Task scheduling is a hot issue in computing fields, aiming at improving time efficiency of algorithms or resource utilization of computing platforms. This article mainly deals with the data processing scheduling problems in scientific workflow systems. In this paper, data processes in the scientific workflow will be analyzed based on the process activity graph. According to the dependencies between the processes, some processes will be merged before scheduling. During the execution of the scientific workflow tasks, the processes that can be performed in parallel will be recognized automatically by the proposed method, which will be further executed by the developed scheduling approach via virtual machines with the goal of minimizing the workflow run time. Experiments are conducted to evaluate the proposed methods. The results show the effectiveness and positiveness of the scheduling strategy.
Paper
Full text
Towards task parallelizing in scientific workflow systems
Semantic Scholar · Computer Science · 2022
Abstract
Task scheduling is a hot issue in computing fields, aiming at improving time efficiency of algorithms or resource utilization of computing platforms. This article mainly deals with the data processing scheduling problems in scientific workflow systems. In this paper, data processes in the scientific workflow will be analyzed based on the process activity graph. According to the dependencies between the processes, some processes will be merged before scheduling. During the execution of the scientific workflow tasks, the processes that can be performed in parallel will be recognized automatically by the proposed method, which will be further executed by the developed scheduling approach via virtual machines with the goal of minimizing the workflow run time. Experiments are conducted to evaluate the proposed methods. The results show the effectiveness and positiveness of the scheduling strategy.