Simulation-based experimentation for designing reliable and efficient Web service orchestrations in supply chains

Abstract

The world today is witnessing a growing interest in conducting supply chain business processes electronically using Web services orchestration technology. Fast adoption is often hampered by the need for experimentation to make efficient use of this technology in supply chain processes. In this paper, a simulation-based approach supporting experimentation with the efficiency and reliability of Web service orchestrations in supply chains prior to implementation are presented. The approach simulates the de facto business process specification standard for Web services, BPEL4WS, using Java-based simulation building blocks. A supply chain case study is used to demonstrate and evaluate the approach.

Introduction

The ability to create flexible alliances with partners to form supply chains or business networks becomes more and more important for businesses [30], [35]. When products or services are followed through the network, the successive steps of value addition in the network are often described by the term supply chain [31]. The choice of the term supply chain is unfortunate because it might result in confusion. The term “chain” is often associated with linear invocations and static images. The term supply is inappropriate and suggests mass production rather than the creation of customized business processes. Better names like demand network or customer driven Webs have been proposed. Yet, the name supply chain holds its own [15]. Strictly speaking, a supply chain is a network of multiple businesses and relationships [18]. The creation of flexible, temporarily arrangements result in the creation of business processes that are no longer self-contained within a single organization. The reliability of business processes depends more and more on the performance of external partners that are often unknown and viewed as black boxes. In addition, the dynamic nature of supply chains, for example, the changing number and/or types of partners, and the involvement in several networks, increases the difficulty and complexity to understand the dependencies in the supply chain. Consequently, the re-design of the business processes in supply chains becomes difficult, as the impact of changes cannot easily be determined. More insight into business processes spanning multiple organizations is needed.

Nowadays, one of the most promising technologies for ensuring interoperability at a process level is Web service orchestration (WSO) [39]. Web services are self-contained, Web-enabled applications capable of performing some kind of functionality, and possessing the ability to engage with other Web services to complete higher-order business transactions [38]. WSO coordinates the sequence of Web service invocations. WSO and Web services technology enable loosely-coupled business processes and applications with much less effort compared to previous technologies. However, acquiring the true value of efficient and reliable orchestration of the various Web services based business processes remains a challenge. There is a lack of experience reports, simulation experiments and case study material concerning WSO [12]. Moreover, in practice, it often requires a lot of experimentation to leverage the benefits of WSO to support businesses processes [e.g., [9].

Inefficient and unreliable WSO can easily result in inordinate waiting times for multi-step business processes, a lack of synchronization of events, limited coordination and cooperation, and excessive control at high operational costs. Consequently organizations are often in need of instruments they can use to speed up the design process and improve application effectiveness [9]. Supply chain performance can significantly benefit from decision-making processes that dynamically evaluate design options [2]. Discrete-event simulation is a suitable decision-support instrument, as it allows decision-makers to understand the essence of business systems, to identify opportunities for change, and to evaluate the effect of proposed changes on key performance indicators [4], [19]. In discrete-event simulation models the passage of events during time plays a crucial role. Chandrasekaran et al. [5] simulated Web service compositions and concluded that discrete-event simulation provides many benefits if experimentation in practice is not feasible or too costly. They did not investigate the use of Web service orchestration technology nor did they research the use of Web services in organizational settings such as supply chains.

In this paper, a simulation-based approach is presented that is aimed at improving the efficiency and reliability of Web service orchestration in supply chains. This paper is structured as follows. In Section 2 Web service orchestration, service-oriented architecture, Web service technology and simulation are discussed. Thereafter, we present our simulation-based approach. In Section 4 this approach is demonstrated and evaluated using a case study. In Section 5 conclusions and recommendations for future research are drawn.

Our main finding from the research is that our simulation approach provided a suitable instrument to experiment with Web service orchestration in supply chains prior to implementation. Using the approach it was possible to gain insight into the dynamics of Web service orchestrations, to assess the efficiency of the Web service orchestrations, and to conduct “what if” analyses to estimate the reliability and the effect of a service failure on to the overall orchestration.