How can enterprise architecture repositories support large-scale collaborative modelling across distributed teams without becoming unmanageable?
Collaborative modelling is well established as a more effective approach than individual analysis — but as enterprise architecture programmes scale to involve hundreds of contributors across distributed teams, the tools and repositories supporting them face a different order of challenge. This paper draws on experience deploying a web-based collaborative EA modelling platform with a global IT services organisation managing a transformation involving over ten thousand applications, and identifies the core problems: information overload, ownership and rights management, version conflicts, multi-language support, and the need to present radically different views to different user communities.
To address these challenges, the paper formalises a set of extensions to meta models and meta-meta models — including the concepts of context, domain, filter, version, scenario, opinion, and journaling — and presents a unified conceptual model integrating them. These constructs allow teams to control what information is visible to whom, manage competing versions of the truth, track changes non-destructively, and support iterative refinement from personal ideas through to ratified corporate policy.
Grounded in both prototype work and production experience, the paper offers a rigorous theoretical foundation for architects and tool builders working on the next generation of scalable, collaborative enterprise modelling environments.
Pages: 9
Originally published as a conference paper by Graham McLeod at the EMMSAD 2008 workshop (co-located with CAiSE), 2008.