Abstract :
A grid resourcepsilas domain split management method (GR-DSMM) is presented, which based on association rule model to group the sub-tasks. Firstly, in order to enhance the association between the subtasks in one sub-task-group and strengthen the independency between the groups, the method groups all sub-tasks, which have been syncopated in userspsila tasks according to the association between the subtasks. Secondly, the GR-DSMM takes the sub-task group as one execution unit, which can reduce the frequent communication between the sub-task-groups, and improve the systempsilas efficiency. At the same time, the method divides the grid resources into domain split according to the sub-task-grouppsilas resource demands. And then, the method manages the classified grid resources taking a domain split as one basic unit. Finally, the performance of GR-DSMM is compared with that of multiple-queue backfilling scheduling algorithm and FCFS scheduling policy by emulator, and it proves that the method has superiority and practicability.
Keywords :
data mining; grid computing; FCFS scheduling policy; GR-DSMM; association rule; grid resource domain split management method; grid resource domain split technique; multiple-queue backfilling scheduling algorithm; Association rules; Computer science; Conference management; Distributed computing; Educational institutions; Grid computing; Processor scheduling; Resource management; Scheduling algorithm; Technology management; Grid; association rules; resource management; task scheduling;