Research On Nginx Load Balancing Technology

The Internet has penetrated into all aspects of people's lives in today's society. In order to cope with the increasing number of network users, the network operators generally use Web server cluster technology. Load balancing is the key technology of cluster. It can reasonably distribute the user request to make the request timely response, and the cluster server can be used efficiently. Nginx as an excellent HTTP and a reverse proxy server, it has been widely deployed in many companies around the world. But as the Internet provides more and more types of services, the heterogeneous cluster composed by the server nodes with different computing ability exist in a large number of network services. The original load balancing strategy in Nginx has been difficult to adapt to today's needs. The research on the load balancing strategy has become one of the hot issues in the academic and industrial circles.In this thesis, we studied the load balancing strategy of Nginx in heterogeneous cluster and multi service environment. We improve the service response of all kinds of network services and provide to users with better service in a complex environment. Specifically, the main work in this paper is as follows:First of all, in view of the problem that the web server cluster load is constantly changing with the user's service request, and the Nginx default load balancing policy does not consider this problem, a balanced strategy based on the minimum residual load is proposed. This strategy reduces the load information of the back-end Web Server by controlling the load reporting cycle and load change rate, so as to further to reduce the pressure of the Load Balancer. The proposed strategy is compared with the Nginx weighted polling strategy and the "rt_dyn_load" algorithm. The experimental results show that according to the clusters of different load to choose suitable load balancing strategy to handle user requests can effectively reduce the average delay of Web Server Cluster in Nginx deployment environment, and improve the system average throughput.Secondly, aimed at the diversity of business types and the heterogeneous cluster has been popularized, a load balancing strategy for heterogeneous clusters in multi service environment is proposed. The strategy first to divide the back-end web server cluster into a plurality of business clusters according to the cluster server processing performance and the estimation of each kind of service requests. Furthermore, user requests will be classified before distribution, then distribute the request to the appropriate server to process based on the minimum residual dynamic load balancing strategy. This paper further puts forward the server dynamic partition strategy in order to effectively avoid the user requests to focus on a particular class of business at a certain time. The proposed strategy is compared with the Nginx weighted polling strategy and the "rt_dyn_load" algorithm. The results showed that the proposed strategy can effectively reduce the average delay of Web Cluster and improve the average throughput of the system in a large quantity of concurrent requests, and the peak value appeared later than that of the existing load balancing strategy.It is shown that the research on Nginx load balancing technology and the improvement of the strategy about single and multi-service application scenarios can effectively improve the throughput of Web cluster system based on Nginx deployment, and reduce the average time delay of the system. The research have a good application reference value.