Research And Application Of MQTT Server Load Balancing Technology

With the development and rise of the Internet of Things technology,more and more devices are connected to the network.Compared with the traditional Internet,the Internet of Things device imposes higher requirements on communication technologies because of its many limitations in processor capabilities and network bandwidth.In recent years,mobile devices and smart terminals that use MQTT as a communication protocol have shown rapid growth.MQTT server load balancing technology has been proposed to solve the problem of concurrent access by a large number of users and to eliminate the phenomenon of unbalanced load on the nodes in the system.Nginx as a good HTTP server and reverse proxy server,load balancing strategy is the core technology,using Nginx as a load balancer is also a better solution currently,load balancing strategy directly affects the performance of the system.This paper aims at the disadvantages of the current common load balancing algorithm that does not consider MQTT server status information,as well as single selection of load parameters and fixed weight values,through the analysis of the source of Nginx,working principle of the main module,improve on the basis of existing WRR(Weighted Round Robin Scheduling),this paper proposes a dynamic load balancing algorithm for MQTT server,it mainly includes the load parameters collection module,the weight calculation module,and the weight modification module.The load balancing algorithm designed in this paper achieves the calculation of real time processing capabilities of each server by collecting performance parameters,and the introduction of MQTT server status parameters,through the design of the weight calculation formula is converted to the corresponding weight,Combining the load parameter collection strategy and the weight value reporting strategy reduces the processing pressure of the load balancer on the load parameters,and proposes a good solution for the problem of unbalanced load in the MQTT server cluster.Finally,this paper also builds the actual test environment of MQTT server cluster system,using the test tool Tsung and compares the proposed algorithm with the existing WLC(Weighted Least Connection Scheduling)and WRR(Weighted Round Robin Scheduling).The test results show that the load balancing algorithm proposed in this paper can achieve better load balancing effect and achieve the expected goal.