Research And Implementation Of A Grid Simulation Platform Using GridSim

Grid computing is an important branch of distributed computing. As the grid itself is highly complex, it is difficult to verify the grid computing theory. Most of the grid resources are very valuable, Grid researchers must pay a long time with higher cost to do their theoretically verified tests. So, it is not conducive to the conduct of research. The experimental verification of grid computing theory research has become one of the major constraints in grid computing.GSP (GridSim-based Simulation Platform) refers some of the existing GridSim-based software's features and integrates them into a comprehensive grid experiment simulation platform. In order to improve the accuracy of experimental data, this paper researches and develops a massive grid data modeling tool using Hadoop technology. Using this modeling tool, researchers can dig out the core data which really needed in experimental from the huge amounts of data, and established the random events model these data meet.Concerned about the widespread load-generation's lack of GridSim-based simulator for grid workflow simulation, this article researches the common type of grid workflow, and then designs and implements a GridSim-based workflow load generator. The generator can generates three types of workflow load for users to choose, and provides researchers with the extension interface to generate other types of workflow.In order to simplify the experimental operation, GSP provides a graphical user interface and integrated part of the Matlab mapping function. It provides a convenient tool for user to analysis the experiments.Application shows that GSP makes up some deficiencies of the existing Grid simulation tools, and provides researchers a comprehensive, easy-to-use, reliable grid simulation platform.