Improved dynamic load balancing algorithm based on Least-Connection Scheduling

At present, the commonly used load balancing algorithms are mainly round_robin and least_conn, but they do not take into account the constant changes of server performance over time. The paper on the basis of the Least-Connection Scheduling, proposes a dynamic load balancing algorithm, collection server static load factor and dynamic load factor to calculate the weight of the closest server real-time performance parameters, and then introduced the response time and the number of connections as statistics class load factor, finally calculated the current server load. By testing the built-in round_robin and least_conn strategies of Nginx and the dynamic load balancing strategies proposed in this paper, it is found that the improved load balancing strategy makes more efficient use of resources in the server cluster and improves the throughput and processing capacity of the system.