
2. Allows bypass of stateful inspection and other advance 1433 MS-SQL 2%
application controls
1521 Oracle DB 0%
3. Allows inbound initiated communication through an
egress-only device 3306 MySQL 1%
4. Allows exploitation of vulnerabilities that may exist
3389 RDF 2%
Trustwave SpiderLabs analyzed 250,000 public IP addresses from
132 countries to determine whether or not BNAT exists on open Note: Percent value is the number of instances of each service
Internet and how prevalent it really is in the wild. over all instances found.

This analysis included a port scan of each host on each of the This data is not surprising as Web and email are two of the top
services listed in the tables to the right. If the host responded with three of services used on the Internet today. Within these services,
a TCP SYN/ACK response with a matching sequence number a subset of BNAT services existed. This means that the service
(+1 of the ISN) then it was included in our data set. If the host responded, but the response traffic received did not match the IP
responded with a TCP SYN/ACK response matching the port and address requested.
sequence number, but not IP, the service is a BNAT service.

Port Service Percent

Port Service
21 FTP 4%
21 FTP
22 SSH 1%
22 SSH
25 SMTP 8%
25 SMTP

80 HTTP 80 HTTP 9%

443 HTTPS 443 HTTPS 74%

445 Microsoft-DS 445 Microsoft-DS 1%

1433 MS-SQL 1433 MS-SQL 1%

1521 Oracle DB 1521 Oracle DB 0%

3306 MySQL
3306 MySQL 0%
3389 RDP
3389 RDF 1%

Of the 250,000 IP addresses scanned, only those that responded

to a TCP SYN request with a TCP SYN/ACK on one or more the Note: Percent value is the number of instances of each BNAT
services listed above were considered in scope. These yielded service over all BNAT instances found.
approximately 60,000 live hosts with the following distribution of
TCP services:
 Security Weaknesses under the Microscope 50

The most surprising result of the service distribution of the BNAT exists in the wild across various services and multiple
identified BNAT services was that although HTTP and HTTPS geographic boundaries. It is highly recommended that
shared about 35% of the total services identified, HTTPS was organizations that have a publicly facing Internet presence
eight times more likely to yield a BNAT service than HTTP. This is assess their environment to ensure that they do not have exposed
likely due to various asymmetric routes that are introduced when BNAT services. These services effectively hide from modern port
load balancers of e-commerce systems are performed. and vulnerability scanners, and can go long periods without
detection. Recently, Trustwave SpiderLabs added BNAT detection
to its TrustKeeper vulnerability scanning solution and has been
assisting customers in fixing the issues identified.
Country Percentage
It is recommended that professional service organizations, like
IT auditors, penetration testers and security consultants alike
Ireland 0.96% scan for BNAT when performing assessments for their clients to
ensure they are protected. Detecting BNAT with the right tools is
easy and not much more work than a simple port scan.25

Hong Kong 0.81%

Canada 0.72%

Japan 0.53%

Mexico 0.46%

United Kingdom 0.21%

United States 0.05%

Note: Country distribution is not a representative sample of

each country as a whole.
Note 2: 71% of all BNAT instances discovered were located in
one of the seven countries listed above.
Note 3: Percent value is number of BNAT instances found over
all services for each country.

In a number of countries where a series of IP addresses were

scanned, a large number of active services were found but
absolutely no BNAT services. These countries, in order of size,
were: Australia, Germany, Sweden and China.

Overall, Trustwave positively confirmed that all but two of the

services identified above (Oracle DB and MySQL) were present
on the Internet. An average of one BNAT service existed for every
790 live hosts identified by the scans. When comparing these
results to the current number of live Internet hosts (about 850
million hosts) reported by the Internet Systems Consortium in
July of 2011, the number of hosts exhibiting BNAT services is
estimated at more than one million.
25
Trustwave currently maintains a set of open source tools on GitHub called “BNAT-Suite” (github.com/spiderlabs/BNAT-Suite) that can be
used not only to detect, but also to hijack BNAT scenarios to better help organizations and security industry professional understand this
new attack vector.
51 Information Security Strategy Pyramid for 2012

Information
processes to identify attacks when they occur and resolve related
problems quickly. Trustwave SpiderLabs recommends six areas
for all organizations to focus on in 2012:

Security Strategy Each area, if implemented fully, will support and enhance the area
placed above it. Once the top is reached, the process starts over

Pyramid for 2012

at the bottom — it is a continuous effort work to refine security
programs. The pyramid can be explained from the bottom up.

Education of Employees
Improving the security posture of an organization may not be an
easy task. If 2011 was any indication of what the future will bring, Employees are the first line of defense against physical and
it is no longer a matter of “if” you will be attacked, but “when.” digital attack vectors. A lack of proper training and awareness
The security goals for 2012 should be to prevent as many cyber can turn employees from assets to liabilities.
attacks as possible, as well as streamline information security
Attackers are motivated to use phishing
attacks as they require only a small percentage
of recipients to perform actions that assist

Visualization the attacker in gaining unauthorized access.

Untrained employees click links to malicious
websites, opening a backdoor into the
\\\\\\ of Events corporate network, or expose their corporate
credentials by logging into what appears
to be a legitimate (but is actually attacker-

Unification
controlled) corporate website. These actions
can result in loss of intellectual property
and exposure of customer data, leading to
incident response investigations, costly fines,
\\\\\\\\\\\\\\ of Activity Logs
and a detrimental effect on an organization’s
reputation and brand.

Registration An alternative to casting such a wide net is

focused phishing attacks against specific
individuals. For instance, the attacker may
of Assets
\\\\\\\\\\\\\\\\\\\\\\ be very interested in a person with privileged
access to sensitive systems, applications
and data within the target organization.

Homogenization With proper education, a targeted employee

can be the first detector of an attack that
of Hardware and Software would otherwise circumvent preventative
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ technical controls. Security-aware employees
are better able to determine if an email is

Identification malicious and will follow proper procedures

to alert security staff.

of Users Physical security awareness is just as

\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ important. Tailgating and social engineering
are less effective when personnel have been

Education
taught the importance of their individual
responsibilities for physical security. In
many respects, it requires education around
of Employees why social norms (like holding an access
door open for the person behind you) can
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ have a negative impact on security for the
organization.
 Information Security Strategy Pyramid for 2012 52

Employees should be encouraged to challenge unknown an employee forgets their badge, all they need to do is request a
individuals in the correct circumstances and to ensure those temporary keycard for the day. Ensure every single person who
around them follow the correct physical access control has access to facilities and systems is identified as they traverse
procedures. This can help prevent losses and ultimately save the physical and digital environment.
organizations money. In some cases, it can also protect the
lives of employees should the person tailgating or attempting to Initiatives and technology to support this area:
bypass controls have intent to physically harm employees.
Logical Access Management Reviews
By changing the mindset and behavior of employees through Performing periodic analysis of all user and group roles will
education and reinforcement of positive behaviors, businesses improve security around employee access levels and may
build a solid foundation for a security program. In the 2012 even identify accounts that are no longer needed.
security pyramid, Visualization isn’t remotely possible if
employees, especially those tasked with security, do not have Password Complexity Policies
situational awareness to sound the alarm when something Set password policies of high complexity and educate
doesn’t look, feel or sound like it should. staff on best practice password techniques, such as
using passphrases.
Initiatives and technology to support this area:
Two-factor Authentication
Security Awareness Training Two-factor authentication allows users to authenticate by
Regular training of staff on both core security techniques both what they know (a password) and what they have (a
and topical issues is important to build a successful security device or certificate). This should not only be applied to the
foundation. digital world, but also the physical world. Combining key-
card access with PINs is a way to accomplish this.

Security Awareness Campaigns Biometrics

Repetition is key; regularly featured security topics will help Beyond passwords and other authentication mechanisms,
staff awareness levels and help maintain employee vigilance. biometrics may be necessary for more sensitive areas of the
workplace, such as data centers and R&D environments.
Rewards for Incident Identification
Monetary or other rewards can help encourage employees
to be observant and report security events. Homogenization of Hardware
and Software
Attack Simulation Exercises
Like a fire drill, attack simulations can help staff understand Fragmentation of enterprise computing platforms is an enemy
how a security event may appear and what they should do to security. IT departments are often busy just keeping systems
in response. up and running, and security takes a back seat. If platforms
and devices become unmanageable from an inventory
standpoint, keeping tabs on security-related issues can become
Identification of Users unmanageable as well.

Once employees are educated on their roles in protecting their Fragmented environments can require unrealistic maintenance
organization’s assets, they need to be properly identified. Every efforts and cause major security issues. One day it is an issue
user-initiated action should be tagged to a specific employee, with an Internet browser that 20% of employees like to use, the
consultant or other third party. This may seem a lofty goal, but it is next day it is a document reader on all the Mac devices, and on
achievable in most environments. the third day it is a flaw affecting only those using Android on a
certain hardware platform.
The first task is the eliminations of generic, shared, vendor and
default accounts. Every single year a significant number of data The more diverse the hardware and software population in the
breaches occur as the result of an attacker obtaining a user environment, the more IT and security teams need to react to
account for a system. More often than not, attackers utilize a problems. Reducing this fragmentation through standardization
shared vendor or default account that should have been changed and decommissioning of old systems will homogenize the
before the application was placed into production. environment, making it easier to manage, maintain and secure.

This level of security must be adhered to in the physical world

as well. Employees may wear badges and key cards for access
control and logging of movement within a facility, but as soon as
53 Information Security Strategy Pyramid for 2012

Initiatives and technology to support this area: implementing a device registration process and limiting BYOD,
businesses will have better oversight of what devices access
Policies corporate networks when and for what purpose.
Defined based upon risk assessment exercises, policies
should dictate how future decisions are made in regards to Security controls also play a strong role this area. A device should
platforms and software packages used by employees. never be allowed access to a controlled environment unless it
is registered and known. In addition, the patch levels and
vulnerabilities should be assessed on a regular basis, not only
Hardware Standards to work to improve the security of those in the environment, but
Identify standards to adopt a “less is more” strategy. Fewer
also to understand what risks exist when issues can’t be resolved
standards can help to reduce complexity, an enemy of security.
in the short term.

Decommissioning of Old Systems Initiatives and technology to support this area:

Too often, through both our investigation and penetration
tests, we find vulnerabilities in systems that are no longer
needed for business purposes. Work to aggressively turn off
Asset Management
Institute a system to track devices, applications and other
and unplug these systems from the environment. Upgrade
assets owned.
or replace systems to align with policies and standards. In
circumstances where this is not a business option, triage
(understanding the security issues and risks) and treatment Network Access Control (NAC)
(applying compensating security controls) are options that NAC can control access to various network environment based
should be considered. upon defined rules. It can also be used to remove devices from
the network if and when security issues are identified.

Registration of Assets Patch Management

When there is an active threat understanding the patch
Educated, identifiable users armed with specific approved levels of systems and applications are critical.
devices, running specific applications, is a good place to start
to create an easier environment to navigate from a security Vulnerability Scanning
standpoint. We now need to know which devices are entering Even with the above solutions in place, there will still be
our networks and when, and at the same time understand their instances where configurations or the combinations of
individual security posture. various services will introduce vulnerability. Regular scanning
of both internal and external systems should be performed.
Networked devices are widespread in organizations today. It is
more important than ever to have a complete inventory or registry
of valid devices.

Bring Your Own Device (BYOD) is a burgeoning trend; employees

Unification of Activity Logs
are accessing the corporate network via their own smartphones, Combining the physical world with the digital affords
tablets and even laptops. Allowing users to use any device they organizations new ways to identify security events more quickly.
want with no security controls, however, will destine the security Most businesses today treat physical controls and information
program for failure. security controls separately. Badge systems, HR records and
even loss prevention are not typically tied to the same team
Businesses that adopt a BYOD policy are opening the door to that monitors firewalls, intrusion detection systems and other
malicious threats. Take, for example, the announcement of security technology.
malware embedded on the motherboard of a specific laptop
model. Users can be asked to check their laptop type, but some From the results of our investigations, it is clear attacks are
won’t report on this accurately, making a survey of devices becoming more sophisticated and criminals are finding out how
unreliable. Without a device registration process, a BYOD lax security controls are in most organizations. Attackers also
business is never sure that a vulnerable device doesn’t exist on know that certain activities may not be flagged as suspicious.
the network. Consider the following scenario:

From desktops to laptops, from servers to mobile devices, A world-wide business employs many individuals who regularly
anything that can connect to an organization’s systems is capable travel for their jobs. While waiting for a flight, one such individual
of providing a unique identifier. A unique identifier aids in access attempts to access their email on a laptop. A certificate warning
control and can provide an accurate record of what devices have pops up but is ignored (see Education above). Their credentials are
access to the environment and when that access is initiated. By intercepted and stolen. A few days pass and the employee is back
 Information Security Strategy Pyramid for 2012 54

in the New York office. While he is sitting at his desk, the attacker
connects to the environment via the email account information he
Visualization of Events
obtained and begins to download email.
Daily business activities take place millions to billions of times per
day in most environments, but all it takes is one security event for
In most organizations, this scenario would not raise an alarm.
a company to make the headlines for the wrong reasons.
When the attacker logs in, the IT environment registers that act
as an employee accessing his email while outside the office.
Security event visualization in the enterprise isn’t practiced
However, the employee is currently physically located in a New
frequently, most of the time it is just considered log review. Many
York office, logged into the domain from the office network, and
security professionals still use spreadsheets to perform their
not traveling (see Registration above). This scenario becomes
analysis — after the event has occurred and the damage has
more serious when the employee is a high-profile individual with
been done. In the previous section we wrote about the Unification
access to sensitive data.
of data using tools like a SIEM. For most organizations today, this
is where the path ends. The ultimate goal should be to develop
The first step to addressing this attack scenario is to reduce the
an environment that allows for security events to be discovered
number of consoles. Instead of viewing multiple consoles and
by seemingly innate observations by both the people who are
attempting to correlate data across all, feed the logs of these point
tasked at protecting the environment and those who are not.
solutions into a single console. During this process, review each
Data aggregation or correlation as seen in a SIEM is a precursor
point solution to ensure they are: 1) configured correctly, and 2)
to real-time security event visualization and notification.
they are logging as much as possible.

After automating analysis, acknowledge there are tasks

Too often many tools are in place but administrators have tuned
computers can’t do very well and design analysis processes to
them down to reduce the “noise,” and they no longer provide
coordinate employees working with computers. Present items to
anything of value. Instead of tuning, use a tool, such as security
administrators that a computer can’t understand in a way that
information and event management (SIEM) technology, to take
the validity of the action can quickly be determined, or that would
over the processing of these logs; all “noise” will just be data for
encourage further investigation.
this technology.

Consider using colors and sounds as data is presented to

Unification of systems will benefit awareness of the attack scenario
employees. Trustwave SpiderLabs research in this area revealed
described above and other types of events, as well as help improve
that “urgent” flashes of light or beeps are not effective at gaining
the accuracy of Visualization. In a unified scenario of this example,
attention or driving actions. Over time people ignore them.
the office badge swipe combines with the local domain login and
Employees are more likely to notice subtle changes in color or
the company issued computer. Correlating this information with
audible tones.
the act of email accessed from a location outside the office from a
non-company-issued machine yields a red flag.
Initiatives and technology to support this area:
Initiatives and technology to support this area:
Custom Visual / Environmental Controls:
Explore tying the physical environments in which administrators
Logging Option Analysis work with the potential security changes occurring in the
Logs are sometimes turned off or tuned down to the point
digital environment.
where they become useless in identifying security events.
Analysis should be performed to maximize the amount of
events captured. Experimental
Trustwave SpiderLabs’ cerealbox is a tool that demonstrates
a method of tying various events on computer systems to
Point Security Solution Tuning visual indications. The goal is to prompt the user to investigate
Over time security logs may not be reviewed as frequently and
when something is different or looks odd, rather than having
may even be tuned to limit the “noise” they are generating.
to constantly review logs or receive pop-up messages on their
Tuning these solutions regularly to ensure proper data
console to indicate so. The tool can be found at
capture and review is happening is recommended.
https://github.com/SpiderLabs/cerealbox.

Security Information and Event Management:

A SIEM helps achieve log normalization and correlation, and
allows for rules to be applied to trigger security events.
55 Global Conclusions

In 2012 and beyond, some predictions and

recommendations can be made. First, cyber attacks
are increasing, as evidenced by the frequency of media
reports and the growing queue of incident investigations
conducted by Trustwave SpiderLabs. There is no sign of
abatement.

Possession of customer records makes businesses a

target. The risk is even greater for businesses frequented
by consumers and brand name chains. Technology may be
necessary to protect the network, such as Web application
firewalls and network access control, and the data itself,
such as encryption and data loss prevention.

Outsourcing IT and business systems to a third party

increases risk, as many of those organizations may not
have client security interests in the forefront. When those
third-party systems are used for remote access, criminals
are more able to access the corporate environment due to
weak and default passwords. Change default passwords
and work with vendors to ensure they are following security
best practices and adhering to industry requirements.

Global Employees will continue to choose poor passwords.

Enacting and enforcing stronger policies, and encouraging

Conclusions
longer passphrases, will help mitigate this risk.

Out-of-the-box anti-virus is not effective against many

classes of threat. Don’t rely on anti-virus to solve security
problems. Instead, adopt a security plan that uses both
automated and manual testing techniques to identify
unknown vulnerabilities and security gaps.

Finally, firewalls deployed years ago are often no longer

effective due to flaws in the original design or its use.
Review the configuration of firewalls and make a plan to
update or install a modern implementation.

Organizations that approach their security initiatives in a

committed manner and as an integrated requirement for the
entire business will be most resilient to attack. By reducing
risk through education, identification, homogenization,
registration, unification and visualization, organizations will
not only protect sensitive data and their employees, they’ll
also safeguard their reputation.

The Trustwave 2012 Global Security Report identifies many

areas where organizations can improve or implement
new procedures and technologies. By learning from the
misfortunes of others, and applying tactical and strategic
changes outlined in this report, organizations worldwide
can build better security programs and frustrate cyber
criminals everywhere.
 Appendix: What is a Penetration Test? 56

Appendix Why Vulnerability Scanning is not

What is a Penetration Test? Penetration Testing?
Testing data protection strategies by using the same tools and Vulnerability scans are automatic tools that attempt to identify
techniques an attacker would use is one of the most useful vulnerabilities that an attacker might be able to exploit. They are
ways to be assured those strategies work. This method is called often used for finding “low-hanging fruit,” such as basic or common
penetration testing. It is a controlled engagement whereby a configuration mistakes that could be exploited by an attacker.
qualified professional attempts to test network and application
security controls by trying to gain unauthorized levels of access, Vulnerability scanners can’t understand data like a human can,
ultimately to access sensitive data. so an automated scanner would not know if it was dealing with a
mundane document, or highly sensitive board room minutes. Nor
RECONNAISSANCE can vulnerability scanners understand the background context
of a security test, meaning they are very poor at assessing the
Start Assessment impact of a specific vulnerability. Finally they are also not able to
identify certain classes of security issue, such as subtle business
Public Client
logic flaws (perhaps related to a complex approval process within
Records Provided
Search Information a supply chain management application). It is precisely these
Target Gathering
security issues that often have the most serious consequences!

Organizations should ask these questions before engaging in a

MANUAL TESTING penetration test or scanning service:

• Are we trying to defend against a low-skilled attacker

Application Mapping who is able to download and run a vulnerability scanner
against us?
• Are we trying to defend against intelligent adversaries
Automated who can cleverly exploit a chain of vulnerabilities, possibly
Session Analysis Tools across seemingly unrelated assets, to gain access to our
Data sensitive data?
Extraction
The answer is yes to both. Just running a vulnerability scanner
Logic and Fraud Abuse
against IT assets is not enough to secure an organization.
Trustwave SpiderLabs consistently finds critical and high-risk
material vulnerabilities in environments that undergo regular,
Issue Identification
automated vulnerability scanning.

Alert Client Who Needs Penetration Testing?

Vuln. Confirmation on High
or Critical Organizations with sensitive information, such as customer data,
personally identifiable information, payroll data, payment card
data, intellectual property and other data types should consider
penetration testing.
YES Compromise?

Some organizations use a data classification policy (with associated

procedures) that describes how different types of data within the
NO business should be protected and handled. However, even the
most detailed data protection strategies can have vulnerabilities.
REPORTING
Any organization that has electronic data that they would not
Final Report/ Close Out Call want exposed publicly should consider regular penetration
testing. Trustwave SpiderLabs conducts penetration tests against
networks and applications for many different types of sensitive
End Assessment data. The entire testing process is primarily manual to limit generic
results often received from scanners and checklist methods used
57 Appendix: What is a Penetration Test?

in general vulnerability assessments. In this way, Trustwave can

focus the engagement on directed attack logic-based testing
About Trustwave®
against systems and networks
Trustwave is a leading provider of on-demand and subscription-
based information security and payment card industry compliance
What is the Difference Between management solutions to businesses and government entities
throughout the world. For organizations faced with today’s
Network and Application challenging data security and compliance environment, Trustwave
provides a unique approach with comprehensive solutions that
Penetration Testing? include its flagship TrustKeeper® compliance management
software and other proprietary security solutions including SIEM,
A network penetration test typically includes entire networks WAF, EV SSL certificates and secure digital certificates. Trustwave
and many hosts, the testing of which is focused at the network
has helped hundreds of thousands of organizations-ranging from
layer. This type of assessment is typically performed “blackbox”
Fortune 500 businesses and large financial institutions to small
(without any authentication credentials or privileges). Network
and medium-sized retailers-manage compliance and secure
layer penetration tests should be performed both externally
(against Internet-facing servers and supporting infrastructure) and their network infrastructures, data communications and critical
internally (against internal corporate information systems assets, information assets. Trustwave is headquartered in Chicago with
including servers, workstations, routing and switching equipment, offices throughout North America, South America, Europe, Africa,
printers and IP telephony systems). Asia and Australia.

Application penetration testing involves a targeted assessment For more information: https://www.trustwave.com.
of an individual (commonly, although not exclusively, Web)
application. This application could either be on the Internet or
accessible only internally to employees and third-party customers
or partners. About Trustwave SpiderLabs®
Application penetration tests will almost always require that SpiderLabs is the advanced security team within Trustwave
the penetration tester receive authentication credentials to the focused on forensics, ethical hacking and application security
applications, specifically two sets of credentials for each type of testing for our premier clients. The team has performed hundreds
“user role” that exists within the application. The reason for this
of forensic investigations, thousands of ethical hacking exercises
is two-fold:
and hundreds of application security tests globally. In addition,
the SpiderLabs research team provides intelligence through
• Typically the dynamic data creating/reading/updating/
bleeding-edge research and proof of concept tool development
deleting functions of an application are only accessible
to enhance Trustwave’s products and services.
post-authentication. Security vulnerabilities within these
areas of functionality are likely to be most serious;
For more information: https://www.trustwave.com/spiderLabs.php.
• Tests to ensure one user cannot create/read/update/delete
data belonging to, or by pretending to be, another user
require two users at each user role.

Often organizations believe that only the corporate website

needs application penetration testing. However, the corporate
website is typically one of many Web applications an enterprise
would have facing the Internet. Trustwave SpiderLabs has
worked with individual customers that have more than 1,000
business applications.
 About Trustwave ®
Trustwave is a leading provider of on-demand and subscription-
based information security and payment card industry compliance
management solutions to businesses and government entities
throughout the world. For organizations faced with today’s
challenging data security and compliance environment, Trustwave
provides a unique approach with comprehensive solutions that
include its flagship TrustKeeper® compliance management software
and other proprietary security solutions including SIEM, WAF, EV SSL
certificates and secure digital certificates. Trustwave has helped
hundreds of thousands of organizations-ranging from Fortune 500
businesses and large financial institutions to small and medium-
sized retailers-manage compliance and secure their network
infrastructures, data communications and critical information assets.
Trustwave is headquartered in Chicago with offices throughout North
America, South America, Europe, Africa, Asia and Australia.

For more information: https://www.trustwave.com.

Corporate Headquarters EMEA Headquarters LAC Headquarters APAC Headquarters

70 West Madison St. Westminster Tower Rua Cincinato Braga, 340 nº 71 Level 26
Suite 1050 3 Albert Embankment Edificio Delta Plaza 44 Market Street
Chicago, IL 60602 London SE1 7SP Bairro Bela Vista - São Paulo - SP Sydney NSW 2000, Australia
CEP: 01333-010 - BRASIL
P: 312.873.7500 P: +44 (0) 845 456 9611 P: +61 2 9089 8870
F: 312.443.8028 F: +44 (0) 845 456 9612 P: +55 (11) 4064-6101 F: +61 2 9089 8989

Copyright © 2012 Trustwave Holdings, Inc.

All rights reserved. This document is protected by copyright and any distribution, reproduction, copying, or decompilation is strictly prohibited without the prior written consent of
Trustwave. No part of this document may be reproduced in any form or by any means without the prior written authorization of Trustwave. While every precaution has been taken in
the preparation of this document, Trustwave assumes no responsibility for errors or omissions.

Trustwave and Trustwave’s SpiderLabs names and logos are trademarks of Trustwave. Such trademarks shall not be used, copied or disseminated in any manner without the prior
written permission of Trustwave.