Research On Virtualization Resources Scheduling Technology

By utilizing the virtualization technology, the cloud data center can consolidate the heterogeneouscomputing resources, storage resources and network resources to build a dynamic virtualizedresources pool. The cloud data center exploits the resources scheduling and management algorithm toachieve the automatic deployment of resources, dynamic extension of resources, and allocation ofresources on demand. Therefore, research on the technology of virtualized resource scheduling andmanagement is of great significance to improve the resource utilization, save the energy consuming,and reduce operating costs.This thesis focuses on virtualization resources scheduling technology. We propose a dynamicconstraint programming based virtual cloud resources allocation model (VCRA-CP), which takesmeeting users’ QoS requirements and the costs of virtual cloud resources into account. Thesimulation results prove that the model can significantly reduce QoS violations and achieve lowerresource usage costs. In order to reduce the influence of subjective factor to some extent, we design amethod to calculate the resources cost evaluation index weight, which is based on Analytic HierarchyProcess (AHP). For the high energy consumption of the cloud data centers, we propose a balancedscheduling algorithm for virtual cloud resources based on dynamic power-aware (BSA-DPA), whichnot only takes into account how to effectively reduce the power consumption, but also considers howto keep the cloud data centers balanced and meet the users’ QoS requirements. The simulation resultsprove that the algorithm can significantly minimize the amount of violations of QoS and effectivelyreduce the energy consumption. On the basis of existing hosts overload detection strategies, we designthe multi-dimensional host overload detection strategy (MHOD) and the simulation resultsdemonstrate the robustness and effectiveness of the proposed strategy.By using the MyEclipse development platform, we design and complete the CloudSim basedsimulation platform of virtualized resource scheduling, which is designed to verify the effectivenessof the model and algorithm proposed in this thesis. The simulation results prove that the proposedmodel and algorithm have the advantages of improving the resources utilization, reducing the energyconsumption and resources costs, and meeting users’ QoS requirements as much as possible.