Study On Task Scheduling Schemes Based On Time Balancing In Computational Grid

Grid computing technology has emerged as an important new field, distinguished from conventional distributed computing by its focus on large-scale resource sharing, innovative applications, and, in some cases, high-performance orientation. Computational grid is one of the most deeply and widely research fields in grid. Scheduling of tasks search some optimum resource set for some task and schedule it on those resources to execute. Scheduling remains one of the most complex problems in computational grid, which is worth deeply research in theory and practice.This paper proposes a task-scheduling algorithm based on time balancing policy for task farming problem to improve system throughput, in order to serve better in task scheduling and information integration in the future.Firstly, this paper analyses the characters of applications in computational grid and the essence of task farming problem, and compares the existing scheduling algorithms. Secondly, it constructs the model of task execution time considering interruption during the process of task execution. Thirdly, it analyses prediction technology in scheduling, and proposes prediction model to predict task expected completion time using subtasks with maximum mathematical expectation. In order to gain the minimum time of task expected completion, this paper suggests a time-balancing policy of task partitioning. According to this policy, a task is decomposed to several parallel subtasks, which are scheduled to several corresponding resources to execute independently, and the results are gathered to integrate. Based on batch mode and remapping principle, a time balancing based scheduling (TBBS for short) algorithm is proposed to minimize the completion time of tasks for high-throughput computing.This paper simulates the performance of the TBBS algorithm, and compares it with Min-Min algorithm and Sufferage algorithm.Finally, a scheduling service with TBBS algorithm is designed as a part of resource management and scheduling system based on service.