Research On Microservice Cluster Load Balancing Technology

Compared to traditional software architectures,microservice architectures have significant advantages.When using a monolithic architecture development project,it is often faced with problems such as complex business logic implementation and difficulty in controlling the granularity of software modules.The microservice architecture divides large and complex single-frame architecture applications into a set of tiny services.Each microservice is refined into a single business function according to the specific business responsibilities it is responsible for.Each service can be easily deployed into the production environment independently.In the process,and can be easily extended and changed.In the microservice architecture,in order to obtain high-performance network services,clustering technology is usually used to build high-performance servers to accomplish some tasks that the original single server could not accomplish.In a cluster environment,the resource utilization of each server often has a large gap,which causes each server to fail to assume the request tasks issued by users in a balanced manner,and ultimately the overall performance of the cluster is greatly reduced.This paper studies the microservice cluster load balancing technology,which aims to provide an effective load balancing framework and algorithm to improve the processing power and quality of the cluster system.In this paper,based on the analysis and siunnary of microservices,clustering,load balancing and other technologies,the microservice cluster load balancing technology is deeply studied.Firstly,a microservice cluster load balancing framework is designed,and the whole implementation process of microservice cluster load balancing is given.The purpose is to improve the execution efficiency of microservices and achieve the purpose of rational utilization of server resources.A server resource prediction algorithm based on ARIMA model is presented.The algorithl predicts the resource usage of each server for the next second according to the history of the server and the current resource parameters,namely CPU utilization,disk utilization,memory utilization and number of connections.Happening.A dynamic weight-based microservice cluster load balancing algorithm is proposed.The algorithm calculates the weight of each resource parameter by using the next second resource usage predicted by the ARIMA model,and then calculates the load of each server based on the resource parameter weight.Select the server with the least load to execute the request.Calculating the server load with dynamic weights has better timeliness,and it can achieve better load balancing,reduce request response time,and improve throughput when there are more requests.Next,the microservice cluster load balancing framework and the dynamic weight based microservice cluster load balancing algorithm are implemented and compared with some related algorithms.Finally,an application case is given to verify the effectiveness of the microservice cluster load balancing framework and algorithm studied in this paper.