Abstract—Cloud computing, an emerging computing
paradigm, is still in its stage of infancy. It can be considered as
an amalgam of a variety of legacy and evolving technologies
and concepts such as distributed computing, grid computing,
virtualization, service-oriented architecture, software-as-a-
service etc to mention a few. It aims at optimally utilizing the
distributed, dynamically-scalable and often virtualized
resources such as processors, storage and software, which are
made available to the users as services over a network, on a
pay-per-use basis. These services can be easily accessed by the
user through an interface, as simple as a browser. But as the
technologies are in a state of flux, developing the software
services in the cloud directly using existing technologies will
make them obsolete as and when new technologies evolve.
Model-driven Architecture (MDA), an initiative by Object
Management Group (OMG), is an open, vendor-neutral
approach that supports the complete software development
process with the modeling activity. In this paper, the authors
propose to leverage Model-Driven Architecture to specify the
software services in the cloud using platform-independent
model (PIM), from which one or more platform-specific
models (PSM) can be derived using automated transformation
tools. The software services thus yielded will be more robust,
flexible and agile in the wake of changing technologies.
Index Terms—Cloud computing, cloud SAAS, model-driven
architecture (MDA), platform independent model (PIM),
platform specific model (PSM).
I. INTRODUCTION
Cloud computing is an emerging computing model that
came into existence at the onset of this century with the
advent of Amazon’s Web based services. Some other big
vendors who gradually joined the fray include Yahoo,
Google, Microsoft, IBM, Sun, Intel, Oracle and Adobe. A
cloud, in the given context, refers to a complex, internet-
based infrastructure of hardware and software components
which include processors, storage, applications,
development environments etc., and are available remotely
as a service. Accordingly, the cloud services are offered at
various levels – software, platform, infrastructure and
hardware. The Internet serves as a medium to transmit data
and/or information between the user and the cloud. The user
may gain access to the service from anywhere and at
anytime, using a thin client or a laptop or even a mobile
phone, through an interface. (usually the browser), on an
on-demand pay-per-use basis.
The resources in a cloud are dynamically allocated and
Manuscript received January 25, 2012; revised March 5, 2012.
R. Sharma is a Research Scholar at the Department of Computer
Science, Himachal Pradesh University, Summer Hill, Shimla, India. (email:
rituchetan@gmail.com)
M. Sood is an Associate Professor at the Department of Computer
Science, Himachal Pradesh University, Summer Hill, Shimla, Himachal
Pradesh, India. (email: soodm_67@yahoo.com)
reallocated on a sharing basis and hence are optimally
utilized and at the same time are very economical.
With the rapid advancements in the field of Information
and Communications Technology (ICT), there has been a
growing significance of model-based software development
solutions. Model Driven architecture (MDA), a recent
model-driven approach in software development offers the
advantage of developing a platform-independent model
(PIM) of the system and transforming it into any number of
platform-specific models (PSMs), one for each platform or
technology in which the final system would be deployed,
and which eventually can be transformed into code. The
transformations are performed using transformation tools
developed for the purpose.
The MDA approach may be leveraged in developing the
software applications that would run in the cloud as
software services. The PIM of the cloud application would
reflect the structure, behaviour and functionality of the
application irrespective of the technology used for its
implementation. The PSM would be more implementation-
oriented and bound to a given execution platform. The
transformations from the PIM to PSM would be carried out
using (semi)automated transformation tools.
This paper is an attempt to model the convergence of
software development principles of MDA with
implementation of cloud software services. Section II
highlights the fundamentals of cloud computing and its
enabling technologies. Section III briefly discusses the basic
concepts of MDA. Based on MDA approach, section IV
illustrates the computation-independent model, platform-
independent model and platform-specific model of a cloud
software application taken as example. Section V draws the
conclusion of the paper and the future work being carried
out by the authors.
II. C
LOUD COMPUTING AND ENABLING TECHNOLOGIES
Cloud computing owes its evolution to a number of
relevant legacy technologies and concepts such as –
distributed computing and storage, grid computing, Service
Oriented Architecture (SOA), virtualization, Software-as-a-
Service (SaaS), Web services and supporting technologies
mainly SOAP (Simple Object Access Protocol), WSDL
(Web Service Definition Language) and UDDI (Universal
Description Discovery and Integration).
The massive computing capabilities of the clouds are
powered by computer grids (or grid computing
environments) having coordinated resources and open
standard protocols and frameworks. Cloud computing
differs from grid computing in the sense that a cloud is
managed by a computational grid but the reverse is not true
[1]. The grids do not rely on virtualization as much as
clouds do. Also, the grids are more suitable for batch
computing while clouds are suitable for interactive
Modeling Cloud Software-As-A-Service: A Perspective
Ritu Sharma and Manu Sood
International Journal of Information and Electronics Engineering, Vol. 2, No. 2, March 2012
238
computing [2].
Virtualization technology enables abstraction of logical
resources from the underlying physical resources. In cloud
computing this technology facilitates the dynamic allocation
and reallocation of resources from a shared pool, to cloud
service consumers. In order to meet a sudden increase in
demand the existing resources are diverted from a low-
priority application to a high-priority one [3]. This, in turn,
helps to minimize the need for additional investment in
infrastructure. Furthermore, it enables the cloud to give an
illusion of the existence of infinite resources to its
customers accessing the services.
Service Oriented Architecture (SOA) is an architectural
paradigm wherein the services (distinct units of logic)
available in a network such as the web, are used as the core
components in building the logic of software applications.
Each service is characterized by a well-defined interface
which exposes the functionality provided by the service.
Individually, the services may be distributed. The key
principles that govern individual services in SOA are
abstraction, loose-coupling, reusability, composability,
statelessness, autonomy, discoverability and adherence to a
service contract [4]. Cloud based systems, in order to be
more effective and efficient, must be built on sound SOA
principles. The cloud should not be looked at as a new
architecture but instead as another option of storing and
running services within SOA [5].
Software-as-a-Service (SaaS) is a software application
based on multi-tenant model. It is owned and managed
remotely by one or more service provider(s) and delivered
to the customers over a network on a shared, pay-per-use or
subscription basis. A SaaS application must be elastic and
scalable in order to qualify as a true cloud computing
service.
Having evolved from other similar computing paradigms
such as SaaS, SOA and Web services, the cloud software
services are simple, scalable, elastic, autonomous, readily
available, easily accessible and service-oriented. Cloud
computing aims at optimizing the usage of distributed and
virtualized resources to achieve higher throughput and to
tackle large scale computation problems. The services
offered in the cloud are configured dynamically and the
usage is metered. The service interfaces are location-
independent and user-centric. The service level agreement
(SLA) between the service provider and the service
consumer ensures Quality of Service (QoS).
Although cloud computing has been extensively defined
by various authors and researchers, a comprehensive
definition is provided by US National Institute of Standards
and Technology (NIST). The essential characteristics of
cloud computing are – 1) On-demand self-service, 2) Broad
network access, 3) Location independent resource pooling,
4) Rapid elasticity, and 5) Measured Service [6].
A cloud service may be delivered by a private cloud, a
public cloud or a hybrid cloud depending on whether the
resources are provisioned and managed internally within the
organization, or externally by a third party on a shared basis,
or both i.e. some resources internally while others
externally.
Cloud computing is an umbrella term that encapsulates
various services such as – Software-as-a Service (SaaS),
Platform-as-a-Service (PaaS), Infrastructure-as-a Service
(IaaS), Hardware-as-a-Service (HaaS). In the absence of a
standard taxonomy, other suggested categories include
Development, Database and Desktop as a Service (DaaS),
Business as a Service (BaaS), Framework as a Service
(FaaS), Organization as a Service (OaaS) etc [2][6][7].
Some key benefits of cloud computing are:
• The ability to access the resources in the cloud from
anywhere and at any time.
• It minimizes the need for advanced hardware and
software on client side thus eliminating the need for
additional physical space and also minimizing the
capital expense.
• Clients can achieve supercomputing powers from
minimal resources.
• It eliminates the cost incurred on hiring trained
professionals or upgrading the skills of manpower
managing and maintaining the related infrastructure.
• It eliminates complexities of buying, configuring and
managing the resources needed to build and deploy
applications which in turn are delivered as a service.
•
III. M
ODEL-DRIVEN ARCHITECTURE (MDA)
MDA®, an initiative by OMG®, is an open, vendor
neutral approach [8] to enterprise application development.
It shifts the focus of software development from the
solution domain to problem domain thereby bridging the
gap which exists between domain-specific concepts and the
programming technologies used to implement them; and
enhances the rigor, productivity and manageability of
software development process.
The software development process, in MDA, is driven by
the activity of modeling the software system. The models
are the prime artifacts and are used throughout the process
of software development. During the development process
these formal, domain-specific models are transformed into
implementation code using automated transformation tools.
The models in MDA are defined at three levels of
abstraction – 1) Computation Independent Model (CIM) or
the domain model bridges the gap between the domain
experts and system experts. It specifies the business logic of
the application, which remains the same irrespective of
whether the system is ICT based or otherwise. 2) Platform
Independent Model (PIM) specifies the functionality of the
ICT-based system independent of the platform that would
be used for its implementation. 3) Platform Specific Model
(PSM) describes the system with respect to the specific
platform on which it would finally be implemented. Miller
et al in [9] define a platform as “a set of
subsystems/technologies that provide a coherent set of
functionality through interfaces and specified usage patterns
which any subsystem that depends on the platform can use
without concern for the details of how the functionality
provided by the platform is implemented”.
A complete MDA specification consists of a definitive
platform-independent model (PIM), plus one or more
platform-specific models (PSM), sets of interface
definitions, each describing how the base model is
implemented on a different middleware platform and sets of
International Journal of Information and Electronics Engineering, Vol. 2, No. 2, March 2012
239
transformation definitions. The PIM depicting the structure,
behaviour and functionality is modeled only once. And then,
the transformation definitions enable the transformation of
the PIM to one or more PSMs.
The key to the success of MDA lies in automating the
model-to-model and model-to-code transformations. A
transformation is defined as a process of the automatic
generation of a target model from a source model, according
to a transformation definition. A transformation definition is
a set of transformation rules that together describe how a
model in the source language can be transformed into a
model in the target language. A transformation rule is a
description of how one or more constructs in the source
language can be transformed into one or more constructs in
the target language [10].
As this approach focuses primarily on the functionality
and behaviour of the software system, and not the
technology in which it will be implemented, it is not
necessary to repeat the process of modeling system’s
functionality and behaviour each time a new technology
comes along.
The three primary goals of MDA are interoperability,
reusability and portability through architectural separation
of concerns [9].
The key benefits of MDA include – improved quality of
software design and implementation, reusability, enhanced
productivity, portability, interoperability, improved Return
on Investment (ROI), reduced cost and reduced time to
market.
IV. A
CLOUD SAAS – CIM, PIM AND PSM
Cloud SaaS are web-based software applications that run
remotely on the Internet and are provided as services to the
consumers on an on-demand pay-per-use basis. These
applications in the cloud may be as simple as a time zone
converter performing a single discrete function, or as
complex as a Customer Relationship Management (CRM)
system performing a set of related business functions. As is
evident, the technologies are constantly evolving. Rather
than directly developing these software services using
available technologies, modeling them at a higher level of
abstraction will decouple them from the undesired effects of
technology change and enhance their longevity.
Therefore, the MDA-based development of cloud SaaS
(application) is going to play a significant role in improving
the quality of cloud software services, making them more
robust, flexible and agile. MDA will enable defining these
cloud software services in a technology-independent
manner. Encapsulating business logic in a manner that is
independent of the technical mechanisms will formally
capture the essence of the applications; and will also make it
possible to reuse them in a variety of contexts [11].
Keeping this in mind, the authors propose leveraging
MDA to develop Platform-Independent Model (PIM) of a
cloud SaaS (a software application in the cloud) and specify
its functionality, structure and behaviour independent of the
platform used for its implementation. Since the PIM would
be less tied to underlying technologies, it needs to be
modeled only once. From this PIM, one or more technology
specific artifacts, the Platform-Specific Models (PSMs),
could be derived (semi)automatically using open or
proprietary model-to-model transformation tools. The PIM
and PSM would represent the logic of the application,
independent of the implementation technology used to
expose it as a web service, such as WSDL, Hyper Text
Markup Language (HTML), Wireless Access Protocol
(WAP), etc. Eventually, the PSMs may be used for
automated code generation for specific technology
platforms.
We illustrate our approach using an example of a
software application running as a service in the cloud – the
Credit Card Validation System (CCVS). The CCVS may be
accessed by anyone connected to the Internet, through an
interface as simple as a browser. For instance, a customer
may purchase products or services from an online/offline
store and make payment through his credit card. In this
situation, the CCVS may be accessed for validating the card.
We assume a simplified approach to credit card
validation where the various steps in the business process
may be listed as:
• The consumer purchases goods or services from a
merchant (online/offline) through his credit card.
• In case of online shopping, the credit card details are
submitted at the customer’s browser, from where they
are securely transmitted to the merchant site and
finally to the CCVS. In case of an offline merchant
store, the credit card details are submitted to CCVS
through the Electronic Funds Transfer Point of Sale
(EFTPOS) terminal by swiping the card.
• Once received by the CCVS, the credit card
information and the bill amount is sent to the
Clearing House.
• The Clearing House then submits the transaction to
the bank that issued the card.
• The bank verifies the card information and transaction
amount, and sends a message to the Clearing House,
approving or declining the transaction.
• The Clearing House in turn sends the message
(approve/decline) to the merchant.
• In case the transaction is approved, the bill amount is
reimbursed to the merchant by the Clearing House.
• The bank pays the amount to the Clearing House later
on.
• The bank deducts the transaction amount from the
credit available to the customer.
• The bank, in turn, receives the payment from the
cardholder at a later date.
• Any processing fee charged by the Clearing House or
the card issuing bank is recorded on the card
statement as expense on part of the customer.
Although MDA does not restrict itself to Unified
Modeling Language (UML) for modeling the system, we
are using UML for the purpose of illustration.
A computation-independent model (CIM) of a software
system may be represented using Use Case diagrams in
UML. A use case diagram capturing the functionality of the
system under consideration is depicted in Fig. 1.
The characteristics of the actors in the system are:
•
Custo
mer is a person who uses the credit card to make
paym
ents for purchases.
International Journal of Information and Electronics Engineering, Vol. 2, No. 2, March 2012
240
• Merchant is a person who receives the payment from
the customer for the goods or services sold.
• Clearing House is a firm that has contract with the
merchant’s bank to clear charges in exchange for a
flat fee and a percentage of every charge processed.
• Bank is the bank that has issued the credit card to the
customer.
Fig. 1. Use Case diagram for CCVS.
As mentioned earlier, the cloud application is web-based;
the PIM design vocabulary describes the attributes and
operations in ways that are entirely independent of XML,
WSDL, SOAP, UDDI, Java, and other Web Service
implementation technologies. Transformation tools can then
be used to generate XML, WSDL, SOAP, UDDI, and the
technology-specific artifacts and finally the implementation
code from the design input [12].
A PIM specifies the system at a higher level of
abstraction as compared to a PSM. Since, it is independent
of any technology to be used for implementing the system,
it can therefore be reused across several different platforms.
The class diagram for a Platform-Independent Model
(PIM) of the application under consideration is depicted in
Fig.2. The PIM, here, defines the static aspects of the
application through a static view. It shows the various
classes (entities) and the associations among them.
A multiplicity adorns each association. The relevant
attributes and the operations are also listed in the model.
The accessor and mutator operations have been
intentionally excluded from the diagram in order to keep it
simple.
The various classes included in the PIM, for the CCVS,
are – Customer, CreditCard, Merchant, Bank,
ClearingHouse, and Product. The associations among the
classes are defined as under:
• A bank is composite aggregation of customers. A
bank has one or more customers. A customer account
is unique and can belong to only one Bank.
• A customer may have more than one account in a
bank.
• Also, a customer can have different accounts in
different banks.
• A bank may issue one or more credit cards, but a
credit card must belong to only one Bank.
• A customer may have one or more credit cards, but a
credit card belongs to only one customer.
• A customer may purchase from one or more
merchants. Also, a merchant may sell products to one
or more customers.
• A merchant may deal with one or more clearing
houses. A clearing house will deal with one or more
merchants.
• A bank may deal with one or more clearing houses. A
clearing house may deal with one or more banks.
• A customer may purchase one or more products. A
product may be purchased by one or more customers.
• A product can be purchased on one credit card. A
credit card can be used to purchase more than one
product.
• A clearing house accepts more than one credit card. A
credit card is presented at only one clearing house.
•
Fig. 2. PIM with associations
For the sake of simplicity, the following assumptions
have been made by the authors, with regard to the system
under consideration:
• An account in a bank can be owned by only one
person.
• A credit card can be processed by only one clearing
house.
• A customer cannot own a credit card that is not issued
by a bank.
The PSM is derived from the PIM using automated
transformation tools. It contains the technology specific
details for the target platform. The Platform-Specific Model
(PSM) for the application under consideration is depicted in
Figure 3 using a class diagram. It is targeted specifically on
Java platform. The PSM also depicts the classes, the
associations among the classes and the multiplicities of the
associations in a manner similar to PIM. The two models
differ in the sense that the PIM does not mention the
language-specific (technology-specific) details of the
system such as the data type of the attributes, the list of
International Journal of Information and Electronics Engineering, Vol. 2, No. 2, March 2012
241
parameters with their data types to be passed to the function,
or the return type of the data that would be returned by the
function. In contrast to PIM, the PSM includes these details
as is clear from Fig. 3.
As discussed earlier, a number of PSMs targeted on
different platforms such as CORBA, .NET, J2EE, etc. can
be derived from a single PIM. Since these models differ in
their structure, each having its own platform-specific
constructs, different sets of transformation rules is required
to generate different PSMs. Once the PSM has been
generated, the next step is to generate the implementation
code from it. Finally, the software (code) obtained is
deployed in specific environments.
Fig. 3. PSM targeted on java.
V. CONCLUSION AND FUTURE WORK
Cloud computing is leading into a future where the
enterprise applications would run on centralized facilities
managed and controlled by third-party compute and storage
resources, instead of local computers. Besides, the
technologies are evolving at a tremendous pace. With the
emergence of newer technologies, the older ones may
become obsolete and/or get replaced. Consequently, a
technology-specific development of these software
applications running in the cloud is not viable in the long
run.
The illustration of the PIM and PSM of a cloud SaaS
taken as example in this paper reinforces that developing
software applications in the cloud in a manner that is
independent of the specific technologies, using MDA, will
enable to reap the benefits of MDA based software
development. Above all, it will enhance the rigor, longevity
and reusability of the cloud service developed thus.
The authors are presently developing a transformation
tool that would transform the CCVS cloud SaaS PIM into
its PSM targeted on Java platform. A set of transformation
rules has been defined by the authors to transform the PIM
constructs into their respective PSM constructs. In addition,
an SOA-based approach for the development of the
illustrated cloud application is also underway.
R
EFERENCES
[1] F. Maria Aymerich, G. Fenu, and S. Surcis. “An Approach to a Cloud
Computing Network.” In First International Conference on the
Applications of Digital Information and Web Technologies, 2008,
ICADIWT.2008 pages 113–118, August 2008, ISBN: 978-1-4244-
2623-2, doi: 10.1109/ICADIWT.2008.4664329 .2008
[2] I. Foster, Y. Zhao, I. Raicu, and S. Lu. “Cloud Computing and Grid
Computing 360-Degree Compared.” In IEEE Grid Computing
Environments Workshop, 2008. GCE ’08 (Nov 2008), pages 1–10,
November 2008.
[3] R. Mikkilineni and V. Sarathy, “Cloud Computing and the Lessons
from the Past,” in Proc, 18th IEEE International Workshop on
Enabling Technologies: Infrastructures for Collaborative Enterprises,
2009, Pages:57-62 , ISBN ~ ISSN:1524-4547 , 978-0-7695-3683-5
[4] T. Erl, “Service Oriented Architecture: Concepts, Technology and
Design,” ISBN 978-81-317-1490-4, Pearson Education, Inc. 2005.
[5] “Cloud Computing and SOA Convergence in Your Enterprise,”
SearchSOA.com, 20 Nov 2009 [Online] Available:
http://searchsoa.techtarget.com/generic/0,295582,sid26_gci1375000_
mem1,00.html .2009.
[6] P. Mell and T. Grance, “The NIST Definition of Cloud Computing”,
Version 15, 10-7-09, [Online] Available:
http://thecloudtutorial.com/nistcloudcomputingdefinition.html
[7] B. P. Rimal, E. Choi, and I. Lumb, “A Taxonomy and Survey of
Cloud Computing systems”, In Proc. Proceedings of the 2009 Fifth
International Joint Conference on INC, IMS and IDC, 2009, Pages:
44-51.
[8] OMG Model Driven Architecture. [Online] Available:
http://www.omg.org/mda/
[9] J. Miller and J. Mukerji, “MDA Guide Version 1.0.1” [Online]
Available: http://www.omg.org/docs/omg/03-06-01.pdf
[10] A. Kleppe, J. Warmer, and W. Bast, “MDA Explained: The Model
Driven Architecture: Practice and Promise,” 2003, Addison-Wesley
Longman Publishing Co. Inc
[11] D. Frankel and J. Parodi, “Using MDA to develop Web Services,”
IONA Technologies PLC, Second Edition, April, 2002 [Online]
Available: http://www.iona.com/archwebservice/WSMDA.pdf
[12] S. David Frankel, Model Driven Architecture: Applying MDA to
Enterprise Computing” 1st ed., Wiley Publishing Inc, 2003.
International Journal of Information and Electronics Engineering, Vol. 2, No. 2, March 2012
242