Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


TIME ESTIMATE: Varies widely, depending on the status and quality of the data model.


NORMALLY EXECUTED ONCE OR MULTIPLE TIMES: Once.


At the outset, a data model needs to have been defined. The data model needs to have done the following:


■■ Identified the major subject areas


■■ Clearly defined boundaries of the model


■■ Separated primitive from derived data


The following need to be identified for each subject area:


■■ Keys ■■ Attributes


■    Groupings of attributes


■    Relationships among groupings of attributes ■■ Multiply occurring data


■    “Type of” data


The output from this step is a confirmation that the organization has built a solid data model. If the model does not meet the criteria specified, then progress should be halted until the model is brought up to standards of quality.


PARAMETERS OF SUCCESS: The data model will have the following:


■    Major subjects identified


■    Each major subject with its own separate definition of data, including:


■    Subtypes of data


■    Attributes of data


■    Clearly defined relationships of data


■    Defined groupings of data


■    Defined keys


In addition, each group of data that will go into the data warehouse will have DSS data and operational-only data delineated. All DSS data will have its own time-variant key specified, usually as the lower order of a higher key.


DSS2—Breadbox Analysis


PRECEDING ACTIVITY: Data model analysis.


TIME ESTIMATE: From one day to two weeks, depending on how well the scope has been defined, how well the data model has been defined, etc.

Скачать в pdf «Building the Data Warehouse»