Tricia - Printout- Browse this Wiki
- Jul 21
Enterprises comprise complex socio-technical systems with a strongly interwoven structure and a highly non-trivial behavior. In order to support the management of these systems, different models of the systems' architecture, the so called Enterprise Architectures, are created - each capturing constituents of the enterprise relevant to the associated management task. These models can be complemented with derived indicators, i.e. metrics that measure the quality of the enterprise system in respect to a certain desired attribute. Using indicators, key information on the EA quality can be presented, discussed, and communicated in a condensed form.
In the computation of the values for the respective indicators, information on the structure of the enterprise but also measurements of the behavior of its constituents, e.g response times or failure rates, are utilized. Such behavioral information can be collected from a current (operational) system, i.e. can be determined ex post to review the current quality. Nevertheless, indicators should also be used as means for decision support for EA planning and have thus to be applied on planned architectures and systems, which are not operated. In this situation, only plans for the future structure exist, but no information on the will-be behavior is available.
Under the assumption that the structure of a system determines its behavior, simulation methods can be used to supply the missing information, as alluded to before. For doing this, the planned architectures are complemented with executable simulation models, which serve a surrogates for the actual constituents of the enterprise, such as business applications or hardware devices. These models are subsequently executed and the behavior of their model concepts is measured to create a prediction of the will-be behavior of the future system (ex ante). Based on this prediction, the "normal" methods for computing and deriving values of quality indicators can be applied. These values can thus be used as decision support in choosing the appropriate future architecture.
| [Bu08g] | Buckl, S.; Ernst, A.; Schweda, C.M..: An Extension to the Essential Meta-Object Facility (EMOF) for Specifying and Indicating Dependencies between Properties. Technische Universität München, Institut für Informatik, Lehrstuhl für Informatik 19, Technischer Bericht TB0801, 2008. |
| [Bu08f] | Buckl, S.; Matthes, F.; Renz, W.; Schweda, C. M.; Sudeikat, J.: Towards simulation-supported enterprise architecture management. In: Fachtagung Modellierung betrieblicher Informationssysteme (MobIS2008), Saarbrücken, 2008. |
