Abstract:
The transparency created by enterprise models is a valuable support for organizational engineering and especially enterprise transformation.
However, current approaches are hardly suitable to also create transparency of the enterprise transformation process itself.
The paper at hand contributes a systematic approach for capturing dynamics of enterprise transformation in enterprise models.
Therefore we differentiate dimensions of time (modeling time, valid time) of enterprise models on a macro level as well as a set of model transformation operations on a micro level.
The proposed model transformation operations on a micro level are suitable for describing the differences of as-is and to-be models in an enterprise transformation program.
Introduction:
Enterprise modeling (EM) addresses the modeling of business processes, goals, strategy, information entities, business structure, support systems, skills and people of an enterprise [11].
Thus, enterprise models integrate conceptual models of information systems and models of supported business functions and can provide the necessary transparency for a systematic support to organizational engineering [9].
By representing both organizational and technical infrastructure, EM may also provide a broad and consolidated view of an entire corporation or government agency [17], also known as enterprise architecture [15, 20, 26, 32, 35].
The complexity and thus maintainability of enterprise models leads to a trade-off between detailed but isolated models on the one hand side or aggregated but broadly connected models on the other hand side [18].
In order to provide support for transformation in an efficient way, EM has to be driven by business and/or IT oriented application scenarios [34] based on stakeholders concerns [22, 23, 36] (goal orientation) [20, 26].
Since the involvement of heterogeneous stakeholder groups may create conflicting requirements in a complex environment, an appropriate documentation and communication of the enterprise models is vital.
A suitable degree of formalization is needed in order to ensure traceable and repeatable results [33].
Furthermore (semi) formalized models and well structured methods are needed to enable division of labor and common understanding among the stakeholder groups [12, 14].
While documentation and analysis of enterprise models (represented by as-is models) are well covered in academic and practitioner approaches, EM for planning transformation purposes is covered much less so far.
Since neither the corporation or government agency itself, nor its environment remains static during a transformation project, and because to-be models may change as projects are launched, the consideration of dynamic aspects is important for planning a transformation program.
As we will illustrate in the next section there are already a few contributions that give first structure to the challenges of EM for planning purposes, e.g. [1, 6].
Based on these contributions this paper aims at answering the following research questions:
1. How can dynamics be represented in enterprise models?
2. What implications arise for the characteristics of enterprise meta models?
3. How can consistency between enterprise model snapshots of different points in time be assured?
For answering these questions the core contribution of this paper will be a framework of temporal dimensions that need to be covered by enterprise models.
However, the paper does not consider the planning of future states of a corporation or government agency by the means of enterprise models but the transformation of current states to planned states.
Therefore it is presumed that at least a to-be model exists and that the decision about the planned implementation/ transformation has already been made.
The remainder of this paper is structures as follows.
Section 2 will derive requirements for capturing dynamic transformations in EM from related work in the field of enterprise architecture.
Section 3 will review solution components from related work in the fields of software and database engineering, concerned with representing dynamics and time.
In section 4 we will present our proposal for capturing dynamics in EM, including a case study demonstrating its application.
The article closes with a discussion and an outlook.
Macro Level:
Macro Level Steps:
Discussion:
At the macro level, the presented proposal provides a structure for temporal dimensions that are relevant for modeling and representing dynamic transformation.
Thereby, approaches from database and software engineering deliver useful ideas concerning the representation of time in enterprise models.
The use of different timestamps in enterprise models enables analyses of different points in time, and ensures a historiography of enterprise model versions at the same time.
Beyond that the use of a new model type, the transformation model, was deduced from approaches in dynamic software architecture engineering.
By means of this model type the transformation can be planned and the relevant information can be stored separately from the enterprise models.
At the proposed micro level, the transformation planning is performed by analyzing the initial and the target enterprise model and establishing a procedure model for the transformation. In doing so, the consistency between the respective enterprise models is checked and can thus be assured before implementing the transformation.
If conflicts arise during the analysis step, this information should trigger adjustments of the to-be model.
Such an information flow needs to be integrated by a comprehensive planning process. The identified temporal dimensions modeling time and valid time require enterprise models to provide attributes that are able to capture timestamps.
Hence, a prerequisite for enabling transformation of enterprise models is an adjustment of the respective meta models.
While valid time should at least be captured by a valid from timestamp (as shown in Figure 1) modeling time should be represented in the form of a version number in order to enhance readability for the model user.
As current practices indicate that also the modeling of ideal “vision models” is required [1], the meta model should furthermore provide an additional attribute for marking this kind of special model. Using a transformation model also demands an adaptation of the meta model.
The transformation model must be able to capture information on successor relationships, the affected model elements in other enterprise models as well as their interrelations.
The respective attributes need to be provided by the meta model capturing links to other model elements, for example.
In order to identify the necessary transformation steps and establish the transformation procedure model, the support of an adequate modeling tool is recommended.
Such a tool must be able to analyze and compare models and propose a useful sequence of transformation steps depending on the interrelations and interdependencies between the model elements.
Moreover, the tool could support the task of representing temporal information as well as model numbers or label in a manageable way.
For the correct assessment of the affected elements that are part of the transformation, it seems inevitable to clear when two elements are regarded as identical to assess deviations between elements of different models.
From a theoretical perspective, two model elements in two different models are identical if they disclose no changes with regard to the transformation which is to be performed. Consequently, it seems unreasonable to define that only elements having completely identical attribute values are regarded as identical.
On the contrary, attributes that are regarded as relevant to the planning and transformation purposes and therefore for deviation analysis, should be defined situationally.
For example, only the change of a release version of a used database management system software could be relevant for planning, so the comparison would only be performed upon this attribute. The presented approach provides a concept for enabling transformation of enterprise models under certain limitations.
The case study illustrates the application of most of the elements proposed in this paper. It therefore demonstrates the feasibility of the presented approach.
However, some information is not contained in additional models or visualizations but is represented as additional attributes to already established model elements. This especially holds true for the proposed usage of transformation models.
With regards to the highly complex systems that are subject to enterprise modeling, one will presumably come across multiple modeling levels, opposed to the two presented levels in this paper.
Though, the presented macro and micro level are universally suitable to different modeling levels within enterprise modeling (and thus recursively applicable) because they are differentiated by their central subject of consideration: While on the macro level whole models and their temporal development are examined, the micro level focuses on the inner structure, i.e. on the model elements.
However, the question on which modeling levels the macro and micro framework is applied, i.e. on which level the planning is conducted, can only be answered by situational decisions about the planning process and may require additional heuristics.
In order to show a general concept, only a generic type of model element is considered at the micro level.
Although this type is not specified, the addition of model elements of different nature with different characteristics might result in a higher complexity because of additional interdependencies between the different element types.
As stated above, one model element could itself consist of links to other models, for example. Such recursivity is not considered in the presented approach and should be examined and integrated in a further contribution.
As stated by Bradbury et al. [4] the removal and addition of connectors between architectural elements are two out of four possible modifications.
The illustrated approach to plan the transformation from one enterprise model to another does not consider connectors, i.e. interrelations between model elements.
The planning of transformation of relationships requires a formal description in order to ensure consistency and functionality in the planned model. A comprehensive procedure model for transformation planning will have to include this aspect.
Finally, as already stated in the introduction, the presented approach does not cover decisions creating to-be models or choosing a certain transformation path.
Hence, this approach can be understood as the complementary concept to a model planning process.
Summary and Outlook:
With regard to planning purposes, this paper presents a systematic approach to capture dynamic transformation in enterprise models.
The approach distinguishes a macro and a micro level, making the transformation process more transparent.
On a macro level we have proposed a framework for temporal dimensions that are relevant to represent model transformation.
From this framework, valuable propositions for adapting enterprise meta models in order to enable transformation can be derived.
On a micro level, we propose a procedure model for analyzing differences between as-is and to-be mod-els with the objective of deriving effective transformation operations.
The proposed procedure enables a consistency check between as-is and to-be models as well as the realization of enterprise model transformation for planning purposes.
While enterprise models create transparency and support for organizational engineering, the transformation of enterprise models themselves also requires a structured and comprehensive planning method.
Such a planning method must embrace representational aspects as well as procedural aspects.
Additionally, dynamic aspects like unplanned shifts and their effects on interdependent enterprise models must be considered.
Lessons from the field of system theories will probably provide useful input in this context and should be evaluated in further research activities.
The presented approach on the macro level forms a contribution to the subject of representation.
The longterm goal is to establish an information model that is able to capture all planning relevant aspects while taking dynamic changes into account.
This will require further research regarding meta model engineering for enterprise models and especially transformation models.
The construction of transformation models will furthermore require an adequate modeling technique.
Procedural aspects that are concerned with a planning process itself, i.e. with involved roles and activities, will complement this approach.
The presented approach for establishing a procedure model for transformation constitutes a first step towards the planning procedure model, targeting the core aspect of transformation planning.
Besides that, other aspects around the planning and transformation of enterprise models will be subject to further research.
For example, strategic questions on where to start planning future states might be of interest. In this context, the idea of defining a “leading” model type or a “leading” model element seems reasonable.