Multi-objective Optimization Scheduling Model Of Virtual Machine And The Key Algorithms

Virtualization technology is one of the fundamental technologies to build cloud computing environment, especially server virtualization is a powerful technology to put IaaS into effect. Nowadays, as carrier for resource pool of cloud computing, datacenter takes its physical servers to carry virtual servers, while virtual servers are used to carry applications by virtualization technology. As mass physical and virtual resources aggregated together in datacenter, how to reform traditional way to provide and manage IT resources to share these IT resources, improve the whole resource utilization, decrease the cost of service, meanwhile guaranteeing service quality is the urgent important problem by assigning and scheduling the virtual resources.As the number and application load of virtual machines change frequently by requirement of users and applications, static resource allocation will cause resource waste and shortage, otherwise dynamic resource allocation and scheduling will guarantee the quality of service while improving the resource utilization.The essay analyzes the research status about virtual machines scheduling technologies, and summarizes the unsolved issues of present research; then by analyzing the goals and related fundamental factors of virtual machines scheduling, proposes virtual machines scheduling fundamental model, summarize the related determinant, impact factors, influence range, and the fundamental factors, and explore the boundary condition of interactions among the fundamental factors qualitatively and quantificational; then build the index system to estimate the effectiveness of virtual resources allocation and scheduling; and furthermore, proposes resources allocation algorithm based on Fuliye changer and resource scheduling algorithm based on resource utilization feature-matching; and at last, verifies the proposed algorithms are efficient by experimental studies.The main innovations of this essay are as follows:1) Clarifies the relations between the goals to schedule resources of virtual machines and fundamental elements to decide resource scheduling, and proposes "load-resource-power cost" model;2) Builds the indicator system to estimate the effectiveness of virtual resources allocation and scheduling;3) Based on the whole resource requirement rule of datacenter and the feature of resource requirement for virtual machines, proposes resource allocation algorithm based on Fuliye changer to acquire the optimal numbers serial of running servers in datacenter to satisfy the applications load and the resource requirement serial of virtual machines;4) Based on the expressions for performance sensitivity of virtual machines to distinguish the types of virtual machines, proposes the virtual machines scheduling algorithm based on resource utilization feature-matching.