针对多核环境下的现有实时虚拟机VCPU(virtual CPU)调度算法不能在保证任务可调度的基础上高效地使用CPU资源的问题,提出一种改进的实时虚拟机VCPU调度算法LBP-EDF(基于独立队列的可负载均衡的最早截止时间优先调度算法).该算法采用了独立队列的方式,以避免共享队列带来的竞争开销.同时算法能够识别紧急任务,并通过负载均衡的机制及时地将紧急任务迁移到负载较轻的CPU上执行,在保证任务可调度的基础上,同时达到充分使用CPU资源的目的.实验结果表明:在保证任务可调度的基础上,LBPEDF较其他算法有更高的CPU利用率.
To address the problem that existing real-time virtual machine VCPU (virtual central processing unit) scheduling approaches in multicore environments cannot use CPU resources efficiently with a guarantee of schedulability, an improved real-time virtual machine VCPU scheduling scheme named LBP-EDF (load balanced partitioned earliest deadline first scheduling algorithm) was presented. The proposed scheduling scheme adopted a partitioned-based pattern to avoid system competition overhead caused by global-based approaches. Meanwhile, the proposed scheduling scheme was able to detect urgent tasks and move the urgent tasks to low-load CPUs through a load balancing technique. With this feature, the proposed scheduling scheme guaranteed the schedulability and improved CPU utilization. The experimental results demonstrate that the LBP-EDF achieves higher CPU utilization than the existing approaches with the guarantee of schedulability.