Monitoring Data Analysis Of Distributed System Based On Visualization Technology

Distributed computing technology is one of the indispensable technologies for the Internet companies,the nodes of networks will generate large amounts of data every day.The visualization of their monitoring data could help user to better understand the status of the node,explore the infor-mation contained in monitoring data and design operational and maintenance strategies.Therefore studying how to apply visualization technology to the distributed nodes is of great significance and value.This thesis studies how to apply visualization technology to the node,solves the problems of data collection,data process and mapping,visual display of the monitoring data,builds a complete data visualization system.Logstash,Elasticsearch and Redis are chosen to achieve data collection and storage.In addition,this thesis chooses D3 to achieve ThemeRiver,Event TimeLine,Circos and Stack,and simulates the real scene to analyze the visualization to verify the feasibility and effectiveness of these visualization modules.The main research work focuses on the following three aspects:1.The data collection and storage of the monitoring data.Analyze and compare the popular server monitoring technology in market,select the appropriate technology to collect and s-tore the monitoring data from principle,adaptability,optimization and other considerations,provide a data source for visualization.2.The design and implementation of visualization.Investigate and analyze the demand for data visualization in the network,select the load data,the traffic data of network and log data as visual data,design and implement data visualization modules such as ThemeRiver,Event Timeline,Circos and Stack,and add interaction to enhance the practicality of these visualization models.3.Scene analysis of visualization model.Simulate some scenarios to demonstrate the effective-ness of these visualization models,verify the effectiveness and feasibility of the visualization model implemented in this thesis.Apply the visualization technology to the monitoring data analysis,could help users to discov?er potential problems in the nodes as soon as possible,adjust the node status,prevent catastrophic accidents of the server and provide a reference for making effective operation and maintenance strategies.