Databases Reference
In-Depth Information
the data for reporting and analytical use. This is the where most of the work
for a data warehouse is done including validation, cleansing, and integration.
The fourth step of data architecture is to publish the data that is ready for
access. Here, data is made available for reporting and analysis. The final step
is to begin using the data itself.
At each layer, several basic things need to be defined regarding the data:
1. What data will be stored here (reference and/or transaction data)?
2. What is the primary purpose of keeping the data here?
3. How will the data be structured?
4. What is the persistence of the data or how much history will be stored?
If it is no longer needed, how is this handled?
5. Who will be able to use the data here?
6. What type of data access will they have?
There are many different alternatives for a data warehouse data architec-
ture. A complete data architecture defines parameters and guidelines for the
preceding questions for each of the layers illustrated previously in Figure 9-2.
A review of the most widely adopted data architectures follows in the next
section.
A Closer Look at Common Data Warehouse Architectures
While there are many philosophical approaches to data warehouse architec-
ture, two are by far the most widely adopted: the
bottom-up
approach and the
top-down
approach. These were introduced as Kimball/Inmon in Chapter 1,
and each is explored in more detail here. Both are viable alternatives for imple-
menting your data warehousing environment. It is important to understand
the basic principles of each in order to select the approach that will work the
best for your organization. Having a good grasp of the most prevalent data
architectures can improve your understanding of any other approaches that
your organization may consider or choose to adopt.
These data architectures are each defined within the context of an overall
data warehouse methodology. The data architectures described next are pulled
from each methodology.
Bottom-Up Data Architecture
The bottom-up approach has been well documented by Ralph Kimball and
his colleagues in his
Data Warehouse Toolkit
series of texts. This approach has a
strong focus on the end user and the delivery of value to the business. The goal
is to ensure support across the enterprise for consistent reporting and analysis.
