Research On Grid Resource Management

Grid technology and application would be the next tidal wave in the Internet information technology. The availability of the resource management system is straight related to the Grid feature and performance. It is also one of the critical factors for Grid coming into practicality phase. In this paper, I did some instructive research on the Grid resource management and obtained the following achievement: 1) Some defect was found in the experiments of the DBC schedule algorithms of the GRACE model that most of the tasks ran only on a minority of machines when the user's tasks number was not far larger than the machines number. The cause is that the GRACE model can't be adaptive to the dynamic resource. 2) The paper discusses about the design ideas for the resource management model for computational economy, and introduces the commodity currency theory to the resource management research. It is a breakthrough for the solution to the dynamic resource management that a self-contained Grid resource market is set up and the resource exchange is performed through the media of commerce. The trading organization is the basis of the Grid resource market. The effective resource allocation should come from the well-constructed trading organizations and the appropriate trading rule. 3) The paper puts forward a virtual trading organization based model for the Grid resource management, which named as VTOM. A trading organization is mainly composed of a resource exchanger and more than one user, user broker, resource provider, resource broker. The resource brokers and the user brokers organize the supply and the demand separately, which deal with self-determination in the resource exchanger. Many trading organizations build up the Grid resource market. There exists some relation between the trading organizations. The participators join the trading organization dynamically. The Grid regulatory committees are market administers who are responsible for totalizing the market information, monitoring the resource trading and regulating the resource price and the resource supply and demand. The VTOM model, mapping by the market framework of the economy to the resource management, concentrates the merits from the distributed schedule and the centralized schedule in some extent. Therefore, it is better than the GRACE model. 4) The VtomSim was developed based on GridSim to implement the basic feature of the VTOM model. The paper puts forward a QoS-based hierarchical management method for computational resources named as QHMM. It can be adaptive to the dynamic resources, which was proved in the experiments. Some contrast experiments was done between the VTOM model and the GRACE model, and the result makes clear that the prior is better than the latter.