JISE


  [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]


Journal of Information Science and Engineering, Vol. 29 No. 5, pp. 1021-1035


Performance Tuning Towards a KVM-based Embedded Real-Time Virtualization System


RUHUI MA1, FANFU ZHOU1, ERZHOU ZHU2,+ AND HAIBING GUAN1
1Department of Computer Science and Engineering
Shanghai Key Laboratory of Scalable Computing and Systems
Shanghai Jiao Tong University
Shanghai, 200240 P.R. China
2School of Computer Science and Technology
Anhui University
Hefei, 230601 P.R. China

 


    Virtualization is a fundamental component in cloud computing because it provides numerous guest VM transparent services, such as live migration, high availability, rapid checkpoint, etc. Utilizing virtualization technology to combine real-time operating system (RTOS) and off-the-shelf time-sharing general purpose operating system (GPOS) is attracting much more interest recently. Such combination has the potential to provide a large application base, and to guarantee timely deterministic response to real-time applications, yet there remain some issues, such as responsiveness of RTOS running on top of a virtual machine (VM), system performance and CPU resource utilization rate, etc. In this paper we propose an embedded real-time virtualization architecture based on Kernel- Based Virtual Machine (KVM), in which VxWorks and Linux are combined together. We then analyze and evaluate how KVM influences the interrupt-response times of VxWorks as a guest operating system. By applying several real-time performance tuning methods on the host Linux, we will show that sub-millisecond interrupt response latency can be achieved on the guest VxWorks. Furthermore, we also find out that prioritization tuning results in waste of CPU resources when RTOS is not executing real-time tasks, so we design a dynamic scheduling mechanism - co-scheduling to improve system performance. Experimental results with SPEC2000 and bonnie 1.4 load, show that this new architecture tuned by CPU shielding, prioritization and co-scheduling, can achieve better real-time responsiveness and system performance.


Keywords: KVM, muti-core, real-time, co-scheduling, virtualization

  Retrieve PDF document (JISE_201305_13.pdf)