EVOLUTION OF ENTERPRISE INFORMATION SYSTEMS.

AN INDUSTRIAL APPROACH

Pietro D’Ambrosio

NET FIRST srl, Salerno, Italy

Keywords: Enterprise Information Systems, Legacy Systems, Evolutionary Maintenance, EIS Management, Software

Development Methodology, Enterprise Architectures.

Abstract: The problem of the evolutionary maintenance of large information systems is very critical. The time and

costs required to adapt the system to the market’s evolutive dynamics are no longer compatible with the

objectives of the business. The experimentation, on small-scale, in the laboratory of new technologies and

new methods is ineffective for this type of systems and it is rare that the customer accepts the risk of

innovations in an industrial project. In this paper we propose an intervention strategy and a reference

architecture to transfer, using an approach based on "small steps", innovations obtained from research into

new industrial applications.

1 INTRODUCTION

The increasing complexity of today's competitive

scenario is frequently associated with social,

organizational and cultural phenomena that

introduce discontinuities and critical issues in the

evolution process of business organizations and their

related information systems.

This results in a significant reduction in the life

cycle of the services provided by the company and

in a growing need for information to be collected

and analyzed in order to take effective decisions.

All this leads to a greater workload for the

information system, to an increased need of

evolutionary maintenance and a progressive

decrease in the quality of the system.

Thus, the time and costs of software

development are no longer compatible with the real

benefits that they bring to the system.

In this paper we introduce a new strategy of

intervention and a new architectural model tested in

an industrial environment to achieve new levels of

effectiveness in the evolutionary development of

complex information systems.

The techniques and the methodology adopted, in

part derived from the study of the Ultra Large and

autonomic systems (IBM, 2006 and Müller et al.,

2009), use an approach based on a logic wrapping of

the existing information system and on the

introduction of new components that enable new

features and add new "perceptual capacities" to the

system as a whole.

More generally, this approach tends to move the

human intervention to a higher logical level in order

to reduce drastically the management and

development costs.

2 PREVIOUS RESEARCH

AND INTERVENTION

STRATEGY

The intervention strategy described in this paper

derives from a series of research activities on tools

and methodologies for software development carried

out by some IT companies and university

departments (D’Ambrosio et al., 2010).

This work also benefits of a series of experiences

gained over the years on industrial projects of

various sizes carried out in situations highly critical

in terms of continuity of service, level of

functionality and security such as banking

information systems.

The heterogeneity of application environments

and technologies that we have met and the constant

need to minimize costs and timing of the project,

suggested a policy of innovation based on "small

steps", with a continuous monitoring of project risks

357

D’Ambrosio P..

EVOLUTION OF ENTERPRISE INFORMATION SYSTEMS. AN INDUSTRIAL APPROACH.

DOI: 10.5220/0003414003570362

In Proceedings of the 13th International Conference on Enterprise Information Systems (ICEIS-2011), pages 357-362

ISBN: 978-989-8425-55-3

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

related to the activities or processes involved with

the innovations introduced.

This constant attention to identify other small

areas of innovation and apply, in real project tasks,

what has been experienced in research activities,

created in our company an innovation "culture" and

a new open attitude that in a mid-long-term period

has undoubtedly produced satisfactory results.

In this sense we believe that our experience may

provide a useful contribution to define a more

general methodology for industrial research

activities on complex information systems. In this

context, in fact, the laboratory simulations and the

demonstration on a small scale performed via

prototypes almost always are ineffective and do not

provide the necessary certainty requested for the

immediate industrial application of technology.

The criticality and complexity of such systems

consists mainly in their large size, in the service’s

continuity requested, in the time and cost needed for

any evolutionary intervention and in the choice of

the best intervention strategy. All these issues are

not really reproducible in simulated environments.

On the other hand, it is unlikely that users decide

to introduce significant innovations (unless they are

required to do so) since it may affect the continuity

of service and increase the cost and time of the

project.

In addition, we have to consider the resistances

raised by the actors of development or business

processes anchored on their consolidated available

skills, by the corporate organizational cultures and

by the natural needs of the business.

This, unfortunately, has created a growing gap

between the availability of new technology and its

industrial transfer and has also significantly slowed

the evolution of this type of information systems.

Over the years we developed an intervention

strategy for gradual introduction of process and

product innovations in development activities and

maintenance of complex software systems.

This strategy, schematically represented in figure

1, is essentially composed of 5 steps:

 Step 1. Intuition. Discussion around the

innovation idea. Research hypothesis

formulation. Evaluation of research

hypothesis formulated. Structuring of the

scientific problem. Verification of the

theoretical foundations. Definition of research

objectives and verification criteria for all

results. Project implementation and

development of prototypes.

 Step 2. Evaluation of results and definition of

one or more cases of large-scale

implementation of the results to understand

the theoretical limits and the possible benefits

to be obtained. This may result in:

o Obtaining unsatisfactory results. The

process ends here.

o Obtaining partially satisfactory results.

In this case we reaffirm the general

principles, redefine some requirement

and restart the process again from the

step 1.

o Obtaining satisfactory results. The work

continues with the search for a new

project that allows us to introduce one

or more innovative elements derived

from the research.

Figure 1: The proposed methodology.

 Step 3. Analysis of innovation in the context of

a real project. Identification of intermediate

steps with the related risk in terms of: i) delay

on the project, ii) implementation strategies

(i.e. the need to maintain in parallel the old

tools and methods), iii) problems of measuring

results. Finding one or more innovation to be

applied in the next project and defining the

expected objectives.

 Step 4. Development of the Project. Monitoring

the activities and achievements.

 Step 5. Post-mortem analysis of the project.

Refinement of the results. Depending on the

case the process restart from step 3 (on a new

project) or from step 1 (on a new research

activity). Almost always, the feedback stops

at level 3 but it is also inevitable an impact on

upcoming activities that start from step 1.

We verified in recent years that on average it

takes at least three projects to consolidate any

innovation. Moreover, generally the first project to

which we apply the innovation has costs and time

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

358

higher of previous ones. As a result we tend to

choose small and not critical project to begin a new

experimentation cycle.

The second project has average cost and time

comparable with the previous ones. From the third,

instead, significant results will be achieved.

3 THE DEVELOPMENT

STRATEGY

Almonaies et al. (Almonaies et al., 2010) propose an

interesting overview on strategies to modernization

of legacy systems towards service-oriented

architectures using comparison criteria that take into

account different parameters. Among the other, we

point out the degree of complexity of the system and

the degree of adaptability of the process to the

specific type of system.

In terms of conceptual approach, the four

strategies identified by Almonaies et al.

(Replacement, Wrapping, Redevelopment and

Migration) remain valid for the context of the

evolution of complex information systems, although

in this case, they are not exclusively composed by

the legacy applications and are not necessarily

obsolete.

In the latter case, the reasons for the evolution of

information system can be profoundly different from

those that motivate migrating or rewriting a legacy

system. Indeed they are mainly oriented to overcome

the "unsustainability" of the costs and time for the

evolutionary maintenance of the system in a market

environment characterized by the acceleration of

competitive dynamics and by the drastic reduction of

the life cycles of services offered to customers (as in

the case of large banking systems).

Among the various modernization strategies, we

believe that the less risky, less invasive and more

suitable to our case is the "Wrapping". The types of

information systems we use as a reference, in fact,

consist of systems adopted in large organizations

that have a strategic value to the business and

generally have a good level of code quality.

Moreover, the aim of our present work is not the

modernization of existing components, but their use

as part of a more advanced system.

In our case, the wrapping is not made with the

aim of improving the existing system but to

"encapsulate" it and to extend its functionalities and

its "perceptual capacities."

In this way, we can add new elaborative

processes to the existing ones, intercept and

elaborate the input events and implement new

system features.

More generally, this strategy favors, over the

time, a gradual and natural shift of system gravity

center to the new "application engine”.

Our line of action then provides indications for:

i) implementing a new architecture (which includes

the existing system as "black box"), ii) extending the

perceptive capabilities of the system (monitoring the

events), iii) increasing capacity of self-control

(autonomic-like engine) and, finally, iv) simplifying

the development of business logic (SOA and

workflow based).

The main goal pursued through these strategies is

to move to an higher logical level the human

intervention, whenever possible, thinking in terms of

events, objectives, rules, logic and workflow

services rather than in terms of transactions and

functions.

4 THE ARCHITECTURE

The hypothesis of architecture that we proposed is

derived from a previous experience carried out under

the research project EMAF (D’Ambrosio et al.,

2010) and contains essentially principles drawn from

cooperating agents architecture, autonomic systems

and SOA.

Figure 2: The proposed approach for EIS evolution.

The reference to autonomic solutions is

essentially conceptual, since everything has been

strongly oriented to the needs of targeting

applicative systems.

The architectural scheme, briefly described in

figure 2, depicts the following components:

 a network of cooperating agents that

monitors specific behaviours and

environmental or relational events (Sensors);

EVOLUTION OF ENTERPRISE INFORMATION SYSTEMS. AN INDUSTRIAL APPROACH

359

 a main component, inspired by autonomic

system, that handles the rules and the

objectives to satisfy (Hybrid application

engine);

 a component of web service orchestration for

the implementation of identified actions and

for collecting feedback;

 a set of application services for the

management of the new features.

In this architecture we will identify at least four

conceptual levels at which the logic of the process

operates: 1) the agents that monitor events, 2) the

rules and objectives of the autonomic engine

(dynamically changed by the collected feedback), 3)

the workflow system of service orchestrator engine,

4) the specific processing services.

Supporting the development cycle with specific

authoring tools, we can "describe" a good part of the

process at a logical level higher than classical

programming techniques. This results in a

significant reduction in time and cost for

development, testing, and for the future evolutive

maintenance of the system.

4.1 The Filter Component

The filter component in the diagram of figure 2 has

been included only to highlight the need to intercept

the transaction's input received by the system and

translate it into events for the new application

engine. In reality this can be done differently

depending on how the information system has been

created.

The filter can operate in a "transparent" mode

with respect of transactions in the existing system or

it can intercept new transactions that can be

deployed exclusively in the new system.

With respect to existing transactions, if the

system includes a "Front Controller" component, the

solution coincides with that shown in the diagram.

Alternatively events can be taken from a log file

of transactions, captured by a transparent proxy or

even through a simple function to invoke before

each component of the existing information system.

The latter solution is certainly the most invasive

but anyway has a low criticality since it is an

intervention that can also be made automatically.

The main objective of this component, however,

remains that of "listening" the requests sent to the

system and eventually turning them into "events" of

interest for the new application engine.

4.2 The Hybrid Application Engine

This is an event-driven "application engine" based

on autonomic principles (which follows essentially

the structure). In particular, the "Monitor"

component collects and controls events sent from the

filter or from the sensors network and, if necessary,

forwards them to the analysis component.

The sensors network, useful to extend the

perceptual capabilities of the system in the "out of

home" environments (if is present) is implemented

through a system of cooperating agents that process

the events themselves and send them to application

engine.

The Analysis component will check which

business rules apply to the treatment of the event.

The choice of rules can be taken in many ways,

especially taking into account a number of state

information related to the environment (environment

variables) and the feedback received from the

execution manager.

In particular, the rules used to decide which

processes to run, are of two types: 1) rule "still

valid" that are evaluated in a predefined order and,

when met, are performed (and therefore more of a

rule can be enforced), 2 ) rules "alternatives" that

are evaluated and weighed, with a fuzzy logic

approach, calculating a truth value for each of them

and choosing the one with the greatest weight (of

course only if it exceeds a certain truth value).

The rules can also use environment variables to

take into account the state of the system and to

contextualize the choices. These rules can be

conditioned (in the calculation of the value of truth)

by the feedback returned by any processing

components.

The "Plan" component is used to schedule the

execution of actions, in the predefined sequence

defined by the selected rule and controls the

outcomes of any action.

The actions are performed using the "Execute"

component that interfaces various "effector" to

manage the internal services execution, the new

applicative features, the feedback and any replies to

be sent to user.

5 INDUSTRIAL APPLICATIONS

In addition to the use cases developed as part of the

research projects mentioned above, the intervention

strategy described in this work was being applied (at

different levels) in various industrial projects and in

particular for two projects addressed respectively to

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

360

a banking organization and to a local Public

Administration.

5.1 Project 1 – Integration of Banking

Applications

The first of these projects was oriented to make an

integration between two products for the banking

market: an application for managing transactions on

POS terminals and a multimedia system for the

presentation of geographically referenced data on a

choropleth map (using colorimetric scales for

observe macro phenomena of interest).

We started the industrial application of these

principles with this project because we had the

ability to install both products in the development

environment and did not have critical deadlines.

The goal of this project was to support the user

while interacting with the main application (pos

management) with a series of additional information

automatically displayed on a second monitor.

Figure 3: Functional integration schema.

In particular, this mechanism can be applied to

the analysis and to the statistic functionalities of the

system.

When the user selects a set of data or makes an

inquiry, the multimedia application managed by the

system automatically proposes on the second

monitor a visual representations of data using

choropleth maps (see figure 3).

The idea is to extend this mechanism to the other

applications of the information system in order to

support the operator with additional information

related with transactions running on the main

system.

For example, while the operator performs front-

end operations on a specific account, the secondary

system can display the photo of the customer, the

authorized signatures, alerts or other information to

be notified.

Figure 4 shows a scheme of the integration

model implemented. Let us point out that the

proposed approach produces a substantial

improvement to the existing information system

without making any invasive change.

Figure 4: The integration model implemented.

5.2 Project 2 – Multimedia System

Enhancement

The second project aims to enhance the

functionalities of a web application for the

management of interactive digital communication in

an urban area.

In this case access to the system takes place

predominantly by mobile terminals and local

systems for the "digital out of home" (video walls,

multimedia totems, LCD panels).

Here the functionality to achieve with the project

is the listening of normal requests that are made to

the system A (figure 5) and the analysis of user

related information (the position in the city, the

expressed interests, the history of performed

activities).

Based on this information, the system B provides

additional information to standard output provided

by system A in response to user requests

(information, advertisements, notices).

The model of integration of new components

with existing components (briefly described in

figure 5) remains essentially unchanged from the

previous project although they are profoundly

different in the application environment, objective

pursued and type of improvements made.

EVOLUTION OF ENTERPRISE INFORMATION SYSTEMS. AN INDUSTRIAL APPROACH

361

Figure 5: The integration model implemented in the

second project.

6 FUTURE DEVELOPMENTS

With the increasing complexity of the scenario and

the increasing competitiveness of the market in the

near future we expect a significant "discontinuity" in

technology and methodology adopted for the

information systems development and evolutionary

maintenance.

To make information systems more consistent

with the emerging functional requirements and to

reduce time and cost of their maintenance it is

necessary to adopt evolutionary approaches,

methodologies and tools that are more suited to new

levels of complexity.

More generally, we believe that it is time to

seriously consider creating a new level of IT

infrastructure, shared among multiple organizations,

which can significantly improve the effectiveness of

evolutionary interventions on the large enterprise

information systems.

This new IT infrastructure can ensure new

"levels of transparency" to the system with respect to

processes and events, and go further in the direction

already taken by the service orchestrator engines and

by the event-driven EAI systems.

The overall goal of our approach remains,

however, to reduce human intervention in the

management of the system and in the evolutionary

development, moving this operation on a higher

logical level. In this way we can achieve significant

benefits without introducing new risks and without

compromising the continuity of existing services.

ACKNOWLEDGEMENTS

The research has also been carried out thanks to:

 The "tomato-WAD" project, designed by NET

FIRST srl and cofunded by the Italian Ministry

of Research. This project aimed to create a

software framework for rapid development of

complex web applications.

 The “EMAF” project (Measure 3.17-POR

Campania 2000/2006), an ICT Cluster Project

funded by the Following partners: DMI Unisa,

Gruppo Net srl, Net First srl, Acube Lab srl.

The project had a number of goals including the

introduction of new levels of functionality into

existing information systems and the drastic

reduction in development time and maintenance

of the new components. In addition, this project

had defined an architectural pattern, a

component of run time and some authoring

tools specifically oriented to the introduction of

"Digital out of home" features in the traditional

information systems.

REFERENCES

D’Ambrosio P., Ferrucci F., Sarro F., Tucci M., “An

Approach for the development of Dooh-oriented

information systems” - 12th International Conference

on Enterprise Information Systems – ICEIS 2010, June

8-10, 2010, Funchal Madeira Portugal.

Almonaies A. A., Cordy J. R., Dean T. R.. “Legacy

System Evolution Towards Service-Oriented

Architecture”. Proc. International Workshop on SOA

Migration and Evolution (SOAME 2010), Madrid,

Spain, pp. 53-62

IBM Corporation 2006. “An Architectural Blueprint for

Autonomic Computing”, 4th ed, http://www-03.ibm.

com/autonomic/pdfs/ACBlueprintWhitePaper4th.pdf

Müller H. A., Kienle H. M., Stege U., 2009. Autonomic

Computing Now You See It, Now You Don't. In A. De

Lucia, F. Ferrucci (Eds.) Software Engineering,

International Summer Schools, ISSSE 2006-2008.

LNCS 5413Springer, pp. 32-54.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

362