I/O performance is an important factor in virtualization technology. However, the hypervisor and the scheduler brings impact on the I/O performance, since the scheduler considers I/O-intensive Virtual Machine (VM) and CPU-intensive VM as the same VM which results in scheduling latency. Most researches address this I/O problem but they do not take dynamic workload into consideration, leaving the facts that their solutions do not work well when the workload is volatile. In this paper, we present a Dynamic time Slice Scheduler (DaSS) as a software solution which brings a significant improvement on the I/O performance. In order to maintain good flexibility and availability, DaSS requires no modification on the Guest OS but only a few changes on the VMM. We implement the prototype on Xen 4. 3. 0 and conduct detailed evaluations on network benchmarks. Our experimental results show that DaSS increases I/O throughput by nearly 40%.
CITATION STYLE
Ma, R., Li, J., Lin, L., & Guan, H. (2015). DaSS: Dynamic time slice scheduler for virtual machine monitor. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9528, pp. 77–90). Springer Verlag. https://doi.org/10.1007/978-3-319-27119-4_6
Mendeley helps you to discover research relevant for your work.