Research And Implement Of A Content-based Load Balancing Technology

With the growing popularity of Web2.0, B/S architecture has replaced the traditional C/Sarchitecture, become the first choice of architecture. With the number of concurrent Webrequests is increasing, exceeding the original enterprise server’s maximum capacity,companies may adopt server hardware upgrades to cope with the increasing amount ofconcurrency, but this approach would eliminate the existing server hardware, resulting inwaste of resources. Cluster technology can effectively solve the existing server hardwarereuse, while responding to the growing amount of concurrency.Large web sites generally take load balancing cluster technology. Load balancing cluster isdefined by one or more front-end load balancer will distribute the workload on a group ofservers to the backend in order to achieve high availability and scalability.Firstly, the load balancing cluster technology including load balancing and load balancingscheduling policies related technologies are introduced, and the currently existing loadbalancing scheduling strategies are the pros and cons analysis. Through the Web serverperformance evaluation studies,we determine the response time as a performance evaluationindex, taking into account differences in the content of the request, we propose acontent-aware load balancing based scheduling policy(RCMRT).Through the request contentclassification, and load balancer statistics the response time of each request, and then selectthe shortest average response time of the server, the request is distributed.Secondly,we study the open source load balancer Nginx, implement the Nginx-based loadbalancing scheduling policy(RCMRT), and Nginx-Tomcat server clusters based on TPC-Wspecification performance tests show that the algorithm is better than the weighted RoundRobin algorithm and the weighted least connection algorithm.