Grid Task Scheduling Based On Economic Models

Currently, grid computing is a hot research of the network and has good development potential. The task scheduling of grid is a core issue in grid computing. As the complexity of the grid system itself, the problem of grid task scheduling is very complicated so. The introduction of economic models will help to solve this problem, and allowing the allocation of resources in a more reasonable way.Firstly, this paper introduces the research of grid task scheduling. Most of the current grid projects were found by research institutes in order to large-scale scientific computing,and the scheduling strategy, first consider the overall performance of the system(such as system throughput, the time efficiency of task execution and so on). With the commercialization of grid applications, Economic model was introduced into the grid. In the second chapter of the paper, it describes a grid system based on economic models by the case of GRACE. Different from the traditional grid, grid scheduling strategy in the Economic model needs to consider the economic interests of the participants.Then I conducted a detailed study of scheduling problems, and analysis of the typical algorithms (such as Min-min, Max-min algorithm).In the fourth chapter, it describes the Nimrod-G. It is an economic grid and based on the architecture of GRACE. Then it analyzed the shortcomings of DBC scheduling, which is unbalanced load. This paper from the grid user perspective, proposed a new scheduling strategy which is based on price and time trade-off. This strategy avoids the issue of load imbalance. Then I proposed two new algorithms which are Min-minTCB algorithm and Max-minTCB. In this strategy, users can set up a trade-off factor, to represent the task's sensitivity on time and price. It can meet the QoS requirements of different users. At the same time, the algorithms are also very good inherited the advantages of Min-min and Max-min algorithm.In order to verify the feasibility of the algorithms, a task scheduler was developed which is based on Gridsim. Gridsim is a well-known tool for grid simulation. Finally, Experiments show that the algorithm is feasible. and compared with the DBC strategy has greatly improved.