The Design And Implementation Of Cloud Platform Scheduling Algorithm Based On Openstack

With the rapid progression of the Internet development,cloud computing technology has got attention, researched and progressed by countries in the world, including China. Cloud computing has several advantages including high reliability, meet-the-needs allocation, dynamic allocation, etc.In order to meet the requirements of centralized deployment and hardware/maintenance costs reduction, telecom operators like China Mobile have launched the construction of cloud platforms.Resource requirements often increase as the development of business and expansion of the users. However, platform resources provided to created virtual machines are hardly to meet the growth in application resource requirements according to the. feature of allocation. To solve this problem, commin IaaS platforms such as OpenStack, VMware offer live migration feature to migrate applications to another, virtual machine which can provide more resources. However, applications of telecom operators are mostly real-time, which will be effected by time cost during the progress of migration,therefore causes unnecesssary losses.As a result, a reasonable scheduling algorithm designed in telecom operators’ application scenarios is useful, which imporves overall platform resource utilization while ensuring the stability of business operations, reflacting the value of reducing costs and ensuring reliability.This paper,based on the Ministry of Education-China Mobile Research Fund for2012projects:"Key tecfhnology and business support scheme of the Internet-oriented system",designs and implements a two-phase scheduling algorithm on top of OpenStack,a open source cloud platform framework. First, this paper brings forward the business model based on business size and lifecycle, and with which designs a two-stage virtual machine resource scheduling algorithm.Second, this paper researches the implementation of OpenStack scheduler, implement the scheduling algorithm module used to evaluate the performance of the algorithm. Finally,this algorithm is tested by analyzing the test results to verify the availability of scheduling module and the effectiveness of the algorithm.