The Design And Implementation Of Intenet Advertising Analysis System

With the Internet becoming increasingly popular, more and more enterprises prefermaking online advertisement.The Advertisement Analysis System (AAS) is used tojudge the effect of online advertisement, as the advertiser want to know theproductivity of the advertisement.ASS needs to handle massive information of users'request, and provides the statistics of analysis on massive data.The key of ASS is thesupportiveness the high concurrency and massive data analysis.To meet the needs of online advertisement analysis, this paper will design adistributed, high-capacity and extensible system based on Linux. ASS includes thefunctions of data collection, data analysis and result display, and provides an easy,quick and visualized platform for the advertisers.The structure of ASS is Brower/Servermodel, and its programming language is C++. Based on Linux Epoll, ASS implementsthe function of log collection of user action from web server.ASS uses HTTP protocoland JavaScript to collect data from web servers. To support the high availability andhorizontal scalability, ASS balances the load forward by LVS when collecting serverdata. This paper provides the comparison of the supported user amount for the webapplication development of CGI and HTTP. It proves that HTTP could support higherconcurrency.The thesis covers the introduction and research of open-source software hadoop. Hadoop is platform which is easy for application development and handling of massivedata.AAS abandons the relational database which uses stored procedure to analyze data.Instead, ASS adopts hadoop platform, which supports the massive data analysis.It provides good platform for future deep data mining.The thesis introduces the concept of thermodynamic chart. Besides the regularsheet, ASS innovatively uses thermodynamic chart to display the analysis result.Thepaper includes the detail description of thermodynamic chart implementation.And theresult of optimization trial on thermodynamic chart generation performance is good.Through research of AAS key points, this paper introduces a feasible solution whichcan collect, analyze and display massive log, and proves its feasibility through realpractice.