Use of cloud federation without need of identity federation using dynamic access control

IJRET: International Journal of Research in Engineering and Technology eISSN:2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 03 Issue: 11 | Nov-2014, Available @ http://www.ijret.org 188
USE OF CLOUD FEDERATION WITHOUT NEED OF IDENTITY
FEDERATION USING DYNAMIC ACCESS CONTROL
Rajkumar B Patil1
1
Associate Professor, CSE Dept BKIT, Bhalki
Abstract
Cloud computing is a computing paradigm, where a large pool of systems are connected in private or public networks, to provide
dynamically scalable infrastructure for application, data and file storage. With the advent of this technology, the cost of
computation, application hosting, content storage and delivery is reduced significantly. Cloud Computing is already a successful
paradigm for distributed computing In highly distributed, dynamic and heterogeneous environments, traditional access control
models present problems, such as: scalability, flexibility and the use of static policies many problems still linger in the application
of this model and some new ideas are emerging to help leverage its features even further. One of these ideas is the cloud
federation, which is a way of aggregating different clouds enable the sharing of resources and increase scalability and
availability. One of the great challenges in the deployment of cloud federations is Identity and Access Management. This issue is
usually solved by the creation of identity federations, but this approach is not optimal. In this paper, we propose an access control
system for a highly scalable cloud federation. The presented system is dynamic and risk-based, allowing the use of cloud
federations without the need of identity federations.
Keywords- Distributed, cloud computing, access control, risk, cloud Federation.
-------------------------------------------------------------------***-------------------------------------------------------------------
1. INTRODUCTION
Cloud computing is computing in which large groups of
remote servers are networked to allow centralized data
storage and online access to computer services or resources.
Cloud computing relies on sharing of resources to achieve
coherence and economies of scale, similar to a utility (like
the electricity grid) over a network.[1]
At the foundation of
cloud computing is the broader concept of converged
infrastructure and shared services.
Cloud computing, or in simpler shorthand just "the cloud",
also focuses on maximizing the effectiveness of the shared
resources. Cloud resources are usually not only shared by
multiple users but are also dynamically reallocated per
demand. As cloud computing grows in popularity, new ideas
and models are developed to exploit even further its full
capacity, increasing efficiency and scalability. One of these
ideas is the deployment of cloud federations [2, 3]. A cloud
federation is an association among different Cloud Service
Providers (CSPs) with the goal of sharing data and resources
[4]. Cloud federation is the practice of interconnecting the
cloud computing environments of two or more service
providers for the purpose of load balancing traffic and
accommodating spikes in demand. Cloud federation requires
one provider to wholesale or rent computing resources to
another cloud provider. Those resources become a
temporary or permanent extension of the buyer's cloud
computing environment, depending on specific federation
agreement between providers. However, to make such a
scenario feasible it is necessary to develop authentication
and authorization models for largely distributed, dynamic
and heterogeneous environments. This problem is usually
treated by the deployment of identity federations. An
identity federation is a model of identity management where
identity providers and service providers share users’
identities inside a circle of trust. This solution, nevertheless,
is not optimal, since identity federations present problems
such as the necessity of attribute and trust agreements,
interoperability issues and, in practice, show limited
scalability [5]. This paper shows that it is possible to provide
authorization in cloud federations without the need for an
identity federation. The difference between cloud
federations and identity federations is that cloud federations
are built to share resources and identity federations are built
to share users and identity information. In this paper, we
propose to use a risk-based dynamic access control to enable
authorization in a cloud federation without the necessity, but
allowing the possibility, of using identity federations.
There are two main kinds of work which are related to this
paper: those which study cloud federations and authorization
in these scenarios and those which propose dynamic access
control models. CLEVER Clouds [6, 7, 8] is a “horizontal
federation” model, built on top of a component called Cross
Cloud Federation Manager (CCFM), responsible for the
discovery of clouds in the federation, finding the best match
for resource requests and handling authentication. Based on
this architecture, there is the proposal of using federated
identity management with a third party identity provider to
handle authentication and authorization [9]. The Contrail
project [10] is a framework for the construction of cloud
federations. It is built upon a set of core components: the
Virtual Execution Platform (VEP), the XtreemFS and the
Cloud Federation. Contrail is a big project funded by the
European Union and is under active development. It also
uses federated identity management and provides support
for eXtensible Access Control Markup Language (XACML)

_______________________________________________________________________________________
authorization and the Usage Control (UCON) access control
model. Some challenges for access control in highly
distributed environments are presented in [13], which
compares the Attribute-based Access Control (ABAC),
UCON and Riskadaptive Access Control (RAdAC) models.
2. CLOUD FEDERATION
The cloud computing paradigm has reached a relative
success due to its well-known advantages in scalability and
cost reduction, but to enable its full potential we must step
forward towards cloud federations. As seen in Section II,
there are several proposals of architectures for cloud
federations in the literature, but they all share a common
goal of aggregating different clouds through standard
protocols, enabling their interaction and the sharing of
resources available in each one. Cloud federation comprises
services from different providers aggregated in a single pool
supporting resource migration, resource redundancy and
combination of complementary resources or services [4].
The main benefits of this new approach are an increase in
scalability, availability and interoperability. It also helps in
reducing costs of single providers, since the workload may
be shared among the members of the federation.
Thinking even further, there are already proposals for an
Intercloud, which is a global aggregate of clouds, such as
the Internet is a global aggregate of networks. The
establishment of cloud federations presents challenges such
as the definition of standard protocols and the migration of
virtual resources among diverse providers, but the focus of
this work is in the security aspects of the federations,
especially Identity and Access Management. Cloud security
is a challenge, since providing availability, integrity and
confidentiality for a huge number of users and resources in
an Internet-accessible environment is not easy. Cloud
federations tend to increase concerns because of the increase
in the number of users and resources, the use of different
protocols and the exchange of sensitive data among
providers. Issues such as governance, auditing and risk
management are being actively researched for One of the
most important issues in the establishment and running of a
cloud federation is Identity and Access Management (IAM).
When in a single cloud, it is possible to use traditional IAM
procedures and authorization models to handle access
control because all of the users and resources are within the
same security domain. When resources and subjects a
rescaled to a federation of clouds, nevertheless, there is the
concern with the fact that subjects may come from a
different security domain than the resource to which they are
requesting access.
To implement authorization using models such as Role-
Based Access Control (RBAC) or Attribute-Based Access
Control (ABAC), the cloud must use information provided
by a system about a user. This information may be, for
instance, the user's identity or attributes of this identity, such
as name, organizational role and date of birth. For a cloud to
trust the identity or attribute information of a user that
comes from another cloud, both clouds must share some
agreement of trust. That is why this process I commonly
mediated by an identity federation. With Federated Identity
Management (FIM), every participant of the federation is
expected to agree that the information received by another
participant is correct, in what is called a Circle of Trust
(CoT). eived by another participant is correct, in what is
called a Circle of Trust (CoT).
A problem with this approach is the fact that this agreement
requires previous negotiation, which may be an extensive
process and hinder dynamic collaboration. Dynamic
collaboration is achieved when entities which have a need to
collaborate can instantly form a federation, without the need
for a previous trust agreement.
Another problem faced by identity federations is the
extensive number of protocols and standards, which actually
reduces interoperability. Federations tend to get bigger and
bigger and users may participate in different f derations. All
of those facts combined lead to, in practice, a limited
scalability of identity federations, reducing their
effectiveness in real world.
3. RISK-BASED ACCESS CONTROL
Traditional access control models employ static
authorization, i.e., every decision is pre-established, based
on the policies The idea behind dynamic access control is
that the access requests must be analyzed taking into
account contextual and environmental information such as
security risk, operational need, benefit and others Real
applications may require the violation of security policies,
and the support for exceptional access requests is known as
“break the glass”
Identity and access management encompasses several
processes related to the identification, authentication,
authorization and accountability of users in computer
systems. Authorization or access control is the process
through which a system guarantees that access requests are
validated using well-established rules.
These rules are known as policies and the way through
which the policies are enforced together with the
mechanisms used in this enforcement is known as an access
control model. Classical access control models are known to
present problems in highly distributed and dynamic
environments], especially scalability and flexibility
limitations and the use of static policies. Role-based models,
for instance, lack granularity of control, because roles share
their permissions with every user they are attributed to. To
enable more flexible access control decisions, which reflect
current needs for information sharing and allow for a secure
handling of exceptional requests, dynamic access control
models were developed.
In contrast with classical models, dynamic access control
has the characteristic of using more than predefined policies
to compute access decisions. These models are based on
dynamic characteristics, which are assessed in “real time” as
the subject requests access to a resource. Characteristics

_______________________________________________________________________________________
such as trust, context, history and risk are often used to
reach decisions, and exactly which characteristics to use and
how to measure them is discussed in several works.
Risk-based access control models are often used as a “break-
the-glass” mechanism, allowing for exceptional access
requests to be handled by the system more effectively than
simply granting full access . Exceptional requests and
special access are sometimes necessary in medical and
military applications, among others. A well-known example
is in a healthcare facility where only doctors have access to
patients’ histories, but in the case of an emergency, a nurse
may need to access this information to save a patient’s life.
If this kind of situation was not predicted in any policy,
either the nurse won’t be able to perform his/her duty or the
nurse may be given a doctor’s access, which may grant a
broader access than the necessary in this case, allowing
misuse. In either case, it represents a greater risk to the
system than if a dynamic access control system were used
and the access control needs were evaluated per request.
Granting special access in exceptional cases usually
involves some form of monitoring by the system. It may be
in the form of: obligations, which are post-conditions that a
user must fulfill in order to keep his or her access right ; a
reputation system, which logs users' actions and assigns
rewards and penalties to them ; or a market system, in which
users have a limited amount of points that may be used to
“buy” exceptional accesses.
Fig 1 Risk-based access control overview
The figure is based on common points found in diverse
models, and the main elements present are the subject, the
resource and the risk estimation engine. The subject tries to
access a resource by issuing an access request, which is then
processed by a risk estimation engine that uses all the
information it deems necessary to come to a decision.
Usually there is a risk threshold defined by thsystem
administrators, and if the risk is lower than this threshold,
access is granted. Other variations measure risk versus
benefit of an access, and decide based on which one is
Greater.
4. PROPOSED ARCHITECTURE
In this paper, we propose that it is possible to provide a way
to establish cloud federations without the need for identity
federations, by using risk-based access control and relying
on the authentication provided by each cloud. This can
increase the scalability of this model and handle exceptional
requests.
4.1 Cloud Federations
Fig. 2 presents an overview of the cloud federation
architecture that we are considering. This architecture is
based on the common points found in the main federation
projects currently being developed, some of which were
described in Section II.
The main application scenarios for such federations are
medical, military and scientific collaborations, which
require large storage and processing capabilities, as well as
efficient information sharing. In this architecture we have
the following components:
CloudProvider: this is the Cloud Service Provider (CSP)
itself, who provides the infrastructure over which the virtual
resources are allocated (they are represented by the clouds in
the figure);
CloudManager: responsible for attaching a
CloudProvider to the federation. It is composed of several
services that deal with users, resources, policies, service-
level agreements, security and the CloudProvider. It is
modular so that it can be attached to different cloud
management software just by changing one of its services.
FederationManager: responsible for coordinating the
federation. It acts as a naming service and is also responsible
for message passing.
Fig 2 Overview of the federation

_______________________________________________________________________________________
4.2 Access Control
As shown in Fig. 2, some of the participating clouds may
form identity federations among themselves. Under the
point of view of a user there are two types of clouds in this
architecture: a home cloud (the user’s original CSP) and
foreign clouds (the other clouds in the federation). Users can
deploy and access resources in both types of cloud, but
access control behaves differently for each case. When users
deploy a resource in their home cloud they may choose if it
will be available for users of foreign clouds. In any case the
user must upload an XACML policy file together with the
resource, which will be used for ABAC. Users may also
deploy resources in a foreign cloud and it will automatically
be available to every user of the federation. Finally, users
may access resources in their home cloud or shared
resources in foreign clouds. When a user tries to access a
resource in their home cloud, this request is handled by a
classical ABAC model. Based on user attributes and
XACML policies the system grants or denies the requested
access.
When a user tries to access a resource in their home cloud,
this request is handled by a classical ABAC model. Based
on user attributes and XACML policies the system grants or
denies the requested access.When a user tries to access a
resource in a foreign cloud, the system first verifies if both
clouds are in an identity federation, in which case the access
will also be handled by ABAC, but if there is not an identity
federation between them, the “break-the-glass” mechanism
is activated and the risk-based access control Policy
Decision Point (PDP) is called.
The PDP is located in the cloud handling the access request
(foreign to the requester) and the metrics and parameters of
risk estimation are defined by the administrators of this
cloud and the users who own the resources. These metrics
are informed in an eXtensible Markup Language (XML)
file, containing definitions of risk metrics and how to
measure and aggregate them, as well as a threshold level for
granting access to the resource and possible obligations that
users will have to follow. This file is known as a risk policy.
Each cloud provider must provide a set of basic metrics with
their quantification rules. Those will be used to create a
baseline risk policy for the provider. This guarantees that a
cloud provider is able to maintain their minimal security
requirements.
Each resource has its own risk policy, which must respect
what is defined in the baseline policy, but may be extended
to become more or less restrictive as the user desires. The
XML file of the policy must be uploaded by users when they
choose to deploy a shared resource. The system does not
generate risk policies on the fly and all the risk policies must
follow a predefined XML schema, so that different clouds
can communicate.
5. CONCLUSIONS AND FUTURE WORK
In this paper, we proposed a risk-based dynamic access
control system to enable cloud federations without the need,
but allowing the possibility, of identity federations. By
eliminating the need for identity federations our proposal
eases the use of cloud federations, since it doesn't depend on
the establishment of agreements and circles of trust, also
enhancing scalability, by avoiding the formation of “identity
islands” [5]. The main contributions of this paper are the
definition of a risk-based access control system for cloud
federations and the proposed use of risk policies in the form
of XML files to allow the use of different risk metrics and
quantification methods that are not necessarily predefined.
The proposal is flexible enough to handle the needs of
acloud federation and the performance evaluations indicate
that it is scalable and that the risk estimation process is not a
big hindrance in the process, especially if the quantification
is performed locally.
REFERENCES
[1]. P. Mell and T. Grance, “The NIST Definition of Cloud
Computing”, 2011.
[2]. E. Carlini, M. Coppola, P. Dazzi, L. Ricci, and G.
Righetti, “Cloud Federations in Contrail”, Proc. Euro-Par
2011: Parallel Processing Workshops, 2012, pp. 159-168
[3]. B. Rochwerger et al., “The reservoir model and
architecture for open federated cloud computing”, IBM J.
Res. Dev., vol. 53, no. 4, July 2009, pp. 535-545
[4]. T. Kurze, M. Klems, D. Bermbach, A. Lenk, S. Tai, and
M. Kunze, “Cloud Federation”, The Second International
Conference on Cloud Computing, GRIDs, and
Virtualization, September 2011, pp. 32-38
[5]. K. Lampropoulos and S. Denazis, “Identity management
directions in future internet”, IEEE Communications
Magazine, vol. 49, no. 12, December 2012, pp. 74-83
[6]. A. Celesti, F. Tusa, M. Villari, and A. Puliafito, “How
to Enhance Cloud Architectures to Enable Cross-
Federation”, Proc. 3rd IEEE International Conference on
Cloud Computing, July 2010, pp.337- 345
[7]. A. Celesti, F. Tusa, M. Villari, and A. Puliafito,
“Security and Cloud Computing: InterCloud Identity
Management Infrastructure”, 19th
IEEE WETICE, June
2010, pp. 263-265
[8]. A. Celesti, F. Tusa, M. Villari, and A. Puliafito, “Three-
Phase Cross- Cloud Federation Model: The Cloud SSO
Authentication”, 2nd
AFIN, July 2010, pp. 94-101
[9]. A. Celesti, F. Tusa, M. Villari, and A. Puliafito,
“Federation Establishment Between CLEVER Clouds
Through a SAML SSO Authentication Profile”,
International Journal on Advances in Internet Technology,
vol. 4, no. 12, 2011, pp.14-27
[10]. M. Coppola et al., “The Contrail approach to cloud
federations”, Proc. ISGC'12, 2012

Use of cloud federation without need of identity federation using dynamic access control

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Use of cloud federation without need of identity federation using dynamic access control (20)

More from eSAT Publishing House (20)

Recently uploaded (20)

Use of cloud federation without need of identity federation using dynamic access control