Research On Cluster Management Based On ZooKeeper

In recent years,with the advancement and development of science and technology,the computing and storage requirements of human life are increasing,and the application of cluster technology is becoming more and more extensive.However,as the scale of the cluster expands,the problem of cluster management becomes increasingly prominent.Due to the heterogeneity of nodes in the cluster,the physical dispersion and the uncertainty of the number of nodes,the current cluster management system still has great drawbacks in communication complexity,data consistency and cross-platform visualization.In view of the shortcomings of the existing cluster management system,this paper proposes a cluster management solution based on ZooKeeper,and uses the graphical interface program Qt for visualization,and finally implements a cluster management system.Administrators can use the system to monitor the status of node servers in the cluster and complete the operations related to cluster management.After an in-depth study of the classic distributed selection algorithm(bully election algorithm and its improved algorithm)and traditional cluster management schemes,this paper proposes an election algorithm based on node performance and ZooKeeper.This paper also designs and implements a complete cluster management system according to needs analysis.The cluster management system can not only reduce the communication complexity and communication time of the internal nodes during cluster operation,ensure strong consistency of data between node servers,but also provide an intuitive and simple cluster management interface and meet the needs of the installation and use of cluster management systems on different platforms.Finally,we deploy and test the implemented cluster management system.The results show that the election algorithm based on node performance and ZooKeeper proposed in this paper is superior to the previous election algorithm in terms of communication complexity and communication time during election and the cluster management system designed and implemented in this paper has good practical value in function,operation and cross-platform operation.