Introduction
Enterprise Architecture (EA) serves as the strategic blueprint of an organization, aligning business strategy with IT infrastructure, processes, and personnel. In large-scale industrial environments, maintaining this alignment is fraught with challenges. Information is often fragmented across heterogeneous data silos—ranging from Excel spreadsheets and SQL databases to unstructured PowerPoint slides. Furthermore, while standards like TOGAF provide a foundation, organizations frequently require bespoke extensions to link architecture with specific requirements or business process models (e.g., BPMN). Finally, the sheer volume of EA data necessitates role-specific visualizations to prevent information overload for stakeholders ranging from CEOs to IT engineers.
The TEAP framework, implemented through the SmartEA tool, addresses these challenges by applying Model-Driven Engineering (MDE) principles to the organizational level.
The TEAP Framework: Core Methodology
The TEAP framework is built upon three pillars: model federation, virtualization, and multi-viewpoint visualization.
Model-Driven Federation
To overcome the "Data Silo" problem, the framework employs a semi-automated sourcing process. Instead of manual data entry, "model discoverers" extract information from external sources like Excel or SQL databases. These are converted into intermediate models which are then transformed into a unified TOGAF-based repository using Model-to-Model (M2M) transformation languages like ATL.
Model Virtualization and Weaving
Rather than physically merging disparate models—which can obscure traceability—TEAP utilizes Virtual EMF. This approach creates a "virtual metamodel" that weaves links between different domains (e.g., connecting a TOGAF architectural element to a BPMN process). This allows the system to treat these separate entities as a single integrated model while preserving their original identity and independence.
Multi-Viewpoint Visualization
A critical requirement for EA is providing the right information to the right person. By separating the underlying abstract syntax (the data) from the concrete syntax (the display), the framework supports multiple viewpoints. Using Obeo Designer, the system generates tailored views—such as diagrams, tables, or trees—based on the user's role, all while maintaining a "single source of truth."
Industrial Validation and Insights
The TEAP framework was validated through collaborative industrial projects with partners such as Capgemini, DCNS, and Obeo. Key insights from this experience include:
- MDO as a New Frontier: MDE is evolving beyond code generation to become a vital tool for "Model Driven Organizations" (MDO) managing business complexity.
- Virtualization Superiority: Virtualizing models proved more effective than physical merging for EA, as it maintains departmental independence while providing a global enterprise view.
- The "Messy Data" Hurdle: Unlike source code, EA data is often semi-structured. Developing robust discovery tools for formats like PowerPoint remains a significant technical challenge.
- Social Consensus: The primary difficulty in multi-viewpoint systems is often human, not technical—reaching agreement among diverse stakeholders on the optimal visual representation of data.
Conclusion and Future Directions
The SmartEA implementation demonstrates that MDE can successfully populate and manage an EA repository from heterogeneous industrial data. Future work focuses on enhancing metamodel virtualization to improve usability and developing collaborative tools to help stakeholders define domain-specific modeling languages (DSMLs) and visual syntaxes more effectively.
Summary updated