Research On Several Key Technologies Of Service Resource Management In Cloud Computing Environment

With the rapid development of computing technologies, Internet technologies and Service-Oriented Architecture (SOA), the new generation cloud computing technology, the core of which is the service concept, has become a hot spot and frontier areas of the current international research. There is no doubt that cloud computing promotes the socialization, intensive and specialized accelerating transition of the information technology. Cloud computing flexibly and effectively facilitates the information technology to the resource consolidation, information sharing and cooperation within and between multiple industry chains, and open up an extremely new road which leads to service-oriented computing era. Cloud computing also facilitates evolution from the core simple and edge rich Internet to the network rich, edge simple and interaction intelligent cloud computing center. Besides, through the dynamic reuse and flexible restructure of software, cloud computing center can support optimization and reconstruction of service processes, and promote the utilization of resources. Therefore, on-demand and efficient use of cloud computing driven service resources has drawn much attention of both academia and industry in recent years.However, the performances of cloud computing services, including virtualization, autonomy, dynamic, availability and scalability, etc., lead to more difficult key problems for the analysis and decentralized execution of process services model or the dynamic provisioning of infrastructure service resources in cloud computing environment. Therefore, traditional computing models can not satisfy the new requirement of cloud computing architecture. To this end, this paper presents a survey of SOA, Web services composition and infrastructure service resources. The following issues about compatibility analysis of process services composition, controller-aided composition, model fragmentation for decentralized execution, and performance optimization of infrastructure services are proposed as the focus of this dissertation.Firstly, a Petri net based optimal control policy is proposed for the check of behavioral incompatibility in web services composition. With a real case of multiple services interaction, we analyze the compatibility, and then develop a controller to make them compatible without changed their internal logic. In this paper, a formalized definition of controlled service composition is presented, and the reduced state reachability graph of service composition net is also derived according to the reduce rules, thus the deadlock states and the deadlock-free states are identified. With the maximally permissive feedback control strategy developed, the appropriate control place and arc are appended in the key transition which can lead to deadlock states. Thus the appropriate optimal controller is developed, as to verify the proposed approach. In addition, from the behavioral incompatibility case, a policy of appending optimal controller is presented. And the experiments prove our policy to be a good solution.Secondly, to solve united process customization and collaboration problems of various organizations, dynamic service process models fragmentation approaches were proposed based on multiple Web Services-Business Process Execution Language (WS-BPEL or BPEL) execution engines. Here, horizontal fragmentation and vertical fragmentation approaches based on high-level Petri nets were proposed, and a whole BPEL model was partitioned into several self-contained implementation sub-chips to be assigned on multiple BPEL execution engines respectively according to their organizations and roles. Furthermore, a hybrid fragmentation approach of Petri nets was presented, so as to increase throughput as well as reduce the cost for service response time. The correctness of the proposed model fragmentations was also verified. At the same time, this paper provided allocation policies to fragmentation process. Then an instance was presented to illustrate that the proposed hybrid fragmentation approach for the process implementation is feasible. The results of experiment show that hybrid fragmentation is the most effective fragmentation methodology.Thirdly, dynamic and scalable resources provisioning is necessary for the multi-tier different virtualized application services in shared cloud data centers to meet business requirements of different clients as well as get the maximum revenue of service resources provider, and to meet quality of service for clients. For an appropriate provisioning mechanism, we provided cloud infrastructure service architecture based on virtualized mechanisms for multi-tier applications, by monitoring virtualized service resources, guaranteed the dynamic provisioning and dynamic allocation of virtualized service resources, and guaranteed effective execution of virtualized application services while meeting Service Level Agreement (SLA) requirements of clients. Then, we propose a novel dynamic provisioning technique and develop a flexible hybrid queuing model for virtualized application services to determine the virtualized resources to be provisioned to each tier of the virtualized application services, for satisfying with the optimal utilization of virtualized resource under the varied workload condition for clients from different levels. We further developed meta-heuristic solutions based on the mixed stochastic optimization algorithm, which is according to different performance requirements of clients from different levels. On the basis of high quality initial configuration of virtualized resources, the algorithm can ensure the maximum profit of cloud IaaS. Simulation experiment results show that these proposed approaches can provide appropriate way to provision cloud infrastructure service resources, especially for improving the overall performance while effectively reducing the resource usage extra cost.Finally, to validate the proposed methodologies, this paper designs the technical architectures of the compatibility analysis and fragmentation system for process services, and provides the performance optimization of infrastructure services. The implementations of the key components of the prototypes are described. The prototypes have been successfully applied in the actual projects, which validate the feasibility and effectiveness of the research results of this paper.In this paper, analysis and optimization are proposed for cloud computing based, on business process services and infrastructure services. This research is much significant in development of cloud computing theory and implementation of cloud services system R&D.