Design And Implementation Of Advertising Business Data Management Platform

Big data has closely related with the enterprise in the internet advertising industry, Data Management Platform is a very important part in the system of internet advertising. Through the depth analysis of big data, marketers can understand consumers purchasing behavior, advertising will be more targeted to convey to the "Rigid Demand Customers". Site administrators can find out the website user access behavior and preferences, content will be more focused to push for users, improve the users viscous, and increase page view of website. In addition, The Data Management Platform provides query interface in multiple dimensions, so that advertisers and operators can check all data display.The paper comes from the practical application of AsiaInfo big data department. The advertising data management platform is designed and implemented. According to the existing advertising business status, the advertising data management platform made a detailed requirements analysis, including the division of function point analysis, functional requirements and non functional requirements analysis; From system architecture, system topology and system construction make a summary of the design, a detailed description of the system itself, three layer architecture and dependence on other services and the online server deployment; For the detailed design of each functional module, the important module of the design process description and class design, and to make clear the function of each point in the implementation of method call relation; Finally completed the development of the whole data management platform. In order to provide better flexibility and maintainability, a traditional B/S architecture of commercial system and JavaEE development are used by this project. The database is based on MySQL, use the j Query framework for the development of view page. In order to support massive log fast collection and statistic analysis, the system uses Hadoop and Spark distributed parallel computing framework. In order to improve the efficiency of query, the system use a lot of performance optimization such as Hive, spark SQL, caching. The author finished requirements analysis, design, development and deploy on-line in the project, the main work includes the task scheduling, system log collection, system log processing, data query display, the crowd management, crowd insight and dynamic material, as well as each module design, development, and performance optimization. From the introduction of advertising business so far, data management platform using agile development, has been developed and launched eight versions, all functional requirements at present data management platform are already on the line. The eight version in addition to increased functional, performance is gradually improved, the latest version has been stably operated for two months, provide important data support for the Demand Side Platform. From the point of data, through the data management platform, the Demand Side Platform effect of advertising has been enhanced, and through the visual data display, also have more advertisers accepted our advertising platform, the data application created income for advertisers and company.