The Research On Key Technology Of Virtualization Resource Monitoring And Scheduling In Cloud Computing Platform

Recently, the development of computation, storage and communication technologies is movingfast. As a network computing model of commercial resource sharing, cloud computing is believedto be the key technology of the future IT industry. And by using virtualization technology on theplatform, it can provide elastic and scalable resource service. In a cloud computing environment,the virtualized computing resources are encapsulated in virtual machines and to optimize theutilization of computing resources, the monitoring and scheduling on virtual machines are carriedon. Monitoring and scheduling on virtual machines can be the critical part of promoting theperformance of the entire cloud computing platform.The subject of the thesis is to look into the key technology of virtualized resource monitoringand scheduling, and combining the both effectively. The main points we are about to discuss in thethesis are as follows:Firstly, the thesis proposes a framework of cloud computing platform thatcombine virtual machine resource monitoring with scheduling policy by conducting strict analysison Cloud Computing resources which include computing resources, storage resources and networkresources and the relationships between them. Secondly, an intensive study of monitoring module ofvirtual machine resources is presented. This study concerns with the way of obtaining monitoringdata, the distribution of monitoring nodes, and propose a forecasting method of monitoringcomputing resources which is based on the time sequence mondel in Cloud Computing platform.VAR (vector autoregressive model) is used to monitor and forcast the fine-grained of computingresources in consideration of the mutual effect between various monitoring data and the loadpressure of monitoring nodes. Thirdly, a heuristic VM placement algorithm that concerns with usertask scheduling and resource allocating is proposed. This algorithm allocates user tasks based on thenetwork performance and task characteristics, and dynamically placement user tasks according tothe change of network performance so as to improve resource utilization and reduce the executiontime of user tasks.And at last the system is built, and monitoring and scheduling technologies are implemented byusing the way of monitoring, scheduling and system framework of virtualized resource managementproposed above. The feasibility of the algorithm and stability of the system are verified.