基于OpenStack云平台的资源调度技术研究

发布时间：2018-06-16 21:09

本文选题：云计算 + OpenStack　；参考：《南京邮电大学》2017年硕士论文

【摘要】：云计算技术的发展已经改变了传统IT架构,它给传统物理资源的管理和利用带来了革命性的变化。随着云计算技术的广泛使用,如何让服务器集群中的资源得到充分均衡的利用已成为资源调度研究的热点问题之一。通过改进虚拟机资源调度策略来解决上述负载不均衡度问题,同时还能够改善资源利用率和减少能耗。本文主要基于OpenStack云平台的虚拟机资源调度策略进行研究。首先,本文研究了基于网络带宽感知的虚拟机调度策略。当前主流的OpenStack开源云平台虚拟机调度机制是基于CPU、内存和存储资源来进行部署的,而没有考虑对网络带宽资源的合理调度,因此,本文考虑了网络带宽资源的约束问题,完善了OpenStack云平台针对网络带宽需求的虚拟机调度策略,均衡了网络带宽资源的利用。仿真结果表明,本文提出的调度策略与OpenStack默认的调度策略相比,更有效的提升了虚拟机实例的网络吞吐量。其次,针对OpenStack云平台中多目标约束问题,提出了基于粒子群和蚁群的多目标融合算法的虚拟机调度策略,该策略能够在多个相互矛盾的目标中寻求折中。多目标融合算法是对粒子群和蚁群算法的改进,本文提出了关于物理集群的资源浪费、能源损耗和负载不均衡度模型,通过算法融合来加速迭代搜索虚拟机映射到物理主机的Parato最优解,仿真表明,本文提出的多目标融合算法与传统的元启发式算法粒子群算法和蚁群算法相比,能够改善多维资源利用率、降低能耗和负载不均衡度。最后,本文设计并实现了基于OpenStack的虚拟机调度平台。主要对监控模块和数据库模块进行了设计。该平台能够对虚拟机和物理主机资源使用情况进行监测,通过虚拟机调度算法来优化下一次的位置部署,从而提升多维资源的利用率和均衡负载。
[Abstract]:The development of cloud computing technology has changed the traditional IT architecture, which has brought revolutionary changes to the management and utilization of traditional physical resources. With the wide use of cloud computing technology, how to make full and balanced use of resources in server clusters has become one of the hot issues in resource scheduling research. The problem of load imbalance can be solved by improving the resource scheduling strategy of virtual machine. At the same time, it can also improve resource utilization and reduce energy consumption. This paper mainly studies the virtual machine resource scheduling strategy based on OpenStack cloud platform. Firstly, this paper studies the virtual machine scheduling strategy based on network bandwidth awareness. The current mainstream OpenStack open source cloud platform virtual machine scheduling mechanism is based on CPU, memory and storage resources to deploy, without considering the reasonable scheduling of network bandwidth resources, therefore, this paper considers the constraints of network bandwidth resources. The virtual machine scheduling strategy of OpenStack cloud platform is improved to meet the demand of network bandwidth, and the utilization of network bandwidth resources is balanced. Simulation results show that compared with OpenStack's default scheduling strategy, the proposed scheduling strategy improves the network throughput of virtual machine instances more effectively. Secondly, aiming at the multi-objective constraint problem in OpenStack cloud platform, a virtual machine scheduling strategy based on particle swarm and ant colony fusion algorithm is proposed. Multi-objective fusion algorithm is an improvement on particle swarm optimization and ant colony algorithm. In this paper, a model of resource waste, energy loss and load imbalance for physical cluster is proposed. The algorithm fusion is used to accelerate the iterative search virtual machine mapping to the Parato optimal solution of the physical host. The simulation results show that the proposed multi-objective fusion algorithm is compared with the traditional meta-heuristic algorithm particle swarm optimization algorithm and ant colony algorithm. It can improve multi-dimensional resource utilization and reduce energy consumption and load imbalance. Finally, this paper designs and implements a virtual machine scheduling platform based on OpenStack. The monitoring module and database module are designed. The platform can monitor the use of virtual machine and physical host resources, optimize the next location deployment by virtual machine scheduling algorithm, and improve the utilization and load balance of multi-dimensional resources.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【参考文献】