Design And Realization Of Scalable Content-aware Web Cluster With Transparency

With the development of Internet, the access to Web is rising rapidly. The loads of Web servers become more and more heavy. Web Cluster, which can be used to solve the overload of Web servers efficiently with low cost, becomes an important technology for constructing big Web sites. Web Cluster system becomes a hot research topic in recent years, and now the main research of Web Cluster system focuses on the layer7 (content-aware) Web Cluster system. But there are many disadvantages with the present resolutions, such as the lack of scalability, transparency and integration, new Web Cluster system needs to be researched deeply.Based on this idea and the analysis of the present solutions, a new scalable content-aware Web Cluster with transparency is proposed in this paper. There are some promising virtues with the new Web Cluster system, such as high scalability, transparency to back node and the ease of integration. In this paper, the architecture and mechanism of this Web Cluster system are introduced in detail. To give an optimized solution of the new Web Cluster system, this paper analyzes the data flow process of this system. Then the essential of this process is disclosed: the dynamic switching between M TCP connections and N TCP connections. As a result, the M: N TCP switching protocol based on request packet is proposed. Then an implementation under Linux system is given. The code that I wrote works as a special module in Linux network kernel. The design and implementation of TCP switching protocol and the new Web Cluster system are illustrated in detail by flow charts and explanation. At last the experiments on this Web Cluster system are done. The result shows that there is a good performance about this Web Cluster system with high scalability and good service. The paper summarizes our research and puts forward the future work in the end.