摘要
对于Xen网络工作模式来说,高负荷网络环境下大量的CPU资源将会被频繁到达的网卡中断占用,导致系统遭遇性能瓶颈,提出一种可以根据网络繁忙程度进行轮询的中断自适应调整机制,通过定义面向两种优先级不同的虚拟缓冲区,根据缓冲区中数据队列长度进行判断处理,以保证不同业务的服务质量。该机制不需要增加额外的处理单元。实验表明,当外部网络数据频繁到达时也不会引发过多的CPU中断,从而提高了CPU的利用率。
For the Xen network model, high load under the environment of network causes bottle- neck of the system where most CPU resources are occupied by frequent interrupt from the NIC. To alle- viate this problem, an adaptive interrupt latency scheduling mechanism based on Xen is proposed, which uses the polling or interrupt method in accordance with the queue length of virtual buffer without sup- plementing any additional processing units. And the mechanism can guarantee different qualities of service to some extent through the definition of the two types of priority virtual buffers. Experimental results show that, so that the CPU when frequent interrupt from the NIC occurs, CPU will not be interrupted frequently utilization is improved.
出处
《计算机工程与科学》
CSCD
北大核心
2014年第10期1937-1942,共6页
Computer Engineering & Science
基金
河南省科技攻关资助项目(132102210491)
关键词
虚拟化
中断
延迟
虚拟缓冲区
virtualization
interrupt
latency
virtual buffer