Research On High Concurrency Web System Technology Based On Container Orchestration

With the rapid development of the Internet and intelligent mobile terminals,more and more people around the world are using online user services such as social networking,games and shopping based on the Web system.The access of a large number of users to the Web site brings a large amount of concurrent traffic,which poses a new challenge to the high concurrency,high availability and high performance of the Web system.Due to the numerous application scenarios of Web system,its high concurrency,high availability and high performance are comprehensively affected by many factors and have great uncertainty.Therefore,the research on high concurrency Web system has important theoretical significance and practical value.Based on the Web system architecture,this paper deeply studies the principles and technical routes of Docker and Kubernetes,and determines the main factors that affect the high concurrency performance of the Web system.A web system architecture based on Kubernetes is designed which can effectively improve the performance of high concurrent processing.At the same time,the paper adopts Prophet model to build a set of Web traffic prediction model.LightGBM algorithm is used to build the container workload prediction model.The main work and innovative achievements of this paper are as follows:(1)Design of Web System Architecture Based on Kubernetes.Paper based on the technology of Kubernetes,from web traffic and container load forecast,the use of micro service technology architecture,at the same time monitoring module is added in the container cluster,web traffic prediction module,workload forecast module,designed a high concurrent processing can effectively improve the performance of the web system architecture,guarantee the quality of service web system.(2)Web traffic prediction based on the Prophet model.Based on the Prophet model,the paper builds a Web traffic prediction model.This model firstly studies the relationship between data sets through data visualization method,and then predicts future traffic based on historical Web traffic data through Prophet model,and decompositions trend items,seasonal and holiday effects of Web traffic data.Theoretical research and experimental results show that compared with traditional time series prediction algorithms,Prophet model shows great advantages in terms of prediction accuracy,model training time and stability,and can predict the trend of Web traffic extremely well.(3)Container workload prediction based on LightGBM algorithm.This paper builds a container workload prediction model based on LightGBM algorithm.By processing the data set,the model integrates or deletes the data features to reduce the noise interference of the model.At the same time,the training data set is used to train LightGBM algorithm.Finally,the test data set is used to verify and evaluate the prediction effect of the model.Theoretical research and experimental results show that the model can predict container workload with high accuracy,fast training speed and less memory.