Dynamic fault tolerant scheduling with response time minimization for multiple failures in cloud

With the increasing demand for large amount of computing resources, the cloud is widely used for executing large number of independent tasks. In order to successfully execute more tasks and maximize the revenues, the cloud service providers (CSPs) should provide reliable services, while maximizing the resource utilization. Providing better Quality of Service (QoS), while maximizing the resource utilization in the event of failures is a critical research issue which needs to be addressed. In this paper, an Elastic pull-based Dynamic Fault Tolerant (E-DFT) scheduling mechanism is designed for minimizing the response time while executing the backups during multiple failures of independent tasks. A basic core primary backup model is also used and integrated with the backup tasks overlapping (BTO) and backup tasks fusion (BTF) techniques to tolerate multiple simultaneous failures. Simulation results show that the proposed E-DFT scheduling can achieve better performance in terms of guarantee ratio and resource utilization over other existing scheduling algorithms.