Design And Implementation Of Lightweight Container Cloud Platform For Distributed Applications

Compared with the virtual machine-based IaaS cloud,the container cloud has higher platform resource utilization and more convenient application deployment process.Therefore,in recent years,more and more enterprises have begun to use the container cloud platform to host their distributed applications.However,in order to use the container cloud platform,users must have knowledge of the container domain,so the platform has a high threshold.At the same time,due to the complex technical architecture of container cloud,it is difficult to troubleshoot applications and platforms.Therefore,the operation and maintenance cost of the platform is relatively high.In response to the above problems,a lightweight container cloud platform for distributed applications is designed and implemented.In order to solve the problem of high use threshold,this platform adopts a lightweight distributed application management architecture to realize oneclick deployment and management of applications.In order to solve the problem of high operation and maintenance cost,a root cause localization method named SHARP-RCA(Snapshot comparison and HieArchical peRsonalized Pagerank based Root Cause Analysis)is proposed.The method first uses a graph generation method based on snapshot comparison to build the observable data of the application into an aggregated graph containing component topology and container distribution.Then the method uses the root cause ranking algorithm based on Hierarchical Personalized PageRank(HPPR)to search the aggregated graph to obtain the root cause ranking.Through the comparison of normal and abnormal snapshots and rule filtering,this method effectively solves the problems of data missing and graph structure changes caused by various container dynamic scenarios.Through hierarchical personalized random walks,the method not only accurately captures the root cause of the container dimension,but also locates the root cause of the infrastructure dimension.A series of experiments show that SHARP-RCA performs well in all common container scenarios,achieving an overall accuracy of 78%.Compared with the traditional manual localization method and the random walk-based root cause localization method named MicroRCA,the accuracy rate is increased by 254.5%and 30.0%respectively.Firstly,this thesis introduces the research background of the lightweight container cloud platform for distributed applications.Secondly,based on the research on other container cloud platforms in the industry,the requirements of the lightweight container cloud platform are analyzed.Then,the SHARP-RCA method and its experimental results are detailed.Next,the design and implementation of a lightweight container cloud platform for distributed applications is described.Finally,the effectiveness of the platform is verified through a series of tests.