Adaptivity of Business Process Charif Mahmoudi Laboratory of Algorithms, Complexity and Logics, Paris 12th University Créteil, France charif.mahmoudi@u-pec.fr Fabrice Mourlin Laboratory of Algorithms, Complexity and Logics, Paris 12th University Creteil, France fabrice.mourlin@u-pec.fr AbstractEnterprise service bus is a software architecture middleware used for implementing the interaction between software applications in a Service Oriented Architecture. We have developed a strategy to dynamically manage business processes. Administrators of service bus need to reconfigure sites where the business processes are placed. This evolution has to be done during execution of service through the bus. We ensure the availability of process definition. Moreover, business process can also be autonomous. This means a process which is able to move from one site to another one, where the business process engine is installed. This provides another approach to design business process. With our "mobile process migration" template, we separate two concerns, on one side architectural features and on the other side business features. The business process can become mobile between two service busses and we improve the availability of business processes. Keywords-business process; BPEL; orchestration; middleware; message exchange pattern; code migration. I. INTRODUCTION Today, companies have tools to model and automate business processes. This type of tools allows formalizing the company's business rules to automate decision-making, that is to say, the branch of the workflow to choose from, depending on the context. The objective of this initiative is to achieve a better overall view of all enterprise business processes and their interactions in order to be able to optimize and, wherever possible, to automate up with business applications. The lifecycle of a business process can be roughly broken down as follows: design, modeling, implementation, execution, control, and optimization. An approach of Business Process Management (BPM) is based on tools such as a tool for process modeling, tools support the implementation, a runtime loaded to instantiate processes, management tools and reporting. These reports show accurate and relevant indicators on the current deployment of business process definitions. Our first remark is on the lack of scalability of this deployment. Thus, the load of messages that flow through the middleware clearly shows an unbalance that affects the entire information system. So the first point is: how to adapt the workflow running. A second remark is about the number of messages exchanged increases as a function of the initial placement of business process definition. Thus, a business process using local services is less costly in a number of messages than a business process using remote services. Blockings are also less numerous, and, therefore, the execution of a business process is more efficient. This remark highlights the dependencies between two concepts, the location of business processes and its own definition. The designer should not consider his work in the placement constraints. In addition, the administrator cannot take into account all the dependencies of a process definition to find a better placement. Also, our second point is: how separate the two. These conclusions led us to consider an initial configuration of business processes is not satisfactory. This placement must be scalable over time to adapt to client needs. The implementation of this idea is described in this paper through a technical framework described subsequently. It allows the validation of the concepts presented here and provides a sample application. The content of the paper is structured as follows. First, the following section discusses work related to our topic. In Section 3, we provide the definitions on which our work is based. Next, we describe the technical framework of our work. Finally, we provide a simple case study to validate our approach. We end with a point on the goals achieved and those that remain to be addressed. II. RELATED WORK The construction of information systems is usually performed by the department, each business building a subsystem adapted to its own needs and supported by heterogeneous technologies, rarely interoperable. To quickly meet the growing computerization of procedures, systems integration issues has emerged, and with them two questions: How to trigger in response to an event in a given subsystem, a treatment in another subsystem that is foreign? How to ensure consistency and spread data across multiple subsystems? A number of technical solutions have been found to answer these questions. The implementation of these integration solutions is most often done on an opportunistic basis, to meet the immediate goals of a particular application. As these ad hoc solutions have been implemented, the problems of localization or global management have emerged: 19 Copyright (c) IARIA, 2013. ISBN: 978-1-61208-246-2 ICONS 2013 : The Eighth International Conference on Systems