Font Size: a A A

Research And Implementation Of Auto-scaling Web Application System Based On Cloud Platform

Posted on:2019-07-07Degree:MasterType:Thesis
Country:ChinaCandidate:X Y WangFull Text:PDF
GTID:2348330542498752Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the development of Internet technology,the rapid development of e-commerce,the rapid increase in workload and the increasingly complicated business request have posed severe challenges to the high availability and service quality of applications.The traditional Web cluster model cannot cope with the load pressure caused by the sudden increase of access due to the fixed cluster size.Emerging elastic scalable Web architecture makes the cluster size can be dynamically adjusted with changes in work pressure,cloud computing technology makes it possible to obtain resources on demand,Docker technology greatly improves the speed and stability of application deployment.Therefore,this paper uses these technologies and elastic scalable model to solve the high availability problem of mall in high load scenarios.Based on the Web-based e-commerce system developed by the lab,this paper studies how to ensure the usability and service quality of Web applications in the current environment.The paper researched the current development of cluster mode,the cloud computing technology and Docker container technology.Then studied the theory of flexibilities,the current status of elastic scalability,and analyzed the modules of current common elastic scalability schemes.After analysis,improved and designed the load indicator selection module,load forecasting algorithm module,cooling time setting module:for the problem that the selection of workload indicators is too simple,innovatively proposed to select load indicators based on the type of workload which is forecasted by the recent workload data.Aiming at the deficiency of the current load forecasting algorithm,an improved ELR(Enhanced Linear Regression)algorithm based on algorithm fusion is proposed to correct the predictive value of the current forecasting algorithm.The accuracy of ELR algorithm was verified by comparing with the experimental results of control groups.Besides,the problem caused by using fixed time to cool after adjusting the cluster size has been improved.With all the above studies,an improved dynamic strategy was proposed and the ADDS system(Auto-Scaled Docker Deploy System)was developed based on the strategy,so that the application ensure the availability of highly available services.At last,the paper designed experiments to verify the availability of elastic scalable web applications and to test the dynamic expansion capabilities.It has certain research value to improve the high availability of applications.
Keywords/Search Tags:Cloud computing, Docker, Auto-Scaling, ELR
PDF Full Text Request
Related items