Real-Time and Preemptive Workflow Scheduling in Cloud Systems

With the advancement of virtualization technologies, cloud computing environments have become more dynamic and real-time, enabling efficient resource allocation and task execution. However, real-time workflow scheduling in such environments remains a significant challenge due to the uncertainties in task execution time and data transfer delays. In the paper, the real-time and preemptive workflow scheduling problem is investigated. User requests are modelled to various workflows and arrive at the system stochastically. The time information about the workflows, such as the execution time and the data transfer delay are uncertain. Only the expected base values are given, and the real time information are certain only when these tasks are completed. The tasks in these workflows are allowed to be interrupted for at most one time. To address the problem, a real-time algorithm framework is proposed. The framework is integrated with a simulated annealing algorithm to obtain a high-quality feasible solution. A task interruption and resource re-allocation strategy is designed to interrupt tasks at the appropriate time and migrate interrupted segments of tasks to appropriate resources. Experimental results show that our approach achieves the best RPD of 0.424%, outperforming RMWS (12.866%) and GLS (14.862%).