An Adaptive Scheduling Method of Weight Parameter Adjustment on Virtual Machines

In the virtual machine architecture including Service OS and Guest OSs, Guest OS can access the real hardware by the aid of Service OS. But the optimizations of the current scheduling algorithms are focused on I/O-intensive domains. They neglect CPU-intensive domains and the effect of the I/O processing capacity of Service OS on the whole performance of virtual machines. Aiming at these problems, an adaptive scheduling method of parameters on virtual machine based on credit scheduling algorithm is given in this paper. It takes the I/O processing capacity of Service OS as an important parameter and pays attention to I/O-intensive domain and CPU-intensive domain, which is advantageous to make system resources get used more reasonably. The experiment results show that the proposed method can make the performance of CPU processing and accessing real hardware of Guest OS better than that of some given ones, such as Credit scheduling algorithm, through the reasonable adjustment of virtual machines’ weight according to the current I/O requests of Guest OSs and the processing capacity of Service OS. The scheduling algorithm of virtual machines has direct effect on the performance, and it is a significant job to go into scheduling algorithms of virtual machines.