The Research And Application Of Dynamic Feedback Load Balancing System Based On Cluster

With the development of information technology, especially the Web technology,there are higher requirements for network servers. Bottlenecks emerge on the serverside. Cluster technology is an effective way of realizing high performance service.The application of load balancing in the cluster can profoundly improve theavailability, scalability and reliability of servers, expand the bandwidth, increase thethroughput, enhance data processing and raise the flexibility and availability of thenetwork.Through complete study on the Linux-based server cluster, this paper proposed adynamic feedback load balancing algorithm based on the general dynamic feedbackmodel by considering the real-world needs and features of the static algorithm in theexisting open source projects LVS(Linux Virtual Server). Taking into account thereal-time load and response of each server, the algorithm, via periodic access to theirstatus, dynamically calculated their weight values and constantly adjusted the ratio ofrequest processing, thus improving the throughput and reducing the response time.The article firstly studied the dynamic feedback algorithm and its general principles.Then, by studying the load balancing technology related to large-scale cluster systems,it put forward the framework of the system architecture and method of realizing it.The paper built the system by using network programming and dynamic load librarytechnology under Linux. Finally, performance of the system was tested byexperiments. The results demonstrated that the dynamic feedback algorithm caneffectively improve the performance of the load balancing system, proving thecorrectness of the algorithm.This paper thoroughly analyzed the load balancing technology in server clusters.The proposed dynamic feedback algorithm can reduce response time and raise theutilization of system resources. The application of the dynamic feedback loadbalancing system in high performance computing and information service will bringsubstantial social and economic benefits.