Web-based Applications Lvs Dynamic Load Balancing Scheduling Policy Research

With the development of computer and communication technology, Internet has presented the explosive growth. Especially along with the appearance of the e-ecommerce, online demand and other fashion web applications, people always encounter the server downtime because of the large load when they enjoy quickly, rich and comfortable network service. And this seriously affects the quality of service.In order to solve the server performance bottlenecks, we have proposed two different upgrading methods to solve the problem. One is the single-server strategy, which improves the server performance by continuous hardware promotion. The other is the cluster strategy, which uses a group of computer to compose a big cluster to provide the quality of service through certain software or hardware, which used to provide by high-performance servers. Compare the two thoughts of solution; the cluster strategy can use limited funds to provide better quality of service.This article's main research is about the popular Linux Virtual Server used in high-load Web server cluster, in view of the shortage of the existing load balancing algorithm in LVS, this paper proposed a new improved load balancing strategy based on the existing WLC strategy. In this improved strategy, we propose to use the dynamic feedback thought which can dynamic fixed the servers'weight by special load collection software. Through apply this in weighting algorithms to realize the improving performance of cluster system.In this dissertation, we discuss the character of the Web server cluster system, besides we also analysis the implementation of Web servers'load balance cluster. We also studied the three IP packets forwarding ways in Linux Virtual Server and the IPVS software architecture. Meanwhile, we also analyze the shortage of the LVS load balancing algorithm used in high-load Web server cluster, and proposed an improved scheme. At the end of this article, we construct a small cluster in the laboratory to compare the improved algorithm and the original algorithm, the test data shows that the improved algorithm can enormously enhance the performance of the Web server cluster.