=Paper=
{{Paper
|id=None
|storemode=property
|title=Using Intentional Actor Modeling to Support the Evolution of Enterprise Software Architectures in Organizations
|pdfUrl=https://ceur-ws.org/Vol-586/iStar10-paper08.pdf
|volume=Vol-586
|dblpUrl=https://dblp.org/rec/conf/istar/GrossY10
}}
==Using Intentional Actor Modeling to Support the Evolution of Enterprise Software Architectures in Organizations==
https://ceur-ws.org/Vol-586/iStar10-paper08.pdf
Using Intentional Actor Modeling to Support the
Evolution of Enterprise Software Architectures in
Organizations
Daniel Gross, Eric Yu
Faculty of Information, University of Toronto, Canada
Abstract. The development and evolution of enterprise-wide software
architectures is influenced by multiple stakeholders and decision makers in
development organizations. Architectural design and evolution is embedded in
a larger distributed network of organizational participants, who actively
contribute to the identification, interpretation, delegation, reasoning, and
enacting of decision processes. In this paper we argue that to adequately
support architectural design and evolution in development organizations, it is
necessary to deal with the inherently distributed and interrelated nature of
organizational decision making in development organizations. This paper
proposes treating architectural design and evolution as a distributed and
interconnected decision process among organizational actors. The utility of the
proposed approach is explored through a pilot study at an insurance company
during an enterprise SOA evolution effort.
1 Introduction
Architectural decision making in software development organizations occurs within a
larger context of organizational decision making. Multiple stakeholders, occupying
various positions in the development, client and other third party organizations, are
involved in decision making that directly or indirectly influences architectural
development and evolution. For example, upper management scans and interprets the
organizational environment, such as current customer demands, future market
opportunities and the like, and decides on current and future strategies and goals of
the organization. Upper management then hands off enterprise strategies to other
decision making stakeholders in the organization, such as product management to
exercise their know-how in turning strategy into product approaches. Similarly, upper
management and product management rely on enterprise architects to develop
architectural principles and guidelines for a sound enterprise-wide architecture that
supports current and future strategic goals.
Decision making in development organizations is thus a distributed and
interconnected phenomenon, in which upstream management decisions have
influence on downstream architectural decision making, and where downstream
decisions influence upstream goal achievement. An architectural modeling and
analysis approach that supports architectural design reasoning and decision making
41
Proceedings of the 4th International i* Workshop - iStar10
needs to support representing, capturing and analyzing the distributed and
interconnected decision processes in development organizations[1].
2 Objective of Research
The objective of this research is to explore techniques for supporting enterprise
architects in capturing, reasoning about and communicating the architectural design
goals, principles and guidelines in the context of other organizational decision
processes in the development organization. More specifically, this research proposes
applying and adapting agent and goal modeling and analysis techniques to support
enterprise architects in:
1. representing, capturing and analyzing design reasoning and decision making of
designers and stakeholders from different contrasting viewpoints, such as from a
component designer’s point of view, and from the enterprise architect’s point of
view;
2. identifying higher level stakeholders and decision makers whose goals, priorities
and choices give rise to, and influence the lower level goal and design reasoning
and decision making of designers and stakeholders;
3. systematically analyzing how goals and design approaches compare and contrast
when analyzed from different points of view and in relation to their links to goals,
prioritizations and tradeoff making of higher level stakeholders in an organization
3 Scientific contribution
The main contribution of this research is the application and adaptation of i* [2] to
represent, capture and analyze distributed architectural design and decision making in
an enterprise development organization at different organizational levels such as the
management level, the enterprise architecting level, the enterprise system architecting
level and the component designer level; and the exploration of the utility of the
proposed approach during a study at an insurance company (to simplify in this paper
we use the terms agent and actor interchangeably).
While i* was originally proposed in the context of early requirements engineering,
it has also been applied to representing, capturing and analyzing software architecture
[3-5]. These approaches have focused on mapping agent and goal-oriented concepts
to concepts in the software architecture domain. In these works, agents directly map
onto components, dependencies across agents usually capture some quality
constraints across interactions of components, and architectural design reasoning and
alternatives are captured outside of agents, thereby assuming an anonymous and
global designer of the architecture and all components. This underutilizes the i*
approach given its ability to deal with distributed intentionality, autonomy and
decision making in organizational settings [2, 6].
42
Using Intentional Actor Modeling to Support the Evolution of Software Architectures
Furthermore, existing approaches do not link architectural design to relevant
business and management decision making stakeholders in the development
organization, or to client and third party organizations. The novelty of this work is
that it views the representing, capturing and analyzing of such distributed intents and
decision making across all such stakeholders in organizations as an essential part of
the development and evolution of software system architectures.
Fig 1: Modeling concept overview
This work can be seen as a significant extension of some early works of the authors
[7, 8] in which stakeholders in development organizations contributed intents towards
a design team involved in architectural reasoning and decision making. Here the work
extends these into dealing with distributed business and architectural decision making
in development organizations.
This work also extends i* with intentional viewpoints, a novel intentional actor
type, and an adaptation of the intentional role concept. Different intentional roles
support capturing the reasoning about different non-overlapping design
responsibilities. Intentional viewpoints adapt the intentional role concept to reasoning
about overlapping design responsibilities. This is analogous to the use of (non-
intentional) viewpoints [9] in the usual non-intentional settings, where multiple model
fragments are created by different modelers to describe the same phenomenon of
interest.
Overlaps between intentional viewpoints are identified when they include
overlapping hardgoals (see figure). Given the usual informal design discussions in
organizations, is it however difficult to provide clear rules to determine when two
hardgoals overlap. Instead, we propose some general guidelines. Given two hardgoals
g1 and g2: we consider the goals overlapping if g1 in some way implies by g2. We
43
Proceedings of the 4th International i* Workshop - iStar10
suggest looking at the following (non-exhaustive) list of criteria: if g1 is an instance
of g2; if g1 is a subtype of g2, if g1 is part of g2; if g1 is an implementation of g2. In
the case study we observed that architects and designers adjusted their terminology
for each other during discussion. We therefore used the same hardgoal when
representing each intentional viewpoint.
The intentional viewpoints concept illustrated in this research captures design
reasoning, while non-intentional viewpoints present different design perspectives,
which is the result of decision making. Such viewpoint models do not include the
design intents and reasoning that lead to design outcome (hence non-intentional
models).
4 Conclusions
The utility of the proposed approach was observed during a pilot case study at an
insurance company. Intentional viewpoints were used to clearly show and contrast the
reasoning of two stakeholders – the SOA architect on the one hand, and the designer
of the “Consumer” component on the other. The Consumer component designer is
concerned about local design simplicity and maintainability of the consumer
component, whereas the SOA architect aims to achieve maintainable, evolvable and
scalable enterprise systems, being responsible for the broader enterprise-wide design
problem. By placing these intentional viewpoint models side by side the designers felt
they were able to communicate better with each other over the design issue at hand.
Furthermore, when management issues and reasoning where included (in the form of
a network of “management” agents), linking them to the intentional viewpoints of the
enterprise architect and the component designer, software design reasoning could be
understood within the context of higher level management strategies and
prioritizations. These linkages were also considered useful by stakeholders at the
study site and were seen as contributing to SOA governance within the Enterprise.
5 Ongoing and Future work
The next steps in this ongoing pilot study are to further analyze design discussions
and extend the modeling technique where appropriate; to identify modeling
simplifications so that detailed agent and goal oriented design modeling and reasoning
can be distilled to a few representative agent and goal elements specifically adapted to
the explanatory needs of different types of stakeholders (maintainers, designers, reuse
managers, middle and upper management, etc.); to put these simpler models to test in
communicating and explaining discussion points with designers and stakeholders; and
to obtain relevant feedback and, where necessary, ideas for improvements.
Technical features being explored include developing a systematic approach to
compare and contrast the reasoning captured in different intentional viewpoints, while
taking into account the different scopes and levels of abstraction each intentional
viewpoint may present; the inclusion of different intentional agent types, such as
44
Using Intentional Actor Modeling to Support the Evolution of Software Architectures
intentional roles and intentional positions, as well as other knowledge structuring
mechanisms across agents, in particular inheritance and instantiation linking between
intentional agent types [2], and how these are combined with agent types to support
dealing with larger scale capturing and documenting of architectural decision
discussions and decision making in development organization; the integration of
intentional architectural agent modeling with non-intentional architecture modeling
approaches, and the representation of relevant non-intentional knowledge in a
notation neutral manner [10]; and appropriate tool support in enterprise organization
settings.
References
1. Curtis, W., et al., On building software process models under the lamppost. Proceedings of
the 9th international conference on Software Engineering, 1987: p. 96--103.
2. Yu, E., Modeling Strategic Relationships for Process Re-Engineering, in Department of
Computer Science. 1994, University of Toronto: Toronto.
3. Gross, D. and E. Yu, Dealing with system qualities during design and composition of
aspects and modules: an agent and goal-oriented approach, in Proceedings of the 1st
International Workshop on Traceability in Emerging Forms of Software Engineering.
2002a.
4. Kolp, M. and J. Mylopoulos. Software Architecture as Organizational Structures. in
Proceedings ASERC Workshop on "The Role of Software Architectures in the Construction,
Evolution, and Reuse of Software Systems. 2001. Edmonton, Canada.
5. Grau, G. and X. Franch, On the Adequacy of i* Models for Representing and Analyzing
Software Architectures. Proceedings of the First International Workshop on Requirements,
Intentions and Goals in Conceptual Modeling (RIGiM'07), 2007: p. 296-305.
6. Yu, E., Agent Orientation as a Modelling Paradigm. Wirtschaftsinformatik, 2001. 43(2): p.
123-132.
7. Chung, L., D. Gross, and E. Yu, Architectural design to meet stakeholder requirements, in
Software Architecture, P. Donohue, Editor. 1999, Kluwer: San Antonio, Texas, USA. p.
545-564.
8. Gross, D. and E. Yu, Evolving System Architecture to Meet Changing Business Goals: an
Agent and Goal-Oriented Approach, in Proceedings of the First International Workshop
From Software Requirements to Architectures (STRAW 2001) at the International
Conference of Software Engineering. 2001: Toronto, Canada.
9. Nuseibeh, B., J. Kramer, and A. Finkelstein, A framework for expressing the relationship
between multiple views in requirements specification. IEEE Transactions on Software
Engineering 1994. 20(10): p. 760-773.
10.Gross, D. and E. Yu, Resolving artifact description ambiguities during software design
using semiotic agent modeling, in 12th International Conference on Informatics and
Semiotics in Organisations. 2010: Reading, UK.
45