Research Of Server Load Balancing Algorithm For Internet Of Things

With the rapid development of the Internet and the Internet of Things(Io T)technology,the number of user requests received by the back-end servers has grown explosively,thus,load balancing technology was proposed to distribute these requests properly to reduce the server-side pressure and improve the overall performance of the server system.As a representative of the most used and excellent load balancing technologies,Nginx's low resource cost and high concurrency receive widely praises by the users.To solve the load imbalance problem caused by the existing technologies in high load situation,Nginx is used in this thesis to study two dynamic load balancing algorithms.The main research and contribution are described as follows:1.To solve the problem that the existing Nginx load balancing algorithms cannot reasonably allocate user requests in high concurrency situations,a dynamic weighting algorithm based on weighted round-robin algorithm is proposed.When load balancer distributes tasks to backend servers,the proposed algorithm not only takes into account the differences in the hardware performance of each node in the server cluster,but also reads and quantifys the real-time load of the back-end nodes.Then,the utilization of CPU,memory,disk I/O,and network,combined with hardware performance,are considered to periodically updates the final weights of server nodes,so that each backend server node can be assigned requests that match its current load capacity,thereby enhanced performance of back-end server clusters.Finally,the performance test tools of Httperf and Autobench are used to test the dynamic weighting algorithm proposed in this thesis.Experiment results show that the proposed algorithm can achieve a better load balancing compared with the weighted round-robin algorithm and the least connections algorithm in response time and actual maximum number of concurrent connections.2.Considering that Nginx's original load balancing algorithm did not take into account the different types of user requests and the different server resources occupied,a dynamic load balancing algorithm oriented to request types is proposed.First,the proposed algorithm judges and classifies the user's request type,selects the weight calculation parameter according to the type,and combines the collected server load information to calculate the dynamic weight corresponding to the type of the server.Then,the dynamic weights are used in the weighted least connections algorithm to perform load allocation.Finally,the AHP is used to obtain the parameters corresponding to each request type in the weight calculation model.Experimental results show that proposed algorithm is superior to the other two algorithms in terms of response time and maximum concurrency,thus can better achieve load balancing effect.