Research Of Load Balancing Algorithm On Streaming Server Cluster

Cluster is a new kind of computer system that undertakes the parallel processing appearing in recent years. It utilizes network to join a group of high-performance PCs or workstations, operate deployment under the parallel programming environment and realize the high-efficient computing. Cluster system is very apt to realize the expansion of systematic function. The technology and its application of streaming media that used in cluster server is the one of relevant application.Streaming media is new developing technology in recent years. It is the outcome results from computer technology, network communication, multimedia technology and video compress etc. The application of streaming media employs the amount of great data, and requires high bandwidth with the characteristic such as being strong of real-time, so the streaming media research on cluster has attract hot extensive concern more and more.This paper directs against the characteristic of the streaming media on the basis of studying several kinds of different load balancing algorithms, through studying load balancing theory, combine the advantage and disadvantage of the genetic algorithm and existing achievement of theory and practice about the cluster, have proposed two kinds of new load balancing algorithms on the cluster environment and analyzed and compared between it. The algorithm includes two parts: One part is to study load balancing algorithm based on genetic algorithm, one is to study the adaptive load balancing algorithm based on dynamic feedback mechanism, and at last paper give some analysis between two kinds of algorithms.The research of load balancing based on genetic algorithm is studied and preceded from angle of science of genetics and apply the genetic algorithm to the balanced load of cluster. Because genetic algorithm explains principle and method from the angle of evolvement that the mankind realizes the objective world, it offer the optimization frames for datum and information processing and have the overall situation parallel searching ability, so it has been applied extensively in some relevant fields. This paper tries to adopt the advantage of the genetic algorithm to further explore new method in order to improve the performance of streaming media that used in cluster.