Multiple DAGs Dynamic Workflow Scheduling Based on the Primary Backup Algorithm in Cloud Computing System
Yu Wang, Jia Chang-yun, Yang Xu · 2014
For reliable scheduling problem about multiple DAGs scientific workflow in a cloud computing environment, we propose a competitive, dynamic and multiple DAG scheduling algorithm which takes link communication processor into consideration (CCRH). Firstly, algorithm uses communication competition model to describe the communication between the processors. It calculates the earliest completion time of the primary backup task. It defines the scheduling processor unit and uses dynamic hierarchical approach for multiple DAGs workflow tasks. It is In order to calculate the unfair degree of each DAG factor. When multiple tasks in DAG have large difference weights, the algorithm is an effective solution for previous DAGs would increase the execution time span which caused by the remaining tasks scheduling. Simulation results show that under the premise of ensuring reliable scheduling, the algorithm not only can improve the fairness of the multiple DAGs scheduling, but also effectively short the average multiple DAGs scheduling time, and make the robustness of algorithm is more better.