Design And Implementation Of Management Analysis System Based On Data Mart For Mobile Communication Corporation

Management Analysis System of Mobile Communication Corporation belongs to CRM of Analysis kind. It uses Data Warehouse, ETL, OLAP analysis, Data Mining etc to display the data stored in Data Warehouse to kinds of users. The user can get the analysis results from Email, or from Web.At present, Management Analysis System has been built in Mobile Communication Corporation. There are three parts in the system: data obtaining, data storing, data access. Now the system for each province has been in use.One of the principles of management analysis system is centralized storing and multiple applications. In order to provide corresponding data support for each city, Mobile Communication Corporation presents building the Data Mart for each city.Management analysis system based on Data Mart of mobile communication corporation contains three parts: data obtaining, data storing, data access.Data obtaining layer consists of data source and ETL(extract, transform, load). In order to assure consistency and completeness of data, Data Warehouse of province management analysis system is data source of Data Mart. So, the data in Data Mart and Data Warehouse can keep synchronism. ETL is implemented with TCL programs, which are put in dispatching system, so that the data can be provided in time.Data storing layer is mainly data storing and management. The data in Data Mart is managed by Metadata. The data is divided into eight subjects: guest subject, guest service subject, service subject, accounts subject, service utilization subject, resource subject, competition subject, sales subject. In Data Mart, there are two kinds of data: integrated data and detailed data.In addition, Data Mart is divided into three layers: data base layer, application support layer, data virtual access layer. Data base layer stores the basic data of data warehouse, and it produces data mart interface layer.Application support layer stores data of each citys, and in implementation, each city for a user, either one database together or one database for each city. Data virtual access layer is the support layer of application support layer, which uses synonym to support accessing data in application support layer. Data Mart uses star modal.Data Access is mainly displaying by web in the system. It uses J2EE structure, and there are there layers: Presentation layer, middle layer, and data access layer. In presentation layer, we use the Struts Framework. In middle layer, we use the Spring Framework. In data access layer, we use O/R Mapping, which is implemented with Hibernate. For tables of month and day, we use JDBC.Struts is the most popular framework in presentation layer. It uses MVC(model-view-controller) design pattern, and there are three parts: model, view and controller. Model is used to model the state of the system, view is the presentation of model, and it displays the result to user, controller receives the request from view, and updates the state of system.Spring Framework has several advantages as follows:v Spring can use only one configuration file in the system, and we don't need several files.v Spring can least depend on APIs belongs to Spring, many business objects can be used independently.v Spring provides data persistency framework, and can use JDBC or O/R Mapping to access data in database.O/R Mapping is selected in data access layer. We define one BO(Business Object) for each table. For dimension table, we use the combination of Spring and Hibernate to access database, so that we can shield the dependence of data for upper layers. For monthly table and daily table, we use the combination of Spring and JDBC to access database, because these tables are variable with time. We can only define one BO, and depends time parameter to determine which table to select.Up to now, the management analysis system based on DataMart hasbeen built roughly, and it has some properties:V It has reliable data. Data in Data Mart is derived from Data Warehouse in province management analysis system. So it keep consistency with the data in original system. For special data of each city, it can keep quality.VAs one branch and development of province management analysis system, the management analysis system based in Data Mart has well compatibility and expansibility. Data Mart can share some functions of original system, such as power management, data security. On the other side, Data Mart has its own branch space, and it's easy to add new functions to it.It's a long process to build a perfect management analysis system based on Data Mart. Data quality is the first premise in developing management analysis system. As requirement increases, personalize data will increase, and data quality needs more attention. In data access layer, OLAP, Data Mining etc will add to the system, so that it can provide more useful information to the decision makers.