Research On Load Balancing Algorithm For Web Cluster System

As the popularization and development of the internet,some popular websites increase rapidly.As a result,Supplying high quality and high efficiency service has come to be the top important factor for every Network Operator,especially for E-Commerce Website.Increasing website traffic means increasing load,which willdirectlyextend the site'sresponsetime.As the increasinglyfierce competition between sites,customers haveincreasinglyhighrequirementsonsiteresponsetime,and their patience to wait for a response is day by day running low.Indeed,as the increase in response time,customer churn increased.However,an ordinary server can handle tens of thousands of to hundreds of thousands of requests per second rather than millions of that or even more.But if all these servers were integrated into a system,by reasonable allocation policy all requests equally among all nodes in the WEB cluster system,which would completely have the ability to even more millions of requests per second.Based on the basic purpose of evaluating server performance,the CPU utilization,memory utilization,bandwidth utilization,and disk IO utilization are deeply studied.The current running state of the comprehensive evaluation server is a microcosm of the current cluster's job situation.In this paper,a comprehensive analysis of relevant technology and load balancing strategy,combined with the load balancing algorithm,topology design,request forwarding,node maintenance,connection pool technology,proposed a load balancing algorithm based on fuzzy clustering,and implements a load balancing system.