Multi-Workflow Scheduling Based on Genetic Algorithm

Most of the existing multi-workflow scheduling algorithms combine multi-workflow into one workflow, and then use a single workflow scheduling method to schedule the synthetic workflow. This approach of merging multi-workflow into one workflow requires the first workflow to wait for the workflow behind, which reduces the response time of workflow and user experience. To address this gap, this paper considers the arrival time of different workflows and two conflicting objectives, to proposes a multi-objective genetic algorithm for multi-workflow scheduling. This algorithm divides each of the arrival workflows into tasks, then uses the genetic algorithm to find the optimal mapping tasks to virtual machines based on task execution overhead and the virtual machine idle time slots. The algorithm makes full use of the virtual machine's resources and finds optimal trade-off solutions between the cost of tasks and the resource utilization. In the real-world scientific workflows experiment, the algorithm is superior to the exiting scheduling algorithm.