Analysis And Research In Grid Task Scheduling Algorithm

With the rapid development of computer network technology, grid technology has become the focus and front issues of international computer technology research. It is considered to be the third internet revolution, following by Internet technology and Web technology, and has attracted great attention from the broad research community. Grid can enable geographically dispersed resources to transparently gather together, more suitable for large-scale distributed applications. Grid technology brings many benefits for users:to provide convenient access interface to various resources, to facilitate resource share of heterogeneous environment, and to eliminate isolated islands of information.In grid computing, task scheduling technology is one of the core grid services. Since the grid is dynamic, heterogeneous and diverse, the related method on how to schedule a large number of users to the optimal resource nodes for execution has become the focus of international and domestic academic research.The main work of this paper and innovative results are as follows:(1)This paper studies three grid architecture in-depth, analyzes their differences, and introduces several typical grid resource management models and systems.(2)The paper discusses in detail several classical task scheduling algorithms:Min-Min, Max-Min and Sufferage, and analyzes the advantages and disadvantages of these algorithms respectively from both sides of scheduling span and waiting time.(3)As the actual grid environment is dynamic and heterogeneous, it will bring tremendous difficulties to the research of task scheduling algorithm. Grid simulation tools are usually used to complete this work. This paper focuses on the GridSim grid simulator, and presents five examples of the GridSim gradual development. This will provide many helps for the future work.(4)As an experimental study, this paper carries out multiple sets of comparative experiments from both sides of economic and non-economic factors with GridSim tools, also studies the resource allocation strategy's impact on system performance. Our results provide important evidences to the future improvement of task scheduling algorithms.