摘要
目前Xen下的经典的虚拟机调度算法忽略了服务操作系统的I/O处理能力对系统整体性能的影响,同时未考虑增加延迟敏感型任务的响应速度,为此,设计了一种云计算环境下的改进的Credit调度算法;首先,设计了一种新的Credit值自适应计算方法,然后设计了一种根据服务操作系统的请求处理速度来确定各虚拟客户操作系统和服务操作系统CPU物理时间片的权值确定方法并描述了权值更新时机,最后,对改进的VCPU调度算法进行了定义和描述,在算法中通过优先考虑延迟敏感任务和缓存关联任务作为下一步调度的VCPU以进一步提高算法性能;实验结果表明:文中方法有效实现Xen框架下的虚拟CPU调度,较其它方法相比,将CPU利用率提高了8%以上,是一种有效的适用于云计算的虚拟机调度方法,具有很强的可行性。
The classic virtual scheduling methods ignoring the effect of I/O operating ability on the system performance in Xen and also not considering the response speed of delay sensitive task, an improved Credit algorithm is proposed. Firstly, an adaptive changing method is designed, and a method for assuring the CPU time interval of service OS and client OS based on the operating speed of service operating sys- tem. Finally, the improved VCPU algorithm is defined and described, and the delay sensitive task and cache related task are considered as the priority for scheduling to improve the algorithm performance. The experiment result shows the method in this paper can realize the virtual CPU scheduling, and compared with the other methods, it has improved the CPU usage rate as to 8%, so it is an effective scheduling method for Cloud computing environment with big feasibility.
出处
《计算机测量与控制》
北大核心
2014年第10期3381-3384,共4页
Computer Measurement &Control
基金
校级一般科研项目(YJYB201208)
关键词
调度
虚拟机
权值
资源利用率
seheduling
virtual machine
weigh
resource usage rate