Change Patterns for Supporting the Evolution of Event-Based Systems

Abstract

As event-driven architectures consist of highly decoupled components, they are a promising solution for facilitating high flexibility, scalability, and concurrency of distributed systems. However, the evolution of an event-based system is often challenging due to the intrinsic loose coupling of its components. This problem occurs, on the one hand, because of the absence of explicit information on the dependencies among the constituting components. On the other hand, assisting techniques for investigating and understanding the implications of changes are missing, hindering the implementation and maintenance of the changes in event-based architectures. Our approach presented in this paper aims at overcoming these challenges by introducing primitive change actions and higher-level change patterns, formalized using trace semantics, for representing the modification actions performed when evolving an event-based system. Our proof-of-concept implementation and quantitative evaluations show that our approach is applicable for realistic application scenarios.