Research On Processor Scheduling Algorithm Based On XEN Virtual Machine

In the recent years,with the rapid development of computer hardware and the rise of cloud computing,at the same time virtualization technology which provide the basis for cloud services is also growing.CPU scheduling algorithm is an important resource allocation method in virtualization technology,which has a great influence on the performance of virtual machine.At present there are Xen,KVM and VMware popular on the Virtualization software market.Due to the Xen virtualization management software is open source,therefore it took Xen virtualization management software scheduling algorithm as the research in the paper.First,it analyzed some researches that had done by the domestic and foreign scholars under different application scenarios in the current background which cloud computing and virtualization were developing.Second,it did in-depth source code to study the implementation process according to the different versions of the default scheduling algorithm,and improve the default Credit scheduling algorithm of Xen-4.4.2.version.While it was in I/O-intensive tasks,the Credit scheduling algorithm may cause the virtual CPU stacking of BOOST state to be dispatched without on-time in the run queue,and the virtual CPU run-time slice of the original Credit scheduling algorithm was unreasonable distributed.It proposed an improved algorithm for credit scheduling algorithm in I/O intensive task,after the analysis of the two points.This paper will put forward two suggestions for improving the defects of credit scheduling algorithm,which can not guarantee real-time.First,when there are a large number of I/O tasks,the BOOST state virtual CPU performs load balancing to shorten the system response time.Second,The time slice of the VCPU running in the BOOST state is shortened,and the running time slice of the VCPU in the UNDER state is lengthened.The use of dynamic time slice instead of the original fixed time slice to adapt to the dynamic changes in the virtual CPU.By detecting the average response time and turnaround time of the system to the task,to evaluate the impact of the unimproved and improved response to I/O tasks.Experimental results show that the average response time of the improved Credit scheduling algorithm is significantly shorter than that unimproved,which can greatly improve the performance of I/O delay sensitive applications.