Design And Implementation Of Load Balancing System For Interactive Television Application Based On LVS

At present, with the development of Internet, interactive television application inthe home is becoming increasingly popular. In order to enable users to get a betterexperience, use the Linux virtual server cluster to deploy the interactive televisionapplication, and achieve load balancing system. But the load balancing algorithm hasbeen applied to interactive television application is not perfect, the default algorithmcannot go according to the real-time load of backend server to distribute the systemload, and complete the true load balancing.In this paper, for interactive television application load balancing system that wasbuilt by Linux virtual server cluster technology, proposed an adaptive load balancingalgorithm to optimize the weight distribution, the algorithm included a method foroptimizing the weight distribution and a dynamic adaptive approach. Firstly, analyzedthe default IP-based virtual servers weighted least connections algorithm of system,according to the load tilt that appeared in visiting interactive television applicationsduring the peak period, proposed a method to optimize weight distribution and adynamic adaptive approach to optimization algorithm. The algorithm obtained the truebackground server load and the ratio of request response time and the maximumresponse time of the request that user can tolerate, increased the parameter that couldtruly represent the backend server load values in the default algorithm. Then achievedautomatically modifying the weights of servers in a cluster system, and updated therevised new weights of server to the scheduling algorithm, realized the optimizationof algorithm. The optimized algorithm was applied to the interactive televisionapplication system, in the experimental environment, used a load balancer and threereal back server to build the entire load balancing system, and used the stress testing software Web Application Stress Tool to access the pressure test of system, got andcompared tested data of cluster system before and after using optimization algorithms.The experimental results show that when the adaptive load balancing algorithm tooptimize the weight distribution that have more than400concurrent connections, theentire system of load balancing can get better effect of load balancing.