Research Of Dependent Task Priority-Based Grid Scheduling Algorithm

Grid has been applied to solve the problem of large-scale and complex system, and grid scheduling is to solve the problem of scheduling task to the appropriate resources node, which is not only a key issue of grid system, but also a NP complete problem. Scheduling algorithm is to ensure high performance of the grid system by minimizing scheduling length. Using the static priority grid scheduling is a common form, which is also a method of high performance.This grid system solves he problem of scheduling non-independent task of grid system, the grid task will be represented by the DAG map, which can simplify the description of the grid task. Through referencing the existing grid scheduling algorithm and analysizing the strengths and weaknesses of static scheduling algorithm of grid, a grid based on priority scheduling algorithm (DTPBS algorithm) is proposed. This algorithm calculates the neutron value of the priority of each application and divides non-independent tasks by priority, which schedules non-independent tasks according to the order of priority. Scheduling criteria is to minimize the grid scheduling length, which is makespan. This paper proposes service migration approach to reduce the data transfer time between tasks, thereby reducing the length of the grid scheduling system.This paper compares the proposed grid based on task priority scheduling algorithm (DTPBS algorithm) with the existing Max-min algorithm and the Min-min algorithm in two aspects (resource utilization and the scheduling length) by carrying out simulation experiments,. Simulation results show that this algorithm has a significant advantage in both of them. Finally, the paper also gives a detailed description of application of DTPBS algorithm in the grid, but the algorithmis not fully realized, and then it will be applied to a real grid environment.