Grid task scheduling based on economic models
|Course||Applied Computer Technology|
|Keywords||Grid computing Economic model Task scheduling Economic grid Time cost trade - off|
Grid computing is a hot research network, has a good potential for development. Grid task scheduling is one of the core issues in the grid computing. Since the grid system complexity and heterogeneous, dynamic, autonomous characteristics of grid resources, grid task scheduling problem is very complex. The economic model contribute to the resolution of this issue, so that the whole rationale for the allocation of resources in the grid more. This paper first introduces the current grid task scheduling. Most grid projects is the construction of a number of research institutions in order to achieve large-scale scientific computing, first consider the overall performance of the system (such as the throughput of the system, task execution time, efficiency, etc.) in their scheduling policy. With the application of grid technology in commercial projects, the economic model is introduced into the grid. Articles in the second chapter to GRACE grid system based on the economic model. Unlike traditional grid scheduling strategy in economic grid need to consider the economic interests of the participants of the grid. The next grid task scheduling problems studied in detail, and the typical algorithm (such as Min-min, Max-min algorithm) analysis. In the fourth chapter introduces an implementation of GRACE architecture computing the economic grid model (Nimrod-G), and analysis of the the existing DBC scheduling policy load imbalance problem. From the angle of the grid users, the time prices weighed against scheduling policy for the problems of the DBC strategy, and combined with Min-min, Max-min algorithm the Min-minTCB algorithm and the Max-minTCB algorithm. QoS algorithm to the task execution time and price as standard, users can set up a trade-off factor, said both the consideration of the time and price sensitivity. This strategy to meet the QoS requirements of different users, because the algorithm allows the user time to change the price or price for the time, and the exchange ratio set by the user. At the same time, the algorithm also inherited a very good advantage of Min-min and Max-min algorithm makes better load balancing between resources, to ensure the efficiency of the entire grid system. In order to verify the feasibility and advantages of the algorithm, in-depth analysis in Chapter grid simulation tool Gridsim, and developed based on the above algorithm task scheduler. Finally, the algorithm simulation experiments, experiments show that the algorithm is feasible, and has been greatly improved compared with DBC strategy.