INTEGRATING AGILE AND MODEL-DRIVEN PRACTICES IN A

METHODOLOGICAL FRAMEWORK FOR THE WEB

INFORMATION SYSTEMS DEVELOPMENT

Paloma Cáceres, Esperanza Marcos, Valeria de Castro

Rey Juan Carlos University

Keywords: Agile methodologies, model-driven methodologies, Web engineering.

Abstract: Nowadays, the Web information systems (WIS) development has becom

e an interesting area in both

research and business world. On the one hand, the WIS development needs specific methodologies for

taking into account certain specific aspects for Web systems. Therefore, traditional methodologies do not

facilitate a quick and light development as the agile methodologies. On the other hand, the system modeling

becomes too specific because new technologies are emerging and becoming popular constantly. In this way,

OMG proposes the Model-driven Architecture (MDA), a model-driven framework for software

development. Due to the advantages of both agile and model-driven proposals, a topic of interest is the

approach between them. In this way, we are working in a model-driven methodological framework for agile

development of WIS, named MIDAS which is presented in this paper.

1 INTRODUCTION

Nowadays, Web Information Systems (WIS)

development has become an interesting area in both

research and business world.

On the one hand, the WIS development needs

speci

fic methodologies (Fraternali, 2000) for taking

into account specific aspects of Web systems (Lowe

and Hall, 1999). Therefore, traditional

methodologies do not facilitate a quick and light

development (Fowler, 2000). This is the reason why

the agile methodologies (AM) have appeared, which

provide a sufficient development process but not

excessive.

On the other hand, the system modeling becomes

too speci

fic because new technologies are emerging

and becoming popular constantly. In this way, the

Object Management Group (OMG) proposes the

Model-Driven Architecture (MDA) (Miller and

Mukerji, 2001), a model-driven framework for

software development that proposes to specify the

whole system, with platform independent models

(PIM) to be able to develop the software system for

different platforms. In fact, Model-Driven

Development (MDD) has improved productivity,

quality and platform independence in

the development world, as Kulkarni and Reddy

(

2003) states.

Nowadays, an interesting topic is the approach

bet

ween agile and model-driven proposals (Turk et

al., 2002), collecting the advantages of both. In this

way, we are working in MIDAS (Marcos et al.,

2001a, 2001b, 2002, 2003a, 2003b; Cáceres and

Marcos, 2001; Cáceres et al., 2003; Vela and

Marcos, 2003a; Vela and Marcos, 2003b), a model-

driven methodological framework for agile

development of any kind of WIS, which takes into

account next issues: first, the need of specific

methodologies for developing WIS; second, the

agile methodologies motivates more than heavy

methodologies; finally, modeling first in an

independent way and later in a specific technology.

In this paper, we present a preliminar approach

for in

tegrating agile and model-driven practices in

the MIDAS framework. This approach is based on

five aspects which are very differents between both

practices.

The rest of the paper is organized as follows: in

sectio

n 2, the MIDAS framework is presented; in

section 3, the integration of the model-driven Web

development with agile practices is showed; section

4 sums up the main conclusions and future works.

523

Cáceres P., Marcos E. and de Castro V. (2004).

INTEGRATING AGILE AND MODEL-DRIVEN PRACTICES IN A METHODOLOGICAL FRAMEWORK FOR THE WEB INFORMATION SYSTEMS

DEVELOPMENT.

In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 523-526

DOI: 10.5220/0002637705230526

 SciTePress

2 THE METHODOLOGICAL

FRAMEWORK OF MIDAS

MIDAS is a model-driven methodological

framework for agile development of WIS, which

follows a service oriented approach. The MIDAS

architecture (Cáceres et al., 2003) is based on the

MDA of the OMG. MIDAS proposes: a) specifying

the whole system by Computation Independent

Models (CIMs), Platform Independent Models

(PIMs) and Platform Specific Models (PSMs); and

b) specifying the mapping rules between these

models. Besides, MIDAS suggests using the Unified

Model Language (UML) (OMG, 2003) as a unique

notation to model the whole WIS.

In order to model the different aspects of the

system it is important to identify them at the

modeling level as are stated by Kulkarni and

Reddy (2003). To identify the aspects of a WIS,

we have taken into account the middleware

architectures of the Web service development

platforms, as .NET or J2EE. So, we also propose a

n-tier model architecture. Until now, our approach

considers three aspects corresponding with the three

tiers most commonly accepted: graphic user

interface, persistence and business logic tiers. For

the sake of uniformity with the Web Engineering

terminology, these aspects will be called hypertext,

content and behavior respectively (Retschitzegger

and Schwinger, 2000). The advantage of this n-tier

model architecture is that it is easily scalable; so, to

incorporate new aspects, as security or management,

we have just to introduce a new tier.

Summing up, MIDAS proposes to model the WIS

according to two orthogonal dimensions (see figure

1): a) the MDA approach: that is, the degree of

dependence of the platform; and b) the n-tier model

architecture previously defined: that is, the aspects

to be considered in a WIS. So, it defines PIMs and

PSMs for each of the aspects above identified

(hypertext, content and behavior), as well as the

mapping rules between them.

3 TOWARDS A MODEL-DRIVEN

WEB DEVELOPMENT IN AN

AGILE WAY

As mentioned before, nowadays an interesting topic

is the approach between agile and model-driven

proposals (Turk et al., 2002). We also keep on

considering agile practices. However, as is showed

in figure 2, agile and model-driven process do not

take into account the problem domain and the

solution domain, in the same way (Weneger, 2002).

Then our goal is the approach between these

proposals, remarked as “area of interest” in figure 2.

We focus on the analysis of their several aspects

(Weneger, 2002), showed in table 2: People,

process, technology, models and software.

Figure 1: MIDAS modeling and their mapping rules

<< PIM-PSM mappings >>

PIM

PSM

CONTENT

Data Conceptual

Model

XML Schema

Model

BEHAVIOR

Service

Model

WSDL

Model

HYPERTEXT

CIM

Business

Model

<< PIM-PIM mappings >>

Domain

Model

Service

Composition

Model

BPEL4WS

Model

Xlink Model

<< PSM-PSM mappings >>

Navigation Model

Use Case Model

Slice

Model

ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

524

Agile, incremental,

evolutionary

waterfall

Model-driven,

generative

Don’ t do this

Area of

interest

Problem Domain Abstract Stability

Solution Domain Abstract Stability

Low High

LowHigh

Table 1: Distinguishing general aspects of agile and model-driven development process (Weneger, 2002)

Figure 2: Relation between the problem and solution domain (Weneger, 2002)

Aspects Agile Model-driven

People

Have highest priority; interactions between

customer and developer are facilitated.

No explicit role; problem domain model takes role of

discussion platform for customer and developer.

Process

Has medium priority; ensures consistency

of results of interactions between people;

incremental, evolutionary.

No explicit role; strong tendency towards waterfall, less

incremental processes.

Technology

Has lowest priority; only a means to an

end; must be as simple as possible.

Is at center of approach; problem model (e.g., PIM) is

manipulated, stored in and exchanged between repositories.

Can be used to generate software (e.g., using a PSM).

Model

Secondary artifact; only produced up-front

when absolutely need.

Primary artifact; source of generated implementation.

Software

Primary artifact; sole measure of progress Secondary artifact; depending on solution domain, provides

or adds aspects not covered by specification.

We study these different characteristics, and we

adopt the most interesting practices for us. Next, we

sum up the adopted characteristics in MIDAS:

People. As in agile processes, people have a high

priority for us. We propose that the interaction

between customer and the software project staff had

priority over other interactions. In model-driven

process, the customer only appears from the problem

domain model. In our case, the customer and the

software project start together from scratch and they

also finish together. In this way, we can guarantee

that the final product will satisfy to the customer.

Process. The process is an important aspect in our

proposal. The MIDAS process is incremental,

iterative, adaptive and with prototype, similar to the

agile process and far from the waterfall model

proposed in the model-driven processes. Also, we

can obtain several software versions from an

iterative and incremental process before the final

version of software (Jacobson, 2000). Therefore, an

agile and model-driven development can be also

carried out in an incremental and iterative way, as is

pointed by Ambler (2003).

Technology. Our proposal is addressed towards a

specific technology (XML and object-relational).

Then, unlike the agile process, we adopt different

models from MDA (CIM, PIM and PSM) to

represent platform independent and platform

dependent models,.

Model. Modeling is also prioritary for us. The

system knowledge must be written in a concise

notation, as states Atkinson and Kühne (2003). In

this way, we propose to use UML and UML profiles.

The models and the code are the only documentation

that we propose to generate. Then, in this point, we

disagree with agile process.

Software. Our main goal is to satisfy the customer.

We propose to generate a basic software version

shortly. In this way, we propose to develop a

software prototype in every iteration. We guarantee

the software progress thanks both short development

cycles and an incremental process. In this aspect, we

agree with agile process.

INTEGRATING AGILE AND MODEL-DRIVEN PRACTICES IN A METHODOLOGICAL FRAMEWORK FOR THE

WEB INFORMATION SYSTEMS DEVELOPMENT

525

4 CONCLUSIONS AND FUTURE

WORKS

In this paper, we focus on the integration of model-

driven and agile practices, adopting the most

interesting practices in the MIDAS methodological

framework for WIS development. MIDAS combines

the MDA and a three- layer architecture which

represent different WIS views (hypertext, content

and functionality). Based on this architecture,

MIDAS proposes system modeling with CIMs,

PIMs and PSMs and defines mappings between the

mentioned models.

Now, MIDAS is been validating by the CASE

tool development, named MIDAStool. Our aim is the

model-driven and agile practices integration into a

methodological environment for semi-automating

WIS development.

REFERENCES

Ambler, S. (2003). Agile Model Driven Development is

Good Enough. IEEE Software, Vol. 4, pp. 71-73,

september-october de 2003.

Atkinson, C., Kühne, T. Model-Driven Development: A

Metamodeling Foundation. IEEE Software, Vol. 4, pp.

36-41, september-october de 2003.

Cáceres, P.; Marcos, E. (2001) Agile Process for Web

development. Web Engineering Workshop in Working

days of Software Engineering and Databases, 2001.

(JISBD2001). Ciudad Real (Spain). In

http://www.dlsi.ua.es/webe01/articulos/s112.pdf (in

spanish).

Cáceres, P.; Marcos, E.; Vela, B. A MDA-Based

Approach for Web Information System Development.

Workshop in Software Model Engineering

(WiSME@UML'2003) in conjunction with UML

Conference. October, 2003. San Francisco, USA.

Accepted.

Fowler, M. (2001). The New Methodology. Retrieved

May 2001 from

http://www.martinfowler.com/articles/newMethodolog

y.html.

Fraternali, P. (2000). Tools and Approaches for

Developing Data-Intensive Web Applications: a

Survey. Retrieved July 2000 from the World Wide

Web: http://toriisoft.com

Jacobson, Booch and Rumbaugh. The Unified Software

Development Process. Addison Wesley, 1999.

Kulkarni, V., Reddy, S. (2003). Separation of Concerns in

Model-Driven Development. IEEE Software, Vol. 4,

pp. 64-69, september-october de 2003.

Lowe, D.; Hall, W. (1999). Hypermedia & the Web. An

Engineering Approach. J. Wiley and Sons.

Marcos E., Vela B., Cavero J. M. (2001a) Extending UML

for Object-Relational Database Design. Fourth Int.

Conference on the Unified Modelling Language, UML

2001, Toronto (Canadá), LNCS 2185, Springer

Verlag, pp. 225-239, 2001.

Marcos, E., Vela, B., Cavero J.M., Cáceres, P. (2001b)

Aggregation and Composition in Object-Relational

Database Design. 5 th. Conference on Advances in

Databases and Information Systems. Ed.: Albertas

Caplinskas and Johann Eder, pp.195-209, 2001.

Marcos, E. Cáceres, P., Vela, B., Cavero, J.M. (2002)

MIDAS/DB: a Methodological Framework for Web

Database Design. DASWIS 2001. Yokohama (Japan),

November, 2001. LNCS-2465. Springer Verlag. ISBN

3-540-44122-0. September, 2002.

Marcos, E. , Vela, B. and Cavero J.M. (2003a)

Methodological Approach for Object-Relational

Database Design using UML. Journal on Software and

Systems Modeling (SoSyM). Springer-Verlag. Ed.: R.

France and B. Rumpe. ISSN: 1619-1366. Volumen

SoSyM 2, pp.59-72, 2003.

Marcos, E., De Castro, V., Vela, B. (2003b) Representing

Web Services with UML: A Case Study. International

Conference on Service Oriented Computing. Trento

(Italia). 15-18 december 2003. Accepted.

Miller, J. and Mukerji, J. (Eds). (2001) MDA Guide

Version 1.0. Document number omg/2003-05-01.

Retrieved from:

http://www.omg.com/mda, 2003.

OMG. OMG Unified Modeling Language Specification.

Version 1.5. Retrieved from:

http://www.omg.org/technology/documents/formal/um

l.htm

, 2003.

Retschitzegger, W., Schwinger W. Towards Modeling of

Data Web Applications- A Requirement’s Perspective.

In Proceedings of the America’s Conference on

Information Systems. V.I, pp. 149-155, 2000.

Turk, D., France, R., Rumpe, B., Georg, G. Model-Driven

Approach to Software Development. Advances in

Object-Oriented Information Systems. LNCS 2426.

Springer-Verlag, ISBN: 3-540-44088-7, pp.229-230,

September, 2002.

Vela, B. and Marcos E. (2003a). Extending UML to

represent XML Schemas. 15th Conference On

Advanced Information Systems Engineering (CAISE

’03). Klagenfurt/Velden (Austria). 16-20 June 2003.

Ed. J. Eder, T. Welzar. ISBN: 86-435-0549-8. 2003.

Vela, B. and Marcos E. (2003b) Web Information Systems

Development: From navigational conceptual modeling

to logic modeling in XML. In Working Days on

Software Engineering and Databases. Alicante, Spain.

JISBD 2003 (in spanish).

Weneger, H. (2002). Agility in Model-Driven Software

Development? Implications for Organization, Process

and Architecture. Retrieved from

http://www.softmetaware.com/oopsla2002/wenegerh.p

ICEIS 2004 - INFORMATION SYSTEMS ANALYSIS AND SPECIFICATION

526