A stochastic service composition model for business integration

Abstract

One of the key challenges for contemporary enterprises is how to integrate business functions and IT systems within and across the enterprises. Among other approaches, the Service-Oriented Architecture (SOA) has emerged as a prominent candidate to ensure the resulting business integration in a flexible, simple and technology-agnostic manor. A service is defined as a particular business unit of work or IT unit of system that is accessible by independent users and systems through a standardized interface. Therefore, business functions and IT systems are refactored to meet the requirements of SOA. As a result, business integration is modeled as a composition of atomic or composite business services and IT services. However, with their very nature, the results of using a business or IT service can be non-deterministic, i.e., one can't predict ahead of time the outcomes of invoking a business or IT service. For instance, a business service might provide a result short of the original expectation, which would require the use of another business service to compensate for the shortfalls. Consequently, the service composition model should also take the stochastic aspect into consideration. In this paper, we are going to propose a stochastic service composition model for business integration. This model will extend the existing service composition model to take the stochastic aspect of the services into explicit consideration. We will discuss the structure and potential usage of such a model. We will then provide an example to illustrate how a stochastic service composition model can be utilized in a real-world setting for business integration. © 2005 IEEE.