The Research And Implementation Of Cloud Computing Key Technology Based On Virtualization

With the development of cloud computing and the arrival of the era of big data, more and more data analysis and data mining tasks use the cloud computing technology. This situation leads to the rapid development of cloud computing platforms, in which the Hadoop platform that based on MapReduce is most popular. More and more scholars begin to focus on cloud platforms, and the current studies mainly focused on three aspects, load balancing, the average system response time optimizing, and the choice of fixed cloud resources under the user’s request.Hadoop cloud platform based on MapReduce has many advantages, but in some specific problems it needs researches and optimization. In the problem of parallel processing of pipelined tasks, task scheduling and resource pool, specific analysis is necessary. Therefore, this paper improved the cloud computing model from the perspective of pipeline tasks, based on in-depth analysis of cloud computing platform, virtualization technology, the model and task scheduling algorithm of MapReduce. After that, this paper discussed the technic of Map-Reduce-Reload system:different priorities and different needs of the task scheduling techniques; providing technical differences for different level users under resource pools service.The specific research work is as follows:(1) Introduction of the research background, present situation, content and meaning of virtualization and cloud computing technology. We provide an overview of the present situation of researches, which leads to the contents of this paper. Do research in the cloud platform, virtualization technology and task scheduling algorithm, which is the basis for the next section.(2) This section analyzes shortcomings of Hadoop in pipelined tasks, which is a mainstream instrument of cloud computing. Based on the MapReduce model, Map-Reduce-Reload (hereinafter referred to as MRR) model is proposed to solve the problem. The effectiveness of the proposed Map-Reduce-Reload model in handling pipelined tasks is verified by the results of contrast experiment.(3) On the basis of the subsection (2), this section analyzed the technic of MRR systems. We use the priority-based random multiple demands scheduling algorithm to solve the integrated balance of priorities and multiple demands. At the same time, in order to provide various services for different user levels, we puts forward the choice value and resources matching switching scheme to ensure the VIP users get high-quality resources.This paper finally developed MRR based cloud platform system, based on the analysis of the improved MRR model and related technologies. Through the system test, we proved that the system reached the expected requirement in both tasks and resources aspects. This also shows that the improved model and system research in this paper has certain theoretical significance and value in the practical application.