Study On Virtual Resource Scheduling In Cloud Computing

Developed from distributed computing, especially grid technology, cloud computing is newservice computing technology and provides a platform for processing huge amounts of data andservices, addressing the problem of large-scale computing and resource management. As one of thetechnologies supporting cloud computing, Virtualization can build virtual ones on actual hardwareresources to achieve the maximum utilization, and offer the virtual resources to run the servicessimultaneously. How to achieve effective scheduling of virtual resources is a key issue for cloudcomputing. Because resources in cloud computing have characteristics of heterogeneous andpolymorphism, as well as the diversity of customers’ business needs, how to determine an effectivevirtual resource scheduling strategy becomes a big difficulty.Efficient virtual resource scheduling policy can reduce energy consumption effectively. Thepaper focus on achieving a green cloud computing, and proposed a green cloud model which isbased on multidimensional resource analysis (Analysis of Multi-dimensional Resource, AM-R) onthe premise of a common resource scheduling strategy. The basic idea of the virtual resourcescheduling strategy is: Firstly, in accordance to the near linear relationship of CPU powerconsumption rate, the paper proposed energy consumption model of cloud computing. Besides, thenumber of the running physical servers and the relatively low CPU utilization rate serve as astandard to energy consumption. Secondly, for the heterogeneous feature of resources in cloudcomputing and leakage phenomenon caused by uneven resource utilization, the paper proposed amultidimensional resource analysis algorithm which minimizes resource fragmentation andbalances resource utilization to achieve the purpose of reducing the number of physical servers. Thealgorithm defines multidimensional average utilization rate and fluctuating value of resource useper SLA (service level agreement) between cloud computing user and cloud provider, on which itpaints multidimensional space vector graphics, deviding resource space vector into threesubdomains as selective basis of virtual machine deploy. Thirdly, On the basis of multi-dimensionalanalysis algorithms and model meeting the energy consumption, the paper puts forward areasonable strategy of allocation of virtual servers and failover. Finally, through simulationexperiments, the paper verifies effectiveness of virtual resources strategy which is based on AM-Rgreen cloud model.