Design And Implementation Of Mission Drift Grid Scheduling Model With A Consultative Mechanism

Computing Grid is a collaborative mechanism involving multi-resources. As an important aspect of collaboration, workflow has become a new research field of computing grid in recent years. From view of application of computing grid, it is natural to take commodity market as it's model because computing grid can be seen can commerce oriented computing environment . Due to characteristic of distribution, heterogeneousness and respectively pricing strategies by their owners, the grid resource price is hard to manage. In addition, the free competition and unreasonably price in grid resource market will cause some resources fail to schedule. Meanwhile, some of participants may not abide by rational rules, such as some resources buyer makes other consumer fail to visit some resource and make its long unused, forcing the owner have to reduce it's price . And resource providers may maximize their own interests to price unreasonably even monopoly. For task of workflow, despite many task scheduling model is put forward, the bottleneck of randomness of the grid resource pricing makes traditional scheduling model lack of adaptability.It uses market mechanism to simulate the grid computing environment, and uses price control mechanism to realize dynamic balance control of the supply and demand. The convenience of resources and task flow management can be realized through the descriptor, solving the task of collaboration issue in the flow of heterogeneous environment for task flow. Using MCT matrix to store resources and task descriptor in order to map operation and resources conveniently, Thus operations and resource matching rate is improved. A second task match is performed using price negotiation mechanism in the case of mismatch of the price of resources and tasks, so that failure scheduling due to mismatch is greatly reduced. Combining the resource scheduling management with the price negotiation, this model is improved based on traditional grid resource management. It lays a foundation for realizing efficient and intelligent resource management. Simulation experiment shows that this model is feasible. This model can be applied to the grid computing environment of task flow, reducing outside factors influence on the grid resources and improving the management efficiency, reliability, throughput and success rates.