Research Of Scheduling Algorithm In Distributing System

Grid Computing is a typical Distributing Computing and Grid Computing system is a typical Distributing system. It has been a hot point about Grid Computing various technologies in the Computer filed. Many key technologies were aroused attention of researchers such as task scheduling, task cooperation and resource cooperation, etc. At the same time many task scheduling algorithms based on Grid Computing had been brought.At present, the great mass of existed task scheduling algorithms based on Grid Computing are static heuristic algorithm. The task scheduling based on Grid Computing has been proven to be NP-Complete for most cases (besides rarely especial case). The direct aim about task scheduling is to realize optimization scheduling and try to improve Grid System throughput.Firstly, the similarities and differences were compared between Grid Computing and Peer-to-Peer, applied P2P to Grid Computing and brought a new resource organization management model-P2P_Grid model. Super-Peer divided large-scale Grid System to some small-scale sub-Grid system in the new model, and each Super-Peer was control center of sub-Grid system that it belonged;all Super-Peers in the different sub-Grid system were peer to peer. So, each Super-Peer not only could complete concentrated scheduling for local tasks but also could cooperate with other idle Super-Peer to complete distributing scheduling.Secondly, a new task scheduling algorithm based P2P_Grid model was improved, named P_G_Min algorithm. The new algorithm took advantage of P2P_Grid model's characteristic to schedule the tasks that users submitted in sub-Grid system, at same time Super-Peer always designed local resources to scheduling tasks. As a result, the completion time of tasks scheduling to be reduced and the utilization of system resource to be improved.Finally, three algorithms were tested on GridSim simulator. First,Min_min algorithm was tested by using the simulator, then the Min_min algorithm was modified and QOS-Min_min algorithm was draw out. QOS-Min algorithm was also simulated on GridSim simulator and testing data was registered, comparing outcome with conventional Min_min algorithm. At last, the P_G_Min algorithm was compared with traditional Min_min algorithm and QOS-Min algorithm by simulate experimentation. It shows that the new algorithm has a better quality of system load balancing and the utilization of system resource.