Embedded Files
Abstract:
Meta models are the core of enterprise architecture, but still few methods are available for the creation of meta models tailored for specific purposes.
This paper presents two approaches, one focusing on the stakeholders’ information demand of enterprise architecture and the other driven by causal analysis of enterprise system properties.
The two approaches are compared and a combined best-of-breed method is proposed.
The combined method has merged the strengths of both approaches, thus combining the stakeholder concerns with causality-driven analysis.
Practitioners will, when employing the proposed method, achieve a relevant meta model with strong, and goal-adapted, analytic capabilities.
Introduction:
Meta models are at the core of enterprise architecture (EA) concepts.
They describe the fundamental artifacts of business and IT as well as their interrelationships in a single aggregate organizational model [42].
Such high level models provide a common language and a clear view on the structure of and dependencies between relevant parts of the organization.
Meta models serve three main purposes [23]:
1. Documentation of the enterprise architecture
2. Analysis of the enterprise architecture
3. Planning and design of the enterprise architecture.
These three purposes are in turn crucial to the success of management tasks such as product planning, business development, or business process consolidation [24].
However, devising a good meta model is not trivial.
Obviously, it is important that the meta model is relevant in relation to the management tasks it should support.
At the same time it is also of outmost importance that the meta model employed is kept minimal so that it can be used in practice where time and resources available to spend on enterprise architecture are limited.
Occam’s razor – the famous principle that entities must not be multiplied beyond necessity – is a rule of thumb very much valid in the context of meta modeling.
This paper presents two different meta modeling approaches, both based on the idea that minimal meta models are best obtained by maintaining a strict focus on goals when in the phase of meta model creation.
However, the two methods differ in important respects, and a combined, best-of-breed, method is therefore proposed.
The first, stakeholder-oriented, approach to meta modeling starts with stakeholder concern elicitation and is strongly driven by practitioners.
The resulting meta model seeks to satisfy the stakeholders information demands, each connected to distinct application scenarios.
The second, causality-based, approach is based on causal modeling of goals sought.
Starting from these goals, the resulting meta model provides a range of elements and attributes linked together by causality.
The meta model thus supports the analysis necessary to achieve defined goals [16].
Compared to the stakeholder- oriented approach, focus is set on attributes with causal relations rather than on elements.
According to method engineering literature [4, 5], a method consists of design activities, design results, information models, techniques and roles.
The proposed combined method focuses on the design activities, by introducing a meta modeling procedure.
The method combines the strengths of its two constituent parts, addressing stakeholder concerns through causality driven analysis.
The remainder of this paper is structured as follows.
Section 2 discusses related works, putting the present contribution in context.
Section 3 describes the two parent methods in greater detail, and includes an analysis of their strengths and weaknesses.
Section 4 outlines the combined method, including a concrete example of a possible application scenario.
Section 5 discusses the result and concludes the paper.
Engineering of Meta Model Fragments for Example Scenarios:
Goal Decomposition Method:
Comparison of Stakeholder-Oriented and Causality-Based Approach:
Meta Modeling Method Description:
Discussion and Conclusion:
In this paper we have analyzed existing approaches to EA meta model engineering.
Based on this analysis we have proposed an integrated method to define situational EA meta models on a meta model element level as well as on a level of element attributes.
Our basic assumption is, that EA models are no ends in themselves but have to provide a business value by supporting informed and well-founded decisions on how to continually transform the EA to fit an organization’s goals.
Enterprise architecture data that does not contribute to such decisions should not be maintained in EA models since it often increases the EA maintenance efforts and thus reduces the acceptance of EA in an organization.
Therefore our approach strictly derives relevant EA information from relevant application scenarios and stakeholder’s concerns down to an attribute level. Since this paper presents a first proposal of our enhanced method, the method has not yet been evaluated in a real case study.
However, the partial evaluation of its components as well as the example given in section 4.3 may indicate the applicability of the method as well as the progressivity of the method compared with existing approaches.
Therefore our next steps will include the evaluation of the method in industry projects as well as the improvement of exiting EA maintenance approaches [e.g. 7] concerning the updates of EA element attributes.
We will further investigate into the lifetime of accurate EA model information in order to further enhance EA application and maintenance processes.
Page updated
Google Sites
Report abuse