A proposal to detect errors in Enterprise Application Integration solutions

Abstract

Enterprise Application Integration (EAI) solutions comprise a set of specific-purpose processes that implement exogenous message workflows. The goal is to keep a number of applications’ data in synchrony or to develop new functionality on top of them. Such solutions are prone to errors because they are highly distributed and usually involve applications that were not designed with integration concerns in mind. This has motivated many authors to work on provisioning EAI solutions with fault-tolerance capabilities. In this article we analyse EAI solutions from two orthogonal perspectives: viewpoint (orchestration versus choreography) and execution model (process- versus task-based model). A review of the literature shows that current proposals are bound to a specific viewpoint or execution model or have important limitations. To address the problem, we have devised an error monitor that can be used to provision EAI solutions with fault-tolerance capabilities. Our theoretical analysis proves that the algorithms we use are computationally tractable, and our experimental results prove that they are efficient enough to be used in situations in which the workload is very high.