Research On Grid Task Scheduling Based On Genetic Algorithm

Grid Computing is a hotspot in the research of Internet nowadays, but also is a developing orientation of parallel and distributed processing technology. In grid computing, the task management, the task scheduling and the resource management are three basic functions that the grid must have. The purpose of task scheduling is to assign different tasks to corresponding grid node rationally, simultaneously considering the computing performance of each grid node and the parameter among nodes such as communication performance in grid environment where contains massive different computers. Task scheduling is also been called task mapping. Owing to the variability of processing velocity, load of host computers and the time of network communication, the task scheduling problem is known to be in general NP-complete.At present, the task scheduling research in grid environment is mainly focused on two aspects: resource management and scheduling system, and the algorithms.In this paper, we do some research on task scheduling algorithm, and make the following contributions:◆Analyzing the task scheduling problem and the TITAN scheduling frame deeply.◆Aiming at the task scheduling problems existed, we propose a new genetic algorithm for task matching and scheduling in grid environment. In this scheduling algorithm, we can meet various needs of user and resource provider by adjusting the value of fitness function parameters.◆Performing a experiment based on Gridsim to emulating the algorithm, and doing comparison with the policy and the Nimrod/G algorithm, the experiment results show that the genetic algorithm proposed in this paper is fit for grid environment, and can receive greater affection.