Grid technology combines all kinds of computational resource, data resource, storage resource scattered in the Internet, and it enable users all over the world take Internet as a virtual super computer. However there is a lack of mechanism that make user use these resources more convince and effectively. This promotes the development of Grid scheduling platform whose responsibility is to provide systemic support for heterogeneous grid resource management. Recent Grid Scheduling System like Nimrod/G, Condor/G, AppLes cover some aspect of Grid scheduler, however they do not propose a framework and protocols to support dynamic, scalable and large-scale Grid environment, the centralized resource management mechanism they adopt also introduces single-failure. Meanwhile, P2P technology promotes quickly owing to its excellent characteristic and performance such as extensibility and fault-tolerance. In this paper, we propose P2P-based scheduling framework by introducing P2P topology. Systemic services and protocols will be discussed. We also implemented prototype, case study is performed to evaluate our model.
|