Research On Trusted Service Resources Planning In Cross-organizational Workflow

Recently, rapid development of global information technology gives promotion to the computational models evolvement. Such as grid computing architecture, cloud computing architecture and service-oriented architecture are used widely. All these architectures help to promote the evolution of collaborative works supported by computers. With the deepening of economic globalization, enterprises are difficult to product in the traditional way, and the strategic alliances are emerging. In order to meet the challenges of globalization, workflow technology cannot be limited in a single field. It forms a cross-organizational trend. The appearance of cross-organizational workflow makes the possibility of business process restructuring between different organizations. In the enterprise alliances, there are different standards in enterprises of different alliances. It makes trust be a hot spot in this area. This thesis focuses on how to arrange the service executing resources more reasonably and more reliably. The novel contributions are as follows:1. Trust attribute is considered when scheduling service resources. It is very important but frequently ignored when cross-organizational workflow scheduling.2. Structure-aware Hierarchical workflow scheduling algorithm(SAH) is proposed in this thesis. Through analyzing the concurrent construction in each level, we can distribute the redundancy time into each level to decrease the overall executing cost of workflow.3. Due to the task executing failure and the server overload problem, the executing resources cannot finish in preset time. This thesis proposes a Structure-Aware Hierarchical Rescheduling Algorithm(SAHR) to address this problem. Analyzing the concurrent construction in each level as well, SAHR reschedules the executing resources that haven't started to reduce the executing cost.In order to verify the effectiveness of the proposed algorithm, we designed several simulated experiments to compare the algorithms used in other papers. The results of simulated experiments reveal that SAH shows better performance.