Database Reference
In-Depth Information
Figure 3. Data warehouse bus architecture
than the architectural patterns depicted in Figures
1 and 2. In addition, as being practiced by many
different industries, to use a star-schema (or, a
“snow-flake” schema) to model and keep a large
organization data in a single, multidimensional
model has not been very successful. In fact, many
organizations are struggling between the loading
time of the batch jobs and the query performance
and the end-user side.
Although the different architectural patterns
described in Figures 1, 2 and 3 have both ad-
vantages and disadvantages, all these patterns
are still being practiced in different enterprises.
Table 1 lists the important building blocks of data
warehouse architecture.
•
The
data warehouse must integrate data
from different business areas of the enter-
prise in order to ensure the “one version of
truth” of data.
The
•
data warehouse must provide users a
clear and complete catalogue of metadata.
The
•
data warehouse must ensure the qual-
ity of data.
The
•
data warehouse must have efficient
performance in order to minimize the time
period between the data is delivered from
operational system and the data is harvest-
ed by the data access tools.
The
•
data warehouse must ensure usability
of data when it provides data to any data
access tools, methods and BI applications.
An Architecture Prototype
These requirements can be implemented by
different data warehouse architectures with dif-
ferent practices. Figure 4 depicts one architecture
pattern that can be shaped towards fulfilling the
listed requirements. In this prototype architec-
To provide a background of the prototype data
warehouse architecture, we present the five
requirements that are most commonly used in
building data warehouses.
Search WWH ::
Custom Search