The Research And Implementation Of Resource Management Related Technologies Based On Science And Engineering Computing Application Gird

With the development of computer science and technology, theory, experiment and computing has become the backbone of modern science research. High-performance computing has become the basal intersecting subject to support science study and new technology development, more and more key problems from science research and major projects cannot simply relay on theory and experiment, and large-scale and high-performance computing and simulation provide a new research way of solving these key problems. Due to the heavier and more rigid computing demand, how to organize the heterogeneous resources distributed in different locations or belonging to different organizations, through the high-speed internet to provide users with a large-scale, extensible computing ability, has become our research point.Gird, as a basal computing tool, achieves computing problems by connecting kinds of distributed and heterogeneous computing resources through high-speed internet. It provides ways of solving huge-scale and complex problems, which often have mass computing or data. The aim of computing grid is to realize a uniform way to describe distributed, heterogeneous resources, encapsulate computing programs into service and provide an access mechanism with high efficiency to access grid service, thereby the problems of complex grid resource usage and management will be solved.There are various resources in computing grid system, which have characteristic of dynamic change, wide area distribution and system heterogeneity. The aim of computing grid resource management and scheduling is to solve problems like resource description, organization and management, which is the core of computing gird. Resource management architecture and scheduling has great influence on performance effect of computing gird applications, the effective resolution of such problem is a key science technology for computing gird turning to practicality, it has broad foreground in high-performance computing, distributed computing, parallel computing and high-performance internet.This paper uses large-scale science and engineering computing as background, studies key technologies like gird middleware Globus, gird resource management and scheduling module, grid monitoring, gird accounting; It builds a gird resource allocation module with multi-factors, multi-phases and multi-purposes, through the actual application to regulate the inspiration and restriction mechanism, thereby more resources will be shared effectively.This paper gets supports from "Science and Engineering Computing Application Gird" project, Shandong Province 2004 Science and Technology Plan.