Dynamic Virtual Resource Scheduling Based On The Interference Effect Analysis

With the rapid development of the computer world, cloud computing is becoming a hot topic in computer science. Cloud computing is a popular form through which the service providers supply their service resources to end-users. It shares hardware and software resources to computer and other devices by demand. Based on the virtualization architecture, cloud computing dynamically allocates the resources to end-users. As a business model, cloud computing is involved in the cost problem unavoidably. It will lower the cost of the usage of the resources if we enhance the efficiency of the usage of the resources. So it is important to utilize the resources in the cloud environment.In the dynamic management of virtualized resources, the scheduling of the virtualized resources is the main problem. From the real world, the requirement of the virtualized resources originates from the real business and workload, which makes the demands from end-users dynamically changed. Cloud service providers try to enhance the efficiency of the usage of the resources by reasonable resource scheduling. Therefore, it is very meaningful to research how to allocate resources to end-users efficiently.The interference problem means the contention when multiple virtual machines run on the same physical machine. It is indispensable to focus on the virtual machine interference problem when constructing the resource scheduling process. A heavy interference problem will cause the less efficiency after the resource scheduling process. As a result, the analysis of the detail of the virtual machine interference problem is very significant. At the same time, it is very intuitive to take the result of the resource scheduling into consideration through the detail process of the resource scheduling.When constructing the analysis of the interference problem, we not only need to know the visualized effect of the whole but also the impact from the singles to each other. The effect lowers the efficiency of the usage of the virtualized resources, enhances the cost of the virtualized resources and the most important, unfriendly to end-users. Recent researches have many studies about the scheduling problem of the virtualized resources and theory analysis of the interference problem. However, few researches concentrate on how to define the interference effect accurately and how to introduce the interference effect into dynamically virtualized resources management, which is pretty much the agenda.Aimed at the above problems, this paper first analyzes the drawbacks of the existed dynamically virtualized resources management framework and emphasizes some potential problems in the cloud clusters. Then we concentrate on the building of the interference model with the effort to quantify the interference effect among virtual machines. After we get the model of the interference effect, we propose a enhanced dynamically virtualized resources management framework which contains the new virtual machine placement strategy and the consolidation strategy. At last, this paper verifies the model and the strategy we propose in some scenes. The result of the experiments shows that our solution defines the interference phenomenon reasonably, lowers the interference impact in the cloud cluster and enhances the efficiency of the usage of the dynamically virtualized resources.