Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


NORMALLY EXECUTED ONCE OR MULTIPLE TIMES: Once per table.


SPECIAL CONSIDERATIONS: If the input to this step is incorrect or ambiguous, the amount of work required here can be much more than that estimated.


DELIVERABLE: Tables, databases physically designed.


Now the output from D3 and/or D4 is used to produce a physical database design. Some of the characteristics of the output include the following:


■■ Indexing


■■ Physical attribution of data ■■ Partitioning strategies ■■ Designation of keys


■    Clustering/interleaving


■■ Management of variable-length data


■    NULL/NOT NULL specification ■■ Referential integrity


The output of this step is the actual specification of the database to the DBMS or whatever data management software/conventions are adopted.


PARAMETERS OF SUCCESS: Done properly, this stage of analysis transforms all the considerations of logical design of data, performance, update, access, availability, reorganization, restructuring of data, and so on, into a workable database design.


Pl-Functional Decomposition


PRECEDING ACTIVITY: Requirements formalization.


FOLLOWING ACTIVITY: Context level 0 specification.


TIME ESTIMATE: Depends on the size of the system and degree of ambiguity (and how firmly and unambiguously the scope has been established). As a rule, two weeks for a reasonably sized system is adequate.


NORMALLY EXECUTED ONCE OR MULTIPLE TIMES: Once per phase of development.


From the requirements document comes the functional decomposition. The functional decomposition merely takes the broad function accomplished by the system and breaks it down into a series of successively smaller functions (down to a level sometimes called the primitive level).

Скачать в pdf «Building the Data Warehouse»