Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


Figure 5.8 shows that multiple multidimensional DBMSs are being fed directly from the same legacy applications. So, what is so wrong with this architecture?


The problems are as follows:


■■ The amount of development required in extraction is enormous. Each different departmental multidimensional DBMS must have its own set of extraction programs developed for it on a customized basis. There is a tremendous overlap of extract processing. The amount of wasted development work is enormous. When the multidimensional DBMSs are fed from


appl a



appl b



appl c



appl d



appl e



appl f



appl g



appl h



appl i



b


s


>


>


»


e


a



one of the primary reasons why the direct application to multidimensional DBMS approach


is unworkable

Figure 5.8 There are many applications, and there are many data marts. An interface application is needed between each occurrence. The result of bypassing the current level of detail is an unmanageable “spider web»



finance



marketing



human resources



management reporting



sales



production



engineering



accounting



manufacturing



actuarial



budgeting


the data warehouse, only one set of integration and transformation programs is needed.


■■ There is no integrated foundation when the multidimensional DBMSs are fed directly from the legacy systems environment. Each departmental multidimensional DBMS has its own interpretation as to how different applications should be integrated. Unfortunately, the way one department

Скачать в pdf «Building the Data Warehouse»