The Research And Improvement Of The Load Balancing Technology Based On Nginx

Today we are all living in Internet Age. The rapid development of internet has deeplychanged people’s traditional ways of life. As we know, most jobs can be done through theinternet in modern society right now. Although we are enjoying in convenience which comesfrom the improvement of several kinds of web services, the internet also brings a greatchallenge to many researchers——an explosive increase of web concurrent requests andmore complicated transaction processing on net is severe trouble for real internet servers.In order to meet the challenge mentioned above, it is normal to use a solution that tobuild a cluster system consisting of servers. As a core of cluster system’s research, loadbalancing is mainly used to solve the problem in reasonable distribution of a large number ofrequests. This article states the background, concept and current research situation of loadbalancing. And the article explains some important aspects of technical realization in loadbalancing. And it gives an analysis about some strategy like evaluating, classifying,collecting and managing of load information. Then this article has a detailed descriptionabout types, levels and obtaining method of different load indexes under linux system.In this article, the author puts forward an improved method to evaluate the load ofservers, which is based on summarizing the existing results of research on synthetical load’sevaluation. The method is that to deduce a relational model between response time and loadindexes by multiple regression analysis. The deduced model will be used to predictingsynthetical load and decision-making of load balancing. With the improved method ofsynthetical load’s evaluation, the author implements a dynamic feedback load balancingalgorithm on an Nginx cluster. And the results of dynamic feedback algorithm’s performancetesting prove the effectiveness of the improved method of synthetical load’s evaluation.At the end of the article, the author demonstrates main innovations and some shortagesand future work of this research.