Oracle Tips  by Burleson Consulting

The Data Warehouse Development Life Cycle

The Structured Specification

In a traditional systems analysis, the DFD does not stand by itself. Rather, the DFD is augmented by a data dictionary that describes all of the data flows and files, and a set of process logic specifications that describes how each process transforms data flows. A process logic specification (sometimes called a mini-spec) can be expressed as structured English, decision tress, or any of the many other techniques used to describe how data flows are being changed. Listing 3.1 shows a sample process logic specification using structured English.

Listing 3.1 A Level 1 DFD process logic specification for the place_order behavior.

Minispec for PLACE_ORDER:

IF total_amt > 1000
Check CREDIT_RATING in CUSTOMER
IF CREDIT_RATING = 'BAD' then reject order
ELSE Store ORDER RECORD
end if

FOR (each item on the order)

-- compare QTY_ORDERED in ORDER with QTY_ON_HAND in ITEM
IF QTY_ON_HAND < QTY_ORDERED
Remove item from order
Prepare backorder slip
ELSE

-- Add the item to the order
Subtract QTY_ORDERED from QTY_ON_HAND.
Move QTY_ORDERED to QTY in LINE_ITEM.
Store LINE_ITEM record.
NEXT ITEM

In traditional systems analysis, data dictionary definitions for all data items are normalized, or grouped, into database entities, which become entity/relational models in systems design. Eventually, the E/R models become relational tables during system implementation. The identification and grouping of data items constitutes the entities that will establish the basic entity/relation model for the database engine. For example, consider the interaction between the place_order process and the data files in Figure 3.3. The process place_order uses the data flow total_amnt to check the credit rating in the customer file. It also uses qty_ordered to check the inventory level for the item, then moves qty_ordered to the line_item record, and finally stores this record in the file.