Abstract (eng)
Recent trends in technology, business and society like cloud and mobile computing lead to increasingly distributed applications and processes requiring integration. Enterprise Application Integration (EAI) offers integration capabilities using message passing in form of integration scenarios and constitute a centerpiece of current IT architectures. Due to the complex nature of EAI, the integration scenario logic builds on abstractions derived by best practices, collected as Enterprise Integration Patterns, denoting the current foundation of developing and modeling EAI logic and solutions.
However, due to increasing distribution, not only the number of communication partners, but also the data volume and velocity requirements grow beyond processing capabilities of current systems. Further, personal digitalization trends like social media computing introduce increasing amounts of non-textual data beyond current variety capabilities and add trust requirements into the functional correctness and reliability of technology that is now used by non-technical users. These problems cannot be solved with current informally described integration pattern foundation or implementations.
This thesis defines formal foundations for EAI that enables responsible development of integration solutions, helps businesses to trust their correctness, and shows how to build effective and efficient integration solutions. First, EAI foundations in form of EIPs are revisited regarding their actuality and comprehensiveness in the context of emerging trends, and captured as an extended pattern catalog. Patterns serve as a basis for the formalization of integration patterns and their compositions as integration scenarios. Then the latter are studied for correctness-preserving optimization strategies, ensuring the efficiency of integration logic on the composition level. A prototype realization allows for the formal analysis of integration scenarios, thus addressing the trust challenge. To address the volume, velocity and variety challenges more efficient pattern solutions are conceptually developed using new technology trends and evaluated using a newly elaborated EIP benchmark.
In summary, this thesis builds a formal foundation for EAI and shows how this can be used to meet the trust, volume, velocity and variety requirements of current integration systems.