2.1 Introduction

Readers of this chapter probably fall into one of three categories. They have:

1. Never designed a database before

2. Designed a database for a transaction processing-type system

3. Built a data warehouse system

In the latter case, the reader could skip this text or use it as a refresher, especially if his or her last database used Oracle. Therefore, this chapter is aimed at readers who fall into categories one or two, which may surprise the person who has previously designed a nondata warehouse database. Why? Because the skills and techniques used to create a database for a data ware-house will be different from those required for a transaction processing-style system. Consequently, while you will have a head start because some of the techniques are the same, it is very important to say to yourself: I am designing a different type of database.

So what is different about designing a database in a data warehouse? In a transaction-processing system, the designer's goal is to make the transaction complete very, very quickly, and the designer also has the benefit of knowing how the business will interrogate and use the data. Contrast that with a data warehouse, where, although queries must complete as quickly as possible, they could still take hours.

Another major problem is determining what information should be held in the warehouse and at what level of granularity it should be retained. This book will not discuss the techniques that can be used to determine what should be included in...