Decision Support for Architecture Management

Method and Analysis Technique for Optimizing Enterprise Architecture Models

Start date: 01.09.2012
Funded by: Universität Augsburg
Local head of project: Melanie Langermeier


Enterprise Architecture Management (EAM) is one mean in order to deal with the increasing complexity of todays IT landscapes. Thereby architectural models are used to describe the business processes, the used applications, the required infrastructure as well as the dependencies between them. The creation of those models is expensive, since the whole organization and therewith a large amount of data has to be considered. Thus, it is important to make use of these models and reuse them for architecture work. The models are a solid foundation for various kinds of analyses that support the decision making while optimizing the architecture. Analyses can determine the effects of the shut down of an application or the effects of an application failure. It is also possible to quantify the model using measurements like the IT coverage of business processes or the workload of a server. The generation of views on an architectural model, i.e. the focus on a specific aspect of the model, is also the result of an analysis. An example is the limitation of the model to the processes and applications of a specific organization unit. Architectural models can also be used for planning purposes. Starting from the model of the current architecture of the organization a target architectural model is developed. The goal is then to implement the target architecture throughout the organization. The development of a target architecture can be supported by identifying different planning options and evaluating project proposals.

In this project we want to define a method for architecture optimization and develop the required techniques for the specification and execution of analyses. The method encompasses the identification and evaluation of possible change scenarios as well as the evaluation of project proposals regarding consistency of changes and goal fulfillment. The adoption of these methods is supported by several analyses. These are for example a change impact analyses, the measurement of metrics or the scoping of the architectural model according to different aspects. We want to support the specification and execution of the analyses through the development of an analysis language. Thereby we focus on reuse and generic definition of the analysis. Using a generic meta model as foundation we are able to abstract from the great variety of used meta models in the EA domain. The execution of the analyses is done using Semantic Web Technologies and the Model Analysis Framework (MAF)