6 Strategies of Data Warehouse implementation

Policy | data
6 Strategies of Data Warehouse implementation





In the implementation of the data warehouse, the need for theoretical guidance, with the development of data Warehouse technology, Data Warehouse implementation strategy has been from the original "Top-down" a model developed to six models.

1, "Top-down" mode





In this "Top-down" mode, the relationship between the data mart and the Data Warehouse is One direction, that is, data flows from the data warehouse to the Data mart.





2, "from the bottom up" mode



The bottom-up model starts with data marts that establish various departments or specific business issues, and the global Data Warehouse is built on these data marts.





The "bottom-up" model is characterized by low initial investment and quick results. Because it needs fewer people to make decisions when it comes to building departmental data marts, it solves smaller business problems. The "bottom-up" development model allows a unit to spend as little money as possible at the beginning of a data warehouse, or to assess the benefits of the technology before making an effective investment.





3, "Parallel development" mode







"Parallel development" mode refers to the establishment of data mart and the establishment of a global data warehouse under the guidance of a data model of a global Data warehouse. The parallel development model avoids the blindness of various departments in developing their data marts and reduces the data redundancy and inconsistency between the data marts. The relative independence of data mart in "Parallel development" mode is beneficial to the construction of global database.





The difference between "parallel development" and "Top-down" mode is that it satisfies the requirements of each department in the enterprise to establish a decision support system for the department in a relatively short period of time, so that they do not have to wait for a global data warehouse to establish their own data mart. The experience gained during the establishment of the data mart will help to optimize the data model of the global Data Warehouse and build the overall data warehouse.

The difference between "parallel development" and "bottom-up" mode is that it uses a unified data model of the global Data Warehouse to guide the construction of various departments ' data marts, which can resolve the inconsistency of data between data marts and reduce the data redundancy between data marts.

But the user's needs are not static but are changing with the advent of new technologies and applications. User needs change is not limited to requiring faster hardware, better data mining techniques, higher performance database software, more user-friendly graphical user interfaces include changes in the content of the information required, and changes in the required data, which will result in the addition of some new attributes to the data warehouse or data mart, A new table or a reorganization of a table or property that already exists. The new requirements will have an impact on the data model of the Data Warehouse and data mart.

The three modes of "Top-down", "bottom-up" and "parallel development" do not consider how to continuously respond to the data mart and data Warehouse construction of the user's feedback information. All three models consider the needs of users only in the process of building a data mart or data warehouse.

In order to solve the problem of user's demand change, three kinds of modes have been introduced in the industry.









4, "with feedback from the top down" mode







In this model, the feedback of the user's new requirements is divided into two stages. The first phase of the user's new requirements are constantly being fed back to the department's data mart, where the Departmental data mart generates its own demand changes based on the user's new requirements. The second Stage Department data Mart feeds its own demand change to the global Data Warehouse, and makes the corresponding change in the global Data Warehouse.





5, "with feedback from the bottom up" mode



In the "bottom-up" mode of "feedback", the change of requirement after the global database is established is mainly reflected between data mart and Data Warehouse. Data marts in various departments can maintain data consistency between each other and can adjust themselves according to the user's feedback information, the data Warehouse established in this mode can reduce the inconvenience caused by the change of user's demand after being put into use.





6. "Parallel Development with feedback" mode



In the "Parallel development with feedback" model, in the initial stage of development, the developer mainly establishes the Department data Mart under the guidance of the global Data Warehouse data model, and feeds back to the global Data Warehouse data model the problems encountered in the establishment process, the solutions and the opinions of the users. The global data model, while directing the construction of departmental data marts, also collects feedback from developers and departmental users and adjusts themselves to this information. After adjustment, the construction of the next phase of the global Data Warehouse will be relatively smooth.





The ultimate goal of the enterprise construction Data Warehouse is to enable the business managers to use the integrated Decision support environment of data warehouse conveniently to obtain valuable information, and to help managers to make quick and accurate judgments and find corresponding countermeasures for the changing environment. Therefore, the interface-friendly, powerful front-end tool for user Service should be effectively integrated into this new data analysis environment. Data Warehouse system is a data warehouse as the core, through data extraction tools, data conversion tools, query tools, reporting tools, analysis tools, data mining tools, etc. to meet the needs of users of information.



A typical database system structure diagram









In the selection of the above six types of data Warehouse implementation strategy, can refer to the specific implementation, integrated costs, benefits, results and other factors, the implementation of the merit, in order to achieve the success of data Warehouse implementation.