Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


The data warehouse is designed based on the data model. Some of the characteristics of the ultimate design include the following:


■■ An accommodation of the different levels of granularity, if indeed there are multiple levels of granularity


■■ An orientation of data to the major subjects of the corporation ■■ The presence of only primitive data and publicly derived data


■■ The absence of non-DSS data ■■ Time variancy of every record of data


■■ Physical denormalization of data where applicable (i.e., where performance warrants)


■    Creation of data artifacts where data that once was in the operational environment is brought over to the data warehouse


The output of this step is a physical database design of the data warehouse. Note that not all of the data warehouse needs to be designed in detail at the outset. It is entirely acceptable to design the major structures of the data warehouse initially, then fill in the details at a later point in time.


PARAMETERS OF SUCCESS: When this step is properly done, the result is a data warehouse that has a manageable amount of data that can be loaded, accessed, indexed, and searched in a reasonably efficient fashion.


DSS7—Source System Analysis


PRECEDING ACTIVITY: Subject area analysis.


FOLLOWING ACTIVITY: Program specification; data warehouse design.


TIME ESTIMATE: One week per subject area.


NORMALLY EXECUTED ONCE OR MULTIPLE TIMES: Once per subject area. DELIVERABLE: Identification of the system of record.


Once the subject to be populated is identified, the next activity is to identify the source data for the subject in the existing systems environment. It is absolutely normal for there to be a variety of sources of data for DSS data. It is at this point that the issues of integration are addressed. The following represents the issues to be addressed here:

Скачать в pdf «Building the Data Warehouse»