A Requirements Based Approach for Automating Enterprise IT Architecture Modeling Using Multiple Data Sources

Enterprise Architecture (EA) is an approach where models of an enterprise are used for decision support. An important part of EA is enterprise IT architecture. Creating models of both types can be a complex task. EA can be difficult to model due to unavailable business data, while in the case of enterprise IT architecture, there can be too much IT data available. Furthermore, there is a trend of a growing availability of data possibly useful for modeling. We call the process of making use of available data, automatic modeling. There have been previous attempts to achieve automatic model creation using a single source of data. Often, a single source of data is not enough to create the models required. In this paper we address automatic modeling when data from multiple heterogeneous sources are needed. The paper looks at the potential data sources, requirements that the data must meet and proposes a four-part approach. The approach is tested in a study using the Cyber Security Modeling Language in order to model a lab setup at KTH Royal Institute of Technology. The lab aims at mirroring a small power utility's IT setup. The paper demonstrates that it is possible to create timely and scalable enterprise IT architecture models from multiple sources, and that manual modeling and data quality related problems can be resolved using known data processing methods.