基于Xen的多处理器负载均衡调度算法
发布时间：2018-09-12 15:26
【摘要】：虚拟化技术是创建灵活动态的企业级设施架构的关键机制。随着多处理器技术的发展,计算能力有了很大提高,也加速了虚拟化技术这一关键机制的发展。虚拟化技术可以屏蔽底层复杂的物理环境,允许多个虚拟机之间共享物理机资源,另外,正在运行的虚拟机可以无间断地在两台物理机之间进行迁移。所以对于大多数企业来说,基于虚拟化技术的企业架构能够提供更多的安全稳定性能,并且也是合并服务器资源和应用的理想解决方案之一。 Xen是目前应用比较广泛的开源虚拟化软件之一,它能够使多个客户操作系统以接近物理系统的性能同时运行在相互隔离的环境中,并对底层物理硬件实现共享访问,包括处理器资源。在虚拟化环境中,虚拟机上的CPU调度是以虚拟CPU作为调度单位进行分时调度。SEDF是Xen中缺省的CPU调度算法之一,该算法支持实时性较强的应用,但是缺少SMP架构中多处理器之间负载均衡的控制。 本文考虑到SMP架构中处理器和cache的关系,提出一种虚拟机环境中的CPU调度算法,该算法设计了一个共享等待队列,各个处理器根据自身负载情况,共享等待队列动态调整VCPU的调度顺序,不仅能够选择最优处理器来执行,并且可以避免处理器空闲。仿真测试结果表明,本文提出的调度算法具有较高的系统吞吐量和任务完成率,也能够降低任务执行的平均周转时间和平均响应时间。本文提出的算法不仅减少了cache和内存数据同步带来的性能损耗,而且支持较好的全局负载均衡,能够充分合理利用处理器资源。
[Abstract]:Virtualization is a key mechanism for creating flexible and dynamic enterprise infrastructure. With the development of multiprocessor technology, computing power has been greatly improved, which also accelerates the development of virtualization technology as a key mechanism. Virtualization technology can shield the underlying complex physical environment and allow multiple virtual machines to share physical machine resources. In addition, running virtual machines can migrate between two machines without interruption. So for most enterprises, an enterprise architecture based on virtualization can provide more security and stability, Xen is one of the most widely used open source virtualization software. It enables multiple client operating systems to run simultaneously in isolated environments in close proximity to the performance of physical systems and to share access to underlying physical hardware including processor resources. In virtualized environment, the CPU scheduling on virtual machine is time-sharing scheduling with virtual CPU as scheduling unit. SEDF is one of the default CPU scheduling algorithms in Xen, and this algorithm supports real-time applications. However, there is a lack of load balancing control between multiprocessors in SMP architecture. In this paper, considering the relationship between processor and cache in SMP architecture, a CPU scheduling algorithm in virtual machine environment is proposed. The shared wait queue dynamically adjusts the scheduling order of the VCPU, which can not only select the optimal processor to execute, but also avoid the idle processor. The simulation results show that the proposed scheduling algorithm has higher system throughput and task completion rate, and can also reduce the average turnaround time and average response time of task execution. The proposed algorithm not only reduces the performance loss caused by the synchronization of cache and memory data, but also supports better global load balance and can make full and reasonable use of processor resources.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP332
本文编号：2239447
[Abstract]:Virtualization is a key mechanism for creating flexible and dynamic enterprise infrastructure. With the development of multiprocessor technology, computing power has been greatly improved, which also accelerates the development of virtualization technology as a key mechanism. Virtualization technology can shield the underlying complex physical environment and allow multiple virtual machines to share physical machine resources. In addition, running virtual machines can migrate between two machines without interruption. So for most enterprises, an enterprise architecture based on virtualization can provide more security and stability, Xen is one of the most widely used open source virtualization software. It enables multiple client operating systems to run simultaneously in isolated environments in close proximity to the performance of physical systems and to share access to underlying physical hardware including processor resources. In virtualized environment, the CPU scheduling on virtual machine is time-sharing scheduling with virtual CPU as scheduling unit. SEDF is one of the default CPU scheduling algorithms in Xen, and this algorithm supports real-time applications. However, there is a lack of load balancing control between multiprocessors in SMP architecture. In this paper, considering the relationship between processor and cache in SMP architecture, a CPU scheduling algorithm in virtual machine environment is proposed. The shared wait queue dynamically adjusts the scheduling order of the VCPU, which can not only select the optimal processor to execute, but also avoid the idle processor. The simulation results show that the proposed scheduling algorithm has higher system throughput and task completion rate, and can also reduce the average turnaround time and average response time of task execution. The proposed algorithm not only reduces the performance loss caused by the synchronization of cache and memory data, but also supports better global load balance and can make full and reasonable use of processor resources.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP332
【参考文献】
相关期刊论文 前5条
1 孟江涛;卢显良;;虚拟机监控器Xen的可靠性优化[J];计算机应用;2010年09期
2 怀进鹏;李沁;胡春明;;基于虚拟机的虚拟计算环境研究与设计[J];软件学报;2007年08期
3 时光;郭玉东;王晓睿;李勇;;Xen中VCPU调度算法分析[J];计算机工程与设计;2010年18期
4 姚文斌;郑兴杰;;一种改进的SEDF调度算法[J];小型微型计算机系统;2010年03期
5 宋振华;杨亚军;;Xen虚拟机间的磁盘I/O性能隔离[J];小型微型计算机系统;2011年08期
相关硕士学位论文 前3条
1 曹欣;半虚拟化技术分析与研究[D];浙江大学;2008年
2 阮敏;Xen环境下实时迁移结构和算法研究[D];大连海事大学;2009年
3 沈霆;多核虚拟化的性能分析和改进[D];上海交通大学;2010年
，本文编号：2239447
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2239447.html
