Task Scheduling Algorithm in Grid Environment Based on Duplication and Insertion

Grid resource scheduling theory involving basic theoretical knowledge for grid scheduling was discussed in this paper. Then an intensive study of the scheduling strategy was made. According to the heterogeneous characteristic of grid environment, an improved algorithm DIBS for task scheduling were proposed. In this algorithm, the entire scheduling process was divided into three steps: layering, task priority, and task replication. In the layering stage, according to the characteristics of the DAG, the simultaneous distribution strategy for the multiple DAG images was adopted. In the task priority stage, an improved decision path strategy was proposed. In the replication stage, the previous key path nodes were replaced by the best precursor replication nodes. The effectiveness of this algorithm was verified by Gantt chart. In this paper, the relevant scheduling algorithm simulation was successfully realized by using the basic framework and functions provided by SimGrid and combining with the proposed scheduling algorithm. The availability, validity and stability of the DIBS scheduling algorithm were verified by comparison and analysis of simulation results.