A Model-based Approach to Managing Enterprise
Information Systems
Robert France
1
, Roger Burkhart
2
, Charmaine DeLisser
3
1
Colorado State University, Fort Collins, Colorado
2
Information Systems, Deere & Company, Moline,Illinois, USA
3
University of Technology, Kingston, Jamaica
Abstract. Organizations must evolve their infor
mation systems (IS) in order to
adapt to changes in their environment or to maintain or enhance
competitiveness. The use of modern application integration technologies (e.g.,
middleware) and advanced network technologies has resulted in IS that provide
services at unprecedented levels, but at the price of becoming more complex
and thus more difficult to evolve. By way of concrete examples, this paper
focuses on the use of system models expressed in the Unified Modeling
Language (UML) to effectively manage information systems assets. The system
models capture critical information about an organization and are part of an
overall framework called the Application Mapping Framework or AMF. The
AMF can be used by IT architects and planners to track applications, relate
descriptions of system artifacts across different levels of abstraction and support
redundancy, gap and impact analyses. The paper also identifies management
roles needed to ensure that the AMF repository contains comprehensive and up-
to-date models.
1 Introduction
The mission-critical role that Information Systems (IS) play in accomplishing
business goals requires that they be managed and tracked as organizational assets. In
this paper we describe a framework called the Application Mapping Framework
(AMF), for organizing information about planned and deployed applications in an
organization. The AMF is intended to support disciplined management and evolution
of IS resources and enables business managers, information technology (IT) planners
and architects, and application developers to (1) make decisions that minimize
development risks and costs, (2) identify opportunities for cutting costs, and (3)
identify new business opportunities. The AMF is more than just a static application
portfolio. It provides services that can be used by IT planners and architects to support
redundancy, gap and impact analyses. Proper use of the AMF will enhance the ability
of an organization to maintain a corporate memory and utilize that memory to cost-
effectively evolve its IS resources and business processes to meet business goals. The
France R., Burkhart R. and DeLisser C. (2005).
A Model-based Approach to Managing Enterprise Information Systems.
In Proceedings of the Joint Workshop on Web Services and Model-Driven Enterprise Information Systems, pages 171-180
DOI: 10.5220/0002516201710180
Copyright
c
SciTePress
AMF is intended to provide a single, accurate, organized source of information about
business processes, applications, data and other IT resources.
An overview of the AMF architecture is presented in Section 2 of this paper.
Types of analyses supported by the AMF and management functions required to
build, use and maintain the AMF, are presented in Section 3. Section 4 explains by
way of an example how the AMF could be used to support IS planning and evolution.
The paper concludes with our views on the merits of using a model-based approach to
IS management and an outline of our planned work in this area.
2 An Overview of the AMF
The AMF provides a logical architecture for a repository of information on
applications and data within an organization. Its development is based on experience
gathered on industrial projects that focused on developing application portfolios for
organizations with a large and diverse set of distributed applications. The AMF
specifies an application that is flexible in terms of the physical form or location of
information could be captured and integrated in the framework. To help organize its
wide range of topics and content, the information in the AMF is structured into a
number of core views. Information in a core view can be further organized into sub-
views.
Business
Architecture
Physical Design
Architecture
Application and Data
Architecture
Deployment
Architecture
Tracked Artifact
Business to
Application Mapping
Application to Physical
Design Mapping
Physical to Deployment
Mapping
Model
Management
Fig. 1. Enterprise Application Map Structure
172
The UML (Unified Modeling Language) [1] is used to describe the structure of
information in the AMF. An UML package is used to describe a view. UML Class
Diagrams are used to describe the conceptual structure of information in a view,
where a UML class represents a type of information item, a UML association
represents a conceptual relationship between peer information item types, and an
UML specialization relationship represents a further classification of an information
item type. A view, represented as a package, contains a structure of packages
(representing sub-views) and types (representing information item types).
At the top level, the AMF is organized into three views.
The IT Planning View contains information pertaining to ongoing and planned IS
projects, and includes information on tactical and strategic plans. The IT
Planning View is intended to support the work of IT planners and project
managers.
The Asset View contains information about enterprise-wide and domain-specific
reusable business artifacts. This view is intended to support systematic reuse of
development experiences across an organization.
The Enterprise Application Map is the central component of the AMF and
contains information about the current and planned information system resources
(e.g., applications, data, processes, roles) that are tracked within an organization.
The previously mentioned views utilize information within this view (as
indicated by the dependency relationships – the dashed arrows – between the
packages).
This paper focuses on the Enterprise Application Map. The information in the
Enterprise Application Map is organized into the following primary views (see Figure
1):
Business Architecture: This view contains information about the business
processes and entities that are tracked by an organization.
Application and Data Architecture: This view contains information about the
logical (i.e., technology-independent) aspects of applications and data. The
information includes descriptions of the IS artifacts (applications and data) as
they exist, as well as plans for evolving the artifacts. Descriptions include models
and artifact metadata.
Physical Design Architecture: This view contains information about the physical
design of applications and data, that is, it presents a technology-specific view of
applications. This view allows one to track the technologies that are used to
implement applications and data.
Deployment Architecture: This view contains information about the deployment
and usage of applications within an organization. Information pertains to the “as
is” deployment and usage of applications and data, as well as to planned
deployments and usages.
Relationships between concepts across these views are described by the Mapping
Packages:
Business to Application Mapping: This package links elements in the Business
Architecture view to the Application and Data Architecture view. The mappings
provide traceability of business concepts to logical (platform independent)
application concepts.
173
Application to Physical Design Mapping: The mappings in this package provide
traceability of logical application elements to physical (platform-specific) design
elements.
Physical to Deployment Mapping: The mappings in this package provide
traceability of physical design elements to the artifacts to their deployed forms.
The other packages of information in the Enterprise Application Map contain
information that is orthogonal to the packages described above.
Tracked Artifact: This package contains information about properties that are
common to artifacts that are tracked in the AMF. Currently, this includes only
information pertaining to versioning of artifacts.
Model Management: This package contains information about models, groupings
of model elements used to present views of applications, the tools used to display
models and the organizational roles responsible for maintaining the views.
3 Using the AMF
This section outlines the kinds of analyses that are supported and the management
roles that are recommended for effective management and evolution of the AMF.
3.1 Management Roles
Effective use of the AMF by business analysts, architects and system developers is
possible only when the contents of the AMF are relevant, properly packaged, easily
retrieved, current and accurate. The following are recommended management roles
that address issues related to the relevancy, accuracy, and usability of the AMF:
Content Manager: Responsible for packaging, cataloging, and updating AMF
contents.
Content Collector: Responsible for collecting candidate contents.
Content Certifier/Evaluator: Responsible for evaluating and certifying candidate
AMF contents. The evaluation is carried out to determine, for example, the
accuracy, relevance, and currency of candidate content.
Content Disseminator: Responsible for promoting and facilitating the use of the
AMF.
AMF Strategic Planner: Responsible for developing and maintaining plans for
evolving the AMF. This involves analyzing the usage of the AMF, analyzing
repository contents (e.g., identifying content with diminishing returns), and
identifying opportunities.
3.2 User Roles
Users of the AMF can be classified in terms of the roles they play in system and
business process management and development. Below we list the roles and the types
of interactions they can have with the AMF.
174
Business Analyst: Responsible for defining, documenting and updating business
processes.
IS Architect: Responsible for planning and managing the integration and
evolution of IT systems that support business processes.
System Architect: Responsible for designing and managing the evolution of a
particular system.
System Developer: Responsible for implementing system designs and changes.
3.3 Model-based Analyses
A sample set of IT project planning activities supported by the AMF are listed below:
Impact Analyses: During project planning one needs to determine, among other
concerns, how the system to be developed impacts other systems, what resources
are required and available for the project, what parts of the system functionality
can be provided by existing system components and what parts need to be built or
acquired. The AMF can be used to support impact analysis, determining the
impact of change on the organization’s ability to effectively meet business needs.
The relationships among the artifacts in the AMF (for example, data/object
create, read, update, and delete relationships between applications and
data/objects) can be used to determine the impact of planned changes and new
features on existing applications and data and on other current and planned IT
projects.
Gap Analyses: The AMF can also be used to support gap analyses. As new
processes and system functionality are developed, gaps in the existing integrated
system need to be identified and filled. Gap analysis is concerned with
determining the missing functional and process elements that need to be present
in order to implement new functionality of processes. The repository can be used
to determine what parts of a system are under development or already exists, and
what parts need to be obtained from outside vendors or be built in-house.
Redundancy analyses: As an organization’s pool of systems grows, the need to
identify redundancies to reduce inefficiencies and avoid conflicts arising from
multiple representations of a single concept across an organization becomes
evident. Redundancy analysis is concerned with identifying systems that provide
similar services. The repository can be used to determine whether proposals for
new system features can already be met by existing systems and to determine
wasteful overlaps in system functionality.
Reuse analyses: Order-of-magnitude improve-ment in productivity and system
quality can be accomplished if developers reuse product experiences. A well-
managed integrated system can form the basis for identifying potentially reusable
experiences across an organization. Reusability analysis is concerned with
identifying potentially reusable artifacts. Commonality analyses can be carried
out on the repository to identify organization-wide and domain-specific patterns
that can be packaged for reuse (e.g., as product frameworks, components,
reusable models).
175
4 An order fulfillment process scenario
The following scenario, though fictional, provides a realistic view of how the AMF
can be used to support IT planning and system evolution within an organization. The
Commercial Equipment (CE) Division of a fictitious organization has acquired a new
distribution channel that is located overseas. The need to adhere to reporting
regulations and other standards (e.g., customer addressing) in force within the foreign
territory requires CE to reengineer its order fulfillment processes and systems.
In this scenario the AMF contains a web of artifacts ranging from business models
of the processes to documents describing the deployment and usage of applications,
and the computing infrastructure that currently support the processes. The AMF is
accessed through interfaces that provide reporting functions and browsing starting
points that are particular to the roles of the individuals accessing the repository. For
example, the Business View interface of the AMF provides business analysts with a
business-oriented view of the repository contents from which they can drill down to
more system-specific views if required.
In the absence of an AMF, analysts, planners and developers have the challenge of
locating, relating, and analyzing possibly poorly documented information about the
current order fulfillment processes and supporting systems within CE. They may even
have to revert to source code analysis. These activities are expensive, error-prone, and
time-consuming. More importantly, such an environment is not conducive to the
development of systems that fully exploit resources that can significantly reduce the
cost of development without sacrificing system quality.
4.1 Business Analysis
The business analyst is responsible for defining an order fulfillment business process
that will handle the orders of the new distribution channel. To carry out this task the
analyst needs to (1) consider the impact of the proposed process solutions on existing
processes and systems, and (2) identify possible opportunities for exploiting current
system resources in the execution of the new processes, in order to define a cost-
effective and realizable process.
As a starting point, the analyst uses the AMF to determine the location of
documents that describe the current CE order fulfillment processes. Using the
documentation reporting facility of the Business View interface, the analyst locates
information on order fulfillment processes. A partial view of the table that is
displayed as a result of the interaction with the AMF is shown below:
Table 1. Relationships between Processes and the Responsible Organizational Roles
Subject
Area
Business Activity Responsible
Organization
Order Fulfillment Org1
Order Entry Org2
Order Routing Org2
Order
Order
Management
Org3
176
The order fulfillment processes are contained in the Order subject area and
consists of business sub-activities. Clicking on an activity name in column 2 takes the
analyst to a page that contains model(s) of the activity. These models can be
expressed as Activity Diagrams, Interaction Diagrams, and/or Use Cases. Using these
links the analyst can not only access models that help in understanding the processes,
but also use the information to identify models that are impacted by the change and
that can be reused to describe the changed process. The analyst also needs to work
with the owners of the process descriptions that will be impacted by the change.
Clicking on the items in the third column of Table 1 results in a page that displays
contact information for business process owners.
Table 2 is a partial view of the table that is displayed when Order Entry in column
2 of Table 1 is selected (in this case the process models are organized by the types of
orders processed):
Table 2. Process Model Table for Order Entry
Business
Activity
Order Entity
Type
Essential
Process
Model
Process
Realization
Model
Domestic
Dealer Order
Ess-.mdl Real-
OD.mdl
Export Order
Region 1
Ess-
E1.mdl
Real-
E1.mdl
Order
Entry
Export Order
Region 2
Ess-
E2.mdl
Real-
E2.mdl
There are a number of variants of the Order Entry process, each determined by the
type of order it processes. Selecting an order type in column 2 of Table 2 results in a
page that describes the order type. Columns 3 and 4 contain pointers to models of the
processes. An essential process model describes a process in terms of externally
observable effects (i.e., effects that are observable by users of the business processes
– the external view), and a process realization model describes a process in terms of
how the activities are carried out (the internal view).
The analyst also needs to determine the business entities that are impacted by the
change. To support this task the AMF can be used to produce the following table:
Table 3. Trace relationships between business activities and business entities
Business Activity Business Entity Access Type Responsible
Organization
Order Fulfillment Org 1
Order Entry Order
FDD
Create
Create
Org2
Org2
Order Routing Supplier
Update
Supp
Order Management Customer
Account
Update
Customer Dept
177
Selecting items in column 1 of Table 3 results in a page that shows the realization
process models indicating which entities are created, accessed, updated and deleted by
the business activity. Column 2 lists the business entities that are manipulated by the
business activities, and column 3 specifies the type of access (Create, Read, Update,
Delete).
The analyst also needs to have an idea of the order fulfillment systems and
databases that would be impacted by the change in order to identify a cost-effective
process solution. Another table (not shown), can show the relationship between the
business activities and the systems and databases that support the activities
. Selecting
on the items in System and Database columns can link the analyst to a page that
contains descriptions of the artifacts, contact information for the owners of the
artifacts, and pointers to more detailed information about the systems and databases.
5 Related Work
Other frameworks for information systems architecture are being used today, most
notably being the Zachman Framework (ZF), the Four+one framework and the RM-
ODP. Each has its own merit providing developers of new systems architectural
options for conceptualizing and designing. Zachman Framework is pre-object and
reflects a structured approach to development. It consists of a thirty-six-cell matrix
covering the perspectives of different stakeholders and aspects of the architecture. It is
seen as the best way to conceptualize all the elements of a system but has been
criticized as being process-heavy, requiring years to create. Ambler [3] suggests ways
in which ZF can be used in an agile manner.
The RM-ODP [2] is rooted in object analysis, and covers five viewpoints
enterprise, information, computational engineering and technology. Evitt [4] points
out that the viewpoints are abstract and do not reflect the concerns of specific
stakeholders as the ZF.
The AMF being proposed provides a lower level of detail than the ZF and RM-
ODP. Whereas the frameworks mentioned above can be used for developing new
systems, the AMF is intended for use as a lightweight means to document existing
systems and the way they relate to each other. It is to be used as a management tool
for identifying gaps, redundancies and reuse opportunities, and to be able to perform
impact analysis. The AMF can be used within the context of both the ZF and the RM-
ODP.
6 Further work
The AMF can provide a comprehensive representation of an enterprise’s business and
information systems and the means to conduct relevant queries and analyses. The
proposed business architecture, application and data architecture etc. serve to define a
workable structure for organizing, managing, analyzing and evolving enterprise
information systems. Populating this framework with suitable, well placed and
accurate business and information system design and implementation models
178
however, requires the involvement of skilled modelers, the formulation of and
adherence to standards of operating that will guarantee capture of accurate
information in a timely manner. The discipline required to make the use of the AMF
a success will ultimately result in improved practices, processes and tools and to a
more mature use of IT.
Our next step is to validate the AMF by using it to develop an IS repository for an
industrial partner. We are currently evaluating different development environments
for hosting, populating and querying an AMF repository. We will then develop and
deploy a prototype repository infrastructure and evaluated its usage. The experience
we gain will help refine the architecture and give insights into the types of
mechanisms needed to seamlessly integrate AMF related activities and IS
development and planning activities.
References
1. The Object Management Group (OMG): Unified Modeling Language: Superstructure, the
OMG, {http://www.omg.org}, Version 2.0, Final Adopted Specification, August, 2003
2. Open Distributed Processing Reference Model: ISO/IEC IS 10746.
http://www.joaquin.net/ODP. Visited 3/01/2005
3. Ambler, S.W., Architecture and Architecture Modeling Techniques (2002)
http://www.agiledata.org/essays/enterpriseArchitectureTechniques.html Visited 12/11/2002
4. Evitts, P. : A UML Pattern Language. MacMillan Technical Publishing (2000)
179
180