Nginx-Based Combined Dynamic and Static Load Balancing Algorithm Research

Abstract

To solve the problem that traditional Web cluster load balancing algorithm processes low efficiency under low network load and uneven load distribution under high load, based on the analysis of several common load balancing algorithm and combined with their advantages and disadvantages, a new load balancing algorithm which combines dynamic and static status was presented. In this algorithm, the threshold calculated by simulated annealing algorithm was used as interval partition. In the meanwhile, a static weighted polling algorithm under the base of the weight of each load index such as the performance of CPU, Memory, Disk IO and network bandwidth and a dynamic weight adjusted algorithm which collected real-time load information according to feedback from each server node were accepted. And through the comparative experiments of several algorithms, it is verified that the new algorithm proposed in this paper processes more reasonable load balancing, which has more advantages in the average response time and the actual number of concurrent. All in all, it meets the design requirements.