The Research And Improvement Of Load Balancing For Web Cluster Based On Nginx

The Internet has developed rapidly in recent years,and the increasingly sophisticated network services enable people to deal with affairs through the network more and more frequently.The development of the Internet has become an important driving force of social development.However,with the sharp increase of Internet users and Internet views,the working ability of the server has been greatly tested.It is especially prominent in some complex Web sites.How to improve the working ability of Web clusters in high concurrency scenarios is the main problem needed to be solved.At present,the main solution to the above problem is to increase the number of servers to share the server load by building a load balancing cluster of the web application server.In the process of load balancing,the decisive factor is the load balancing algorithm.This paper analyzes the research status of load balancing,summarizes and analyzes the related research of dynamic load balancing algorithm.With the requirements of Web cluster load balancing in the laboratory project,this article compares various load balancing technical solutions,designes an improved load balancing algorithm based on Nginx.The work completed in this paper are as follows:(1)Based on the research of Nginx architecture model,module system and underlying data structure,this paper analyzes the forwarding process and load balancing algorithm principle of Nginx in the upstream module,studies the improvement schemes for the inadequacies of the Nginx weighted round robin algorithm,and selects load information available for improving load balancing algorithm.(2)The weighted round robin algorithm for Nginx does not consider server performance dynamically,so this article designed a dynamic adaptive load balancing algorithm using LA strategy.The algorithm adjusts the server weights periodically by combining the back-end server state,and sets the judgment condition for load adjusting by the load variance;the learning automata-based strategy is adopted in the cycle,and the response change trend is used to optimize the weight,thereby improving the intracycle imbalance.(3)For the problem of centralized management of load information and the multi-service environment of the system,the paper designs a hierarchy dynamic load balancing strategy based on request.The system adopts hierarchical structure to manage load information,and the sub-domain adjust weight based on entropy.With the special requirements of requests in projects,the system adopts load balancing method in combination with the server state based on request classification within the sub-domain,which improves the cluster performance in a complex system environment.(4)On the project platform,the dynamic weighted round-robin algorithm designed in this paper and the request-based hierarchical dynamic load balancing strategy were tested and analyzed.The experimental results show that the algorithm and strategy designed in this paper have a good performance.