Improving And Realization Of Min-Min Algorithm In Grid

The rapid developing network makes it possible to integrate the geographically distributed and heterogeneous computing resources into tremendous Grid. So the tasks scheduling in Grid has become an important problem. And the research of schedule algorithms, which directly related to the speed and quality, plays a decisive role in the research of computing Grid.Min-Min algorithm, a classics algorithm in heuristic algorithm, which always completes the shortest total completion time task first, has the characteristic of simpleand shortest completion time. So it catches a lot of close attentions in the field ofstudying for tasks scheduling algorithms in Grid.But most of the existing Grid Systems are developed for the purpose of learning research, these systems emphasize the share and cooperate work with resources, the price of resource is not considered. Most of the resources are not volunteer in the actual application. In result of the problem mentioned just now, the Min-Min algorithm already can not satisfy the need for scheduling. So we should consider new algorithms that take the tasks' demands for price of resources in account.Based on this, this paper focuses on the following researches:After we know the research status for task scheduling objects, algorithms, and simulatorsof task scheduling algorithms,then modify the Min-Min algorithm. Firstly, we Expands the concept of shortest finish time in Min-Min algorithms, and makes the job finish at the least cost, which not only include completion time but also the expenses of CPU. Secondly, we define performance price ratio function ,then we can change "distribute every jobs to the computers which finish them in the shortest time" in Min-Min algorithm to "distribute every jobs to the computers which finish them in the best performance price ratio".We simulate Min-Min algorithm,Qos Guided Min-Min algorithm and the two modified algorithms in program. Analyzes and compares the above four algorithms from mufti-aspects through several sets of contrasting experiments. At last, it is proved that the modified Min-Min algorithm has shorter total cost than the other two algorithms, at the same time, the Grid resources are more balanced. So we can conclude they fit for the Grid environment better than the other two algorithms.