Data Modelling

Building a data warehouse is an Iterative process.

It is imperative that the data warehouse developer have the attitude of moving
quickly through the steps of design and development, even though it is known that the
processing and data requirements are not complete at the outset of development.

DSS analysts’ requirements are discovered in a trial and error manner – in a mode of discovery. There is an entirely different mode of development that is required for data warehouse.

Prerequisite:

1) Data Model

2) Technology Selection

3) Sizing the data warehouse

4) Collecting Informational requirements

After Deciding the prerequisites, we have to do the first run:

In first run, we need to consider,

1) How much data needs to be loaded. –

THE FIRST ITERATION SHOULD CONTAIN DATA THAT IS LARGE ENOUGH TO BE
MEANINGFUL AND SMALL ENOUGH TO BE QUICKLY DOABLE.

There can be many ways the size of data can be cut. for example : by unit of time, by geographical area, by product line, by Customer Line, by activity type etc. Practically, there can be infinitely many ways to subset the data. So the meaningful subset depends alot on the fact that who’s going to be the user first Iteration of data.

There is a careful line to be walked here. On the one hand the data architect must take
care not to put too much data in the first iteration of development. On the other hand
the data architect must take care not to include so little data that the spontaneity of
discovery by the DSS analyst is limited.