Research On Grid Based Task Scheduling

The grid is a comparative hot technique developed in recent years,it makes full use of all kinds of available resource which scatter in different geographical position to form a super computer.In the future the internet technique will possibly collect many kinds of idle resources such as computing resources, memorizer resources etc.Grid Computing,a form of distributed computing,has become popular in utilizing and sharing large-scale resources to solve complex scientific problems.Using the Internet for distributed graphics rendering(grid computing problem as example ) has long been a dream of many image and animation producers.Task Scheduling is a critical design issue of distributed computing.A computational grid is a highly distributive environment.To design an efficient task scheduling algorism is one of the most important in grid research.The purpose of task scheduling is to match users needs with full using of available resources in a dynamic grid environment.Finding optimal schedules in grid system has been shown in general to be NP-hard and it is very important to raise some better scheduling algorithm based on existing algorithm which can make the best of all kinds of resources and improve the computing efficiency.Grid computing demands cooperation among the participating resource sites.The concept of "trust" in grids is fundamentally different from that used in p2p.Ray tracing allows for a substantial quality increase by supporting correct shadow calculation, depth-of-field,caustics,reflections,refractions,and so on.The problem of ray tracing is that a lot of computing power is required to create an image,especially for testing large amouts of intersections of light rays with the world geometry.In this paper,the characteristic and performance metric of job scheduling for grid computing is addressed,the existing job scheduling technologies and algorithms are summarized,the problem of load balancing is discussed,after that,the construction of a grid-based rendering system is discussed.A grid scheduling using Ant Colony Optimization(ACO) algorithm and Fuzzy Reputation(FR) Aggregation to improve the decision of scheduling and success job ratio is proposed as well as been experimented by using of GridSim,and a SAH_based kd-Tree is used for the rendering process and experimented by OGRE as well.This way,optimized rendering results which reduce the time spent in the rendering process and,in most cases,do not affect the final quality of images are obtained.The research work of this article has mainly included such several aspects:1.Almost all kinds of grid scheduling algorithm are discussed and compared absolutely.2.A new framework which named ABGRS is constructed by making use of OGRE.3.A new trust model is used in grid by making an assay of trust used in P2P environment.4.A fuzzy-logic based trust model to enable the aggregation of numerous trust parameters is built.5.By combining Fuzzy Reputation Aggregation and Ant Colony Optimization,a new task scheduling algorithm named FRACO is introduced.6.A new cost computing formula for kd-tree construction is designed by putting light to use.7.In addition to grid task scheduling,some coherent topic,such as grid computing,task scheduling principle and architecture and basic scheduling service are involved.8.In addition to rendering,some related topic,such as illumination,ray tracing,Level Of Detail are introduced.Of course,this study is just elementary in supporting fuzzy reputation in ant colony optimization. There are still more works to do,such as taking more refer factors into account,having a look at the cost of completing a task.