Research On Microservice Load Balancing Based On QRT-GA Algorithm

With the wide applicatio n of Internet technology and the rapid development of emerging Internet services,the network traffic increases rapidly.In order to improve the reliability and security of information service system,The micro-service architecture emerges.The micro-service architecture combined with load balancing technology can distribute requests to each service node in a reasonable manner to balance the load and relieve the pressure.In this way,the high efficiency and reliability of the system are guaranteed under the condition of high concurrency.And it is very important to choose a suitable load balancing algorithm.The fastest response time algorithm which is a type of dynamic load balancing algorithm allocates requests according to the response time of requests,which can achieve the effect of dynamic real-time allocation.However,it is not accurate to measure the load of service nodes only by response time.This paper mainly studies the Nginx server and related load balancing algorithms,and makes the following research and improvement on the fastest response time algorithm:Firstly,based on the fastest response time algorithm,the paper introduces the system resource utilization,and proposes the QRT-GA algorithm combining the fastest response time algorithm and genetic algorithm.Aiming at the high concurrency scenario under the micro-service architecture,QRT-GA algorithm not only effectively improves the response speed,but also ensures the load balance of the cluster.Secondly,while improving the load balancing algorithm,the paper adopts the SARIMA model which is a machine learning sequential model to predict the size of the request traffic.The load of service nodes in each working cycle is measured by the sum of the predicted traffic and the actual measured load at the beginning of each working cycle,which solves the problem that the load is not updated in real time.Thus,the load balancing degree of the cluster and the operation efficiency of the whole system are effectively improved.Finally,based on the traditional micro-service architecture,the paper introduces the gateway cluster under the load balancer,and isolates different service clusters by service types,which effectively relieves the pressure of load balancer and load balancing nodes in the highly concurrent environment,and improves the availability and operation efficiency of the whole cluster.Through experimental comparison,compared with other algorithms,the QRT-GA algorithm proposed in this paper not only guarantees the load balancing degree of the whole system,but also effectively reduces the average response time of the system,increases the throughput of the whole cluster,and improves the load performance of the whole system.