Research And Application Of Computing Resource Management In Cloud Service Platform

With the rapid development of emerging internet technologies such as cloud computing and virtualization,the cloud computing platform of IaaS(Infrastructure as a service)model has been widely used in various industries.Nowdays more mainstream IaaS cloud platforms,such as OpenStack and Eucalyptus which is mainly allocation and management the physical machines(PMs)resource through statistics and distribution of the PMs,rather than the resource occupation.Therefore,it cannot meet the requirements of high efficiency,dynamic and scalability of cloud computing service platform,and cannot accurately predict the fluctuation of infrastructure load caused by the change of cloud platform business demand scale.Aiming at the problems of unbalanced resource load and untimely resource adjustment in traditional cloud computing,this paper designs and implements a cloud-oriented computing resource management system.The main work is as follow:1.Firstly,this paper designed a resource scheduling process and algorithm of the system,and optimized the detection of abnormal state in the physical host based on the doublethreshold detection algorithm which makes the results more accurate and robust.As for the virtual machines(VMs)selection algorithm,this paper sorted and grouped the VMs based on their cpu utilization to make the scheduling decision to reduce time-efficient and the overall time complexity of the algorithm.And then,based on the traditional descending optimal fitting algorithm,we design a best fitting method to choose the destination machines by considering of the energy efficiency.In the end,a comparison was made with MBFD(Modified best-fit reduction)on the simulation platform and it turns out our algorithm make a good improving on resource utilization and energy efficiency.2.Secondly,this paper designed a resource management system for cloud service platform based on the idea of discovering resources first and then making behavioral decisions.The design and implementation of this system is mainly based on OpenStack,and the system includs the resource management subsystem which is the core of the system and the Web Services subsystem which provides a convenient and efficient resource management visualization interface for cloud service platform.3.Thirdly,this paper finished the design and implementation of each component module of the resource management subsystem and the Web service subsystem.Then we built a resource management platform based on OpenStack,and carried out the functional test and analysed the test result for each module.