SECURITY RISK ANALYSIS IN WEB SERVICES SYSTEMS
Carlos Gutiérrez, Eduardo Fernández-Medina, Mario Piattini
ALARCOS Research Group. Information Systems and Technologies Department UCLM-Soluziona Research and
Development Institute. University of Castilla-La Mancha Paseo de la Universidad, 4 – 13071 Ciudad Real, Spain
Keywords: Security Risk Analysis and Management, Security Engineering, Software Security Development Process,
Web Services Security.
Abstract: Nowadays, best practices dictate that security requirements of distributed software-intensive systems should
be based on security risk assessments. Web services-based systems supporting network alliances among
organizations through Internet are such type of systems. In this article we present how we’ve adopted the
risk analysis and management methodology of the Spanish Public Administration, which conforms to ISO
15408 Common Criteria Framework (CCF), to the Process for Web Services Security (PWSSec) developed
by the authors. In addition, a real case study where this adaptation was applied is shown.
1 INTRODUCTION
Nowadays, best practices dictate that security
requirements of software-intensive systems should
be based on risk assessments (Butler and Fischbeck
2005). Software systems based on Web services
(WS) technologies have achieved a great popularity
recently in both industry and academic world. Web
services are a natural consequence of the evolution
of the Web and distributed systems. Since its
beginnings as a way to share and distribute
information on a global scale, effectively becoming
a giant distributed content library, the Web has been
progressively widening its reach to enable more
sophisticated forms of interaction between browser
clients and servers: single form-based interactions,
retail ecommerce applications, and more complex
business-to-business interactions. IDC estimates that
$2.3 billion was spent worldwide on total WS
software in 2004, more than double the amount from
the previous year. IDC expects spending to continue
to increase dramatically over the next 5 years,
reaching approximately $14.9 billion by 2009. In
consequence, security in WS development processes
should include a risk analysis so that security
requirements can be elicited and prioritized. In this
paper, we present a risk analysis process on a WS-
based system that is part of the tasks to be developed
during the WSSecReq (Web Services Security
Requirements) subprocess of the PWSSec (Process
for Web Services Security) process created by the
authors (Gutiérrez, Fernández-Medina et al. 2005).
Although WSSecReq subprocess does not demand a
specific risk analysis method we show how the risk
analysis and management method of the Spanish
Public Administration, Magerit2 (Crespo, Gómez et
al. 2005), is applied to a real case study. MAGERIT
2 is the Spanish Public Administration's adaptation
of ISO 15408, Common Criteria Framework.
The rest of the article is organized as follows: i) in
section 2, a little background on those terms the rest
of the article is based on is presented. That is, a brief
explanation about the PWSSec process, a short
introduction on its WSSecReq subprocess, and,
finally, a short presentation of the case study that
section 3 is based on (see (Gutiérrez, Fernández-
Medina et al. 2005)) for more details on the case
study’s context); ii) in section 3, we will explain
how we have adopted Magerit2 methodology when
performing the tasks related to risk analysis defined
by the WSSecReq subprocess; iii) in section 4, final
conclusions are stated.
425
Gutiérrez C., Fernández-Medina E. and Piattini M. (2006).
SECURITY RISK ANALYSIS IN WEB SERVICES SYSTEMS.
In Proceedings of the International Conference on Security and Cryptography, pages 425-430
DOI: 10.5220/0002105004250430
Copyright
c
SciTePress
Figure 1: Activities and taks of the WSSecReq subprocess.
2 BACKGROUND
2.1 PWSSec Overview
The PWSSec process specifies how to define
security requirements for WS-based systems,
describes a security services-based reference
security architecture and explains how to instantiate
it to obtain concrete security architecture based on
the current WS security standards (Gutiérrez,
Fernández-Medina et al. 2005). PWSSec process
is structured in three sub-processes which describe
their inputs, outputs, activities, actors and
sometimes, guides, best practices, tools and
techniques that complement, improve and facilitate
the set of activities and tasks developed within these
stages. WSSecReq sub-process’s main purpose is to
produce, by means of a systematic approach, a
specification (or a part of it) of the security
requirements of the WS-based system. WSSecArch
sub-process is aimed at allocating the security
requirements specified in the previous section to a
WS-based security architecture. This security
architecture will be equipped with the necessary
security policies and architectural mechanisms to
achieve the considered security requirements.
WSSecTech subprocess’s main objective is to
identify the set of WS-based security standards that
will implement the architectural security
mechanisms identified in the previous stage.
2.2 WSSecReq Overview
The main purpose of this subprocess is to produce a
specification (or a part of it) of the security
requirements of the target WS-based system. Its
input is composed by a specification of the scope
that we want to comprise during the current
iteration, the business and security goals defined for
the system as well as the part of the organizational
security policy that we estimate that may impact on
the system design. The output is basically formed
by: i) A threat attack tree (Schneier 1999) associated
with the WS business and application pattern
(Endrei, Ang et al. 2004) identified within the
analyzed functionality; ii) Every built attack tree’s
leaf will show a threat (WS-I 2005) that can refined
by a set of attack scenarios, defined as misuse cases
according to (Alexander 2003; Sindre and Opdahl
2005), organized into attack profiles (Moore, Ellison
et al. 2001), and represented according to the
Quality of Service UML profile (OMG 2004); ii)
every misuse case must have related a set of security
use cases, according to Donald G. Firesmith
(Firesmith 2003), that state how the system should
respond to the associated misuse case; iii) A formal
specification of the security requirements for the
scope of the system based on SIREN (Toval, Nicolás
et al. 2001) (Gutiérrez, Fernández-Medina et al.
2005). These requirements will have been derived
after instantiating the WS security requirements
templates associated with every security use case.
This subprocess defines 4 main activities:
Elicitiation, Analysis, Specification and
PWSSec Process
Sub-process P1 – WSSecReq
Activity A1.1: Elicitation
Task T1.1.1: Decide granularity level and identify the fragment of functional software whose security will be
analyzed
Task T1.1.2: Identify the IBM WS-based business pattern.
Task T1.1.3: Identify the IBM WS-based application pattern.
Task T1.1.4: Identify possible business threats.
Task T1.1.5: Identify possible application threats.
Task T1.1.6: Relate business and application threats.
Task T1.1.7: Identify and assess threats.
Task T1.1.8: Identify type of attackers and their possible types of attack.
Task T1.1.9: Assess impact of attacks.
Task T1.1.10: Estimate and prioritize security risks.
Task T1.1.11: Determine the behaviour the system should have for each attack.
Task T1.1.12: Identify security sub-factors.
Task T1.1.13: Specify security requirements.
Activity A1.2: Analysis
Activity A1.3: Specification
Activity A1.4: Verification and Validation
SECRYPT 2006 - INTERNATIONAL CONFERENCE ON SECURITY AND CRYPTOGRAPHY
426
Validation and Verification. Here, we will focus
on the Elicitation activity (see (Gutiérrez,
Fernández-Medina et al. 2005) for more details on
the others). The Elicitation activity will be
supported by a detailed study of security for each
WS business service identified and considered in the
current iteration. This activity is inspired in the risk
analysis and management process known as
Operationally Critical Attack, Asset, and
Vulnerability Evaluation
SM
(OCTAVE) (Firesmith
2003). This activity defines a set of tasks that
support security risk analysis during elicitation of
security requirements. In this article we will show
how we have adopted Magerit2, a Common Criteria
Framework-compliant security risk analysis and
management methodology, developed by the
Spanish Public Administration.
2.3 Case Study
In this article we present an actual case study that
was applied to a web services-based system known
as WS-BTS (Web Services-based Bank Transfer
System). This system’s objective was the sale of
certain products chosen by purchasers through a
Web application. Payments are made from
purchaser’s bank account which is associated with
the bank account of the sales organization (hereafter
SalesOrg). This use case was developed as a WS-
based system and consists of two types of WS-based
agents: (1) a WS consumer agent, belonging to
SalesOrg, who will be referred to as WS-
BTSConsumer (Web services-based Bank Transfer
System Consumer) and (2) the WS provider agent of
the bank service (hereafter BankOrg) that will be
referred to as WS-BTSProvider (Web services-based
Bank Transfer System Provider). These agents
interact in order to fulfil a business workflow called
BTS (Bank Transfer System), whose objective is to
assist the final customer during its payment so
purchase is facilitated. This use case is achieved by a
three-step protocol carried out by the WS-
BTSConsumer and WS-BTSProvider web services
agents as described in (Gutiérrez, Fernández-Medina
et al. 2005). In this article we illustrate how risk
analysis was made as part of applying the
WSSecReq subprocess on this case study.
3 RISK ANALYSIS IN WS-BASED
SYSTEMS
In this section, we’ll show how WSSecReq’s tasks
were carried out during the aforementioned case
study. Concretely, we’ll focus on risk analysis-
related tasks. That is, tasks from T1.1.4 to T1.1.10
(see high-lighted tasks in Figure 1). In tasks T1.1.1-
Threat Tree derived of the IBM WS Application Pattern Exposed Direct Connection
ID: A3Ap-CED-1
Goal: 1. Cause damage on the elements defined in the IBM WS Application Pattern named Exposed Direct
Connection when applied to the WS-BTS system.
1.1. Threat WS-BTS Agents
1.1.1 Threat WS-BTSConsumer/WS-BTSProvider
1.1.1.1 Intentional Threats
1.1.1.1.1 Principal Spoofing of WS-BTSConsumer/WS-BTSProvider
1.1.1.1.1.1 Integrity
1.1.1.1.1.2 Confidentiality
1.1.1.1.1.3 Service’s User Authentication
1.1.1.1.1.4 Message Origin Authentication
1.1.1.1.2 Manipulation of Configuration of WS-BTSConsumer/WS-BTSProvider
1.1.1.1.2.1 Integrity
1.1.1.1.2.2 Confidentiality
1.1.1.1.2.3 Service’s User Authentication
1.1.1.1.2.4 Message Origin Authentication
1.1.1.1.2.5 Service Traceability
1.1.1.1.2.6 Message Traceability
1.1.1.1.3 Denial-of-Service to WS-BTSConsumer/WS-BTSProvider
1.1.1.1.3.1 Availability
1.1.1.1.4 Privilege Abuse by WS-BTSConsumer/WS-BTSProvider
1.1.1.1.4.1 Integrity
1.1.1.1.4.2 Confidentiality
1.1.1.1.5 Unforeseen use of WS-BTSConsumer/WS-BTSProvider
1.1.1.1.5.1 Availability
1.1.1.1.6 Re-routing of messages to WS-BTSConsumer/WS-BTSProvider
1.1.1.1.6.1 Integrity
1.1.1.1.6.2 Confidentiality
1.1.1.1.6.3 Service’s User Authentication
1.1.1.1.6.4 Message Origin Authentication
1.1.1.1.7 Non-authorized Access
1.1.1.1.7.1 Integrity
1.1.1.1.7.2 Confidentiality
1.1.1.1.7.3 Message Origin Authentication
1.2. Threat Network Zone
1.3. Threat Connection Rules
Figure 2: Threat Tree derived from the IBM WS Application Pattern Exposed Direct Connection– View of threats on the
existing run-time software systems.
SECURITY RISK ANALYSIS IN WEB SERVICES SYSTEMS
427
T1.1.3, business and application IBM WS-based
architectural patterns were identified (Endrei, Ang et
al. 2004). The novelty of our approach resides in
showing how a risk analysis method conformed to
the Common Criteria Framework was integrated into
PWSSec in such a way that security requirements
and security engineering disciplines for Web
services-based system were successfully aligned,
integrated and developed. Few previous approaches
have been proposed on the subject of applying
security risk analysis in WS-based development
processes up until now. The problem with them is
that they explain how this subject from a very
abstract level of detail (Christopher Steel 2005). In
this paper, we provide a reusable, real and practical
solution on this area showing how we adjusted
Magerit2 to security analysis-related tasks of
PWSSec.
3.1 A1.1. Elicitation - T1.1.4:
Identify Possible Business
Threats
Rigorous risk analysis relies on an understanding of
business impacts, which requires an understanding
of laws and regulations as well as the business
model supported by the software (Verdon and
McGraw 2004). The main purpose of this task is,
from the business-level description elaborated
during task T1.1.2, to define the set of potential
business-level threats that applies to the system
under development. We’ve associated an abstract
business threat tree to every IBM WS business
(Endrei, Ang et al. 2004; Gutiérrez, Fernández-
Medina et al. 2005). This way, once the WS
business pattern has been identified its potential
threats are systematically discovered. These threats
are organized in a tree-like form (Moore, Ellison et
al. 2001). This task’s output is a Business Threat
Model containing the description of the identified
threats organized in the business threat tree. The
chosen notational language representation is based
on the Quality-of-Service UML Profile (OMG
2004).
3.2 A1.1. Elicitation - T1.1.5:
Identify Possible Application
Threats
Risk analysis on modern distributed paradigms such
as WS, requires a functional decomposition of the
application into major components, processes, data
stores, and data communication flows, mapped
against the environment across which the software
will be deployed (Verdon and McGraw 2004). In
this task, the application-level threat tree, which
provides such a functional decomposition, will be
created based on the IBM WS-based application
pattern identified during task T1.1.3 (see Figure 2).
The set of IBM WS application patterns and their
associated abstract threat trees are part of the WS
Security E&A (Elicitation and Analysis) Resources
Repository of WSSecReq subprocess (Gutiérrez,
Fernández-Medina et al. 2005). In Figure 2, the
fragment of the application threat tree that unfolds
branch 1.1 is presented. Under this branch, the set of
threats to be considered on WS agents that
participate in the WS-BTS system: Agent WS-
BTSConsumer (WS-BTSC) and agent WS-
BTSProvider (WS-BTSP) are organized according
to their types. The set of threats on the network
organized under branch 1.2 and 1.3 are omitted due
to space-limits. These threats have been extracted
from the catalogue of threats defined in Magerit2.
Under branch 1.4 the set of threats to be considered
on the WS-based interactions is presented. Here, the
division proposed by the abstract threat tree is based
on the set of threats on the messages of each one of
the interactions that support the functionality whose
security is under analysis (threats have been
extracted from (WS-I 2005) and (Crespo, Gómez et
al. 2005)). This task’s output is an Application
Threat Model. The description of these threats will
give place to a threat model at the application level
that will mainly contain: i) An application threat tree
specific for the system under analysis; ii) UML QoS
model of threats and assets (OMG 2004).
3.3 A1.1. Elicitation - T1.1.7: Threat
Assessment
Task T1.1.7 of WSSecReq is completed by applying
the following Magerit2’s steps: i) Identification of
Assets: According to the application threat tree, and
just focusing on threats on the interactions, the
lowest level assets (those whose risk depends on
higher-level assets) are TNT message (for the
developed branch), TTR Message, TTR Response
Message, RNP Message and RNP Response
Message as well as WS-BTSP and WS-BTSC
agents; ii) Definition of the Dependency Matrix of
Assets: Every (business/application) abstract threat
tree has predefined its own template for its
corresponding asset dependency matrix within the
WS Security E&A Resources WSSecReq’s
repository. The asset dependency matrix allows the
establishment of dependencies between branches
representing assets of the threat tree. The types of
assets considered in a WS context are: a) Web
SECRYPT 2006 - INTERNATIONAL CONFERENCE ON SECURITY AND CRYPTOGRAPHY
428
Services: The purpose of the WS-BTS system is to
offer a service; b) WS agent: From
Magerit2’s viewpoint, we consider it as software
applications; c) Messages: access to data (messages)
is made through WS agents; d) Volatile/Persistent
Structured Storage Services (Databases, directory
services, etc.): It is the base from which certain
messages are created (outgoing messages) and
where the results of processing other messages are
stored (incoming messages); iii) Threat
Characterization: Threat characterization consists
of determining the likelier threats for each one of the
assets and represents them in a System’s Risk Map.
In our case, this step was straightforward since we
just needed to add two new metrics to the
application threat tree: Frequency of Threat
Occurrence and Asset Degradation Ratio. The
Frequency of Occurrence Threat’s value will be
valued during task T1.1.8, when all types of attacks
for each threat are identified and when the highest
frequency of occurrence due to those attacks is
obtained. The asset degradation’s value will be
determined during task T1.1.9 as part of the
calculation of the threat impact. In Table 1, the final
Risk Map (resulting of task T1.1.10) which includes
the set of identified assets is presented. As output
product of this task the Threat Assessment, an
Assessed Global Threat Model consisting of the
aggregation of the security analysis made to the
Global Threat Model is obtained.
3.4 A1.1. Elicitation - T1.1.8:
Identify the Type of Attackers
and their Possible Types of
Attacks
The next step will consist of refining the leaf-nodes
of the threat tree, i.e. further specification of the
threats by means of concrete attacks. Towards this
ends, use will be made of the concept of attack
profile described in (Moore, Ellison et al. 2001). We
use misuse cases in (Sindre and Opdahl 2005) to
defining the sequences of steps which state the
achievement of successful attacks on the system. An
attack profile contains a set of abstract misuse cases
that apply to a reference model defined within the
profile (in our case the IBM WS-based Application
Pattern). Therefore, interactions in every WS-based
application pattern have one attack profile related.
Every WS-based application pattern has one or more
attack profiles related to it which state the potential
attacks that could be targeted at them.
We complete the Assessed Global Model of
Threats with the characterization and frequency of
the attacks that materialize every threat thereby
obtaining the Global Model of Threats and Attacks.
3.5 A1.1. Elicitation - T1.1.9: Assess
Impact of Attacks
In Magerit2 terms, this task will consist of
completing the Risk Map by assigning the value of
degradation on assets as a consequence of threats’
materialization. In addition, the Risk Map is
completed by incorporating an additional value that
represents the accumulated impact on every high-
level asset (WS-BTSProvider/WS-BTSConsumer)
and the repercussed impact on every low-level asset
(WS messages). As output of this task we obtain the
Assessed Global Model of Threats and Attacks
completed with the Risk Map.
Table 1: View of the Risk Map showing degradation ratio, accumulated impact and risk of the WS-BTSC asset. Column F
represents Frequency of the threat.
Security Dimensions
(I=Integrity, C=Confidentiality, A=Availability, S_A=Service’s User Authentication,
M_A=Message Origin Authentication, S_T=Service Traceability, M_T=Message Traceability)
Asset Threat F I C D A_S A_M T_S T_D
1.1.1.1.1.1 5 50% [3] {3} 50% [4] {4} 100%[4] {4} 100% [6] {6}
1.1.1.1.1.2 5 60 % [4] {4} 5% [0] {0} 10% [0] {0} 10% [0] {0} 0% [0] {0} 0% [0] {0}
1.1.1.1.1.3 5 10%[0] {0}
1.1.1.1.1.4 5 0 [0] {0} 0% [0] {0}
1.1.1.1.1.5 5 0%[0] {0}
1.1.1.1.1.6 5 10% [0] {0} 5% [0] {0} 5% [0] {0} 5% [0] {0}
1.1.1.1.1.7 5 0
WS-
BTSC
1.1.1.1.1.8 5 100
%[7] {7} 10% [0] {0} 60% [3] {3}
SECURITY RISK ANALYSIS IN WEB SERVICES SYSTEMS
429
3.6 A1.1. Elicitation - T1.1.10:
Assess and Prioritize Security
Risks
Finally, we estimate and prioritize the risk
completing the Assessed Global Model of Threats
and Attacks. In the case of Magerit2, risk is
computed as a function of the impact and frequency
of the threats. Table 1 shows the computed risks for
every threat and asset and its security dimension.
These risks will guide and provide a basis for the
development of the following tasks defined within
the WSSecReq sub-process. These tasks basically
consist of identifying the expected behaviour of the
system for every attack (task T1.1.11) and eliciting
the security requirement (task T1.1.12). Risks on
every asset will guide what and how resources
should be planned during security architecture
development (in WSSecArch sub-process).
4 CONCLUSIONS
In this paper, we have presented how Magerit2 can
be adapted in the context of the PWSSec process
during elicitation of security requirements within
WS-based systems. This presentation has been
complemented with a demonstration of the
application of the WSSecReq subprocess, one of the
sub-processess defined by the PWSSec process to a
real case study.
ACKNOWLEDGMENTS
This research is part of the following projects
RETISTIC network (TIC2002-12487-E), of
Dirección General de Investigación del Ministerio
de Ciencia y Tecnología, DIMENSIONS (PBC-05-
012-1), financed by the FEDER and the “Consejería
de Ciencia y Tecnología de la Junta de Comunidades
de Castilla-La Mancha” (Spain) and CALIPO
(TIC2003-07804-C05-03) granted by the “Dirección
General de nvestigación del Ministerio de Ciencia y
Tecnología” (Spain).
REFERENCES
Alexander, I. (2003). "Misuse Cases: Use Cases with
Hostile Intent." IEEE Computer Software 20(1): 58-
66.
Butler, S. A. and P. Fischbeck (2005). Multi-Attribute
Risk Assessment. SREIS'05 in conjuntction with
RE'05, Paris, France.
Christopher Steel, R. N., Ray Lai (2005). Core Security
Patterns: Best Practices and Strategies for J2EE™,
Web Services, and Identity Management, Prentice Hall
PTR / Sun Microsystems.
Crespo, F. L., M. Á. A. Gómez, et al. (2005). MAGERIT -
Versión 2. Metodologías de Análisis y Gestión de
Riesgos de los Sistemas de Información. III - Guía de
Técnicas. Madrid, Ministerio de Administraciones
Públicas: 154.
Endrei, M., J. Ang, et al. (2004). Patterns: Service-
Oriented Architecture and Web Services: 345.
Firesmith, D. G. (2003). "Engineering Security
Requirements." Journal of Object Technology 2(1):
53-68.
Firesmith, D. G. (2003). "Security Use Cases." Journal of
Object Technology 2(3): 53-64.
Gutiérrez, C., E. Fernández-Medina, et al. (2005).
PWSSec: Process for Web Services Security. IEEE
International Conference on Web Services 2005,
Orlando, Florida, USA.
Gutiérrez, C., E. Fernández-Medina, et al. (2005). Security
Requirements for Web Services based on SIREN.
Symposium on Requirements Engineering for
Information Security, Paris, France.
Gutiérrez, C., E. Fernández-Medina, et al. (2005). Web
Services Enterprise Security Architecture: a Case
Study. ACM Workshop on Security on Web Services,
Fairfax, Virginia, USA, ACM Press.
Gutiérrez, C., E. Fernández-Medina, et al. (2005). Web
Services-based Security Requirement Elicitation. 1st
International Workshop on Service-Oriented
Computing: Consequences for Engineering
Requirements (SOCCER'05) in conjunction with
IEEE RE'05, Paris, France.
Moore, A. P., R. J. Ellison, et al. (2001). Attack Modelling
for Information Security and Survivability. Survivable
Systems, Software Engineering Institute.
OMG (2004). UML Profile for Modeling Quality of
Service and Fault Tolerance Characteristics and
Mechanisms.
Schneier, B. (1999). "Attack Trees: Modeling Security
Threats." Dr. Dobb's Journal.
Sindre, G. and A. L. Opdahl (2005). "Eliciting Security
Requirements with Misuse Cases." Requirements
Engineering Journal 10(1): 34-44.
Toval, A., J. Nicolás, et al. (2001). "Requirements Reuse
for Improving Information Systems Security: A
Practitioner's Approach." Requirements Engineering
Journal 6(4): 205-219.
Verdon, D. and G. McGraw (2004). Risk Analysis in
Software Design. IEEE Security & Privacy. 2: 79-84.
WS-I (2005). Security Challenges, Threats and
Countermeasures Versión 1.0, WS-I. 2005.
SECRYPT 2006 - INTERNATIONAL CONFERENCE ON SECURITY AND CRYPTOGRAPHY
430