Research On Resource And Service Provisioning With High Efficiently For Massive HRRS Image Under Gloud Environment

As rapid developement of technology on satellite data acquiring and continuous expansion of its application fields in China, along with the characteristics of its massiveness and short periods updating, it is needed to enhance relevant technologies of receiving, processing, filing and dispatching on satellite data responding to the sharp increasing of HRRSI (High Resolution Remote Sensing Image) data and the contradiction of rapid processing. Therefore it is obviously urgent to research and develop a framework and computational model against efficient processing and deployment on massive information for big data HRRSI so as to provide better service for users.The continuous development of computer and Internet technology offers powerful guarantee to process massive data with efficiently. Grid technology is does available to implement heterogeneous resources aggregation then to address the super large scale computing problem. Grid computing makes full use of idle processing capacity and a large number of computable resources on the Internet to form a super computer, and to implement resources all-around sharing, ultimately to solve the problem of low utilization versus plenty of resources, meets the damands of large-scale computing power and massive data processing including HRRSI with big data. Currently, cloud computing as the latest technology is regarded as the promotion of the distribution computing, especially the grid computing. It emphasizes more on service computing. In order to process massive HRRSI with big data and better reflects the service functions, this paper explores a strategy which is fusion the advantages of grid computing and cloud computing to realize the interoperability between GLOUD (grid computing and cloud computing), then builds a GLOUD computing environment. In addition, from the perspective of resource management, this paper proposes a set of effective mechanism on resource expression, resource discovery and resource management, simultaneously further explores the optimal model on the utility of multi-objective under the constraint of QoS parameters, such as resources, cost effectiveness and computational performance, ultimately achieves the goal of the efficient provisioning on resources and services. The main research contents are as follows.1) On the basis of analyzing their respective advantages between grid computing and cloud computing each others, this study proposes a fusion strategy of grid computing and cloud computing to address the problem of HRRSI with big data processing with high efficently. According to the webservice specification, this paper makes research on an integration description mechanism of resources and services, especially the description method of stateful resources, convenient for organization and management of resources and access with high efficently.2) Referencing from the advantages of P2P in distribution, reliability and fault tolerance, this paper studies on a dynamic resource and service discovery algorithm, which based on group spanning tree in P2P under GLOUD to implement the choice with high efficiently for dynamic processing computable resources towards high resolution image data. Then, from the perspective of promotion task efficiency, after logical expressed to the found computable resources, this paper further proposes a resource mapping algorithm based on independent tasks, and forms virtual space of computable resources.3) By analyzing and exploring resources on-demand provisioning on cloud-end and grid computing economy model, this paper proposes an optimal resources and services provisioning model GL-RSPM on the multi-objective optimal utility under the constraint of QoS parameters for users'task request.4) Building an experiment system based on GL-RSPM for disaster application. The system architecture of hardware and software deployed for the system. Under GLOUD environment the system implements registration services for resources and services, meta-service management, workflow and QoS settings, efficient processing of HRRSI data, and so on. Experiments have been done with the algorithms mentioned in this paper, and show that the algorithms apply in the platform is correct and reliable.Study has shown that, from the management and provisioning perspective of resources and services, this paper proposes an integration description method of resources and services, P2P resource and service discovery algorithm and efficient provisioning model for the massive remote sensing image under GLOUD environment, all of them can enable optimal utility between user and resource providers. It provides reference for a large data-intensive computing and resources and services provisioning with efficiently.