The Study Of Improved Ant Colony Algorithm For Task Scheduling In A Grid

Job schedule is a key research subject in computing grid. In grid environments, from task submission to result processing, all events about tasks are under the control of task management. How to schedule job to meet the users'needs is a quite challenging question because grid pronounces the characteristics of large scale, isomerization, dynamics, distribution and autonomy, ect.But Ant Algorithm is a efficient algorithm to solve this kind of problem. This paper offers a load balancing algorithm for quick Ant Algorithm on basis of breadth-first-search and that is based on the results of researching for the grid job schedule algorithm with the characteristics of grid. First, analysis the grid characteristics, grid system structure, grid resources schedule period, offering ant algorithm through approach to solve new intelligent calculation problems and simulating natural ecological mechanism, it is quite suitable for grid job schedule with its parallelism and extensibility. But this method is not good enough at inefficient and load imbalance, and we have improved them already. So that to balance the fast-balance of the two machines, and then through repeated iterations to fast balance the task on each machine in grid system. Finally, the use of grid scheduling simulation toolkit GridSim, carried out a large number of simulation experiments. The results showed that: Fast-Balance algorithm has good performance, and can receive better Min-Min, Max-Min algorithm scheduling results.