Real-Time Fault-Tolerant Scheduling Based on Primary-Backup Approach in Virtualized Clouds
Ji Wang, Xiaomin Zhu, Weidong Bao · 2013
Cloud computing represents a revolutionary paradigm for the great provisioning of computing resources. However, the enormous scale of Clouds increases the probability of failures. So fault-tolerance becomes a significant requirement, especially for real-time applications due to their safety-critical nature. Fault-tolerant scheduling as an efficient fault-tolerant technique attracts extensive studies. Unfortunately, existing fault-tolerant scheduling algorithms, based on the primary-backup approach, do not take virtualization, one of the key features of Clouds into account. To address this issue, we propose a fault-tolerant scheduling algorithm for virtualized Clouds named FSVC using primary-backup approach to tolerate physical host failures. FSVC strives to enhance the performance by employing comprehensive overlapping techniques and Virtual Machine (VM) migration technique. The constraints of the two techniques are elaborately analyzed to realize fault-tolerance. Besides, a two-phase policy is incorporated inFSVC to further improve the performance. Extensive simulation experiments demonstrate that FSVC can improve the schedulability and resource utilization effectively in virtualized Clouds.