Database Reference
In-Depth Information
Figure 1. Point-to-point integration architecture
view of data in the whole organization.
Figure 2 depicts an architecture where all
the operational systems deliver their production
data into a single repository and data marts and
different applications and users can use these
production data for purposes like analysis and
reporting. Here the “production data warehouse”
contains production data from different operational
systems. There is normally no transformation
to the production data after it is loaded into the
production data warehouse.
In this “production data warehouse” architec-
ture, the management of data at different opera-
tional systems is eased by putting the data into
a single repository. It becomes easier to get an
overview of all kinds of data at an organization.
Since the data delivery from operational system
to the production data warehouse can be ensured
through “service level agreement (SLA)” and there
are no further operations over the data after they
are loaded into the data warehouse, the efficiency
for loading the data at data marts, business intel-
ligence (BI) applications or for different users to
directly query the data is quite optimal. On the
other hand, since there is no data transformation
in the data warehouse, there can be quite a big
amount of data redundancy and inconsistency
within the warehouse. Specifically, when two
or more operational systems contain the same
area of data, such as customer information, these
data will all be loaded into different tables in the
production data warehouse without any further
conformations and standardizations.
Figure 3 shows the typical data warehouse bus
architecture defined in Kimball & Ross, 2002. This
architecture is generally divided into 4 layers. The
first layer contains all the operational systems.
These systems deliver production data to the next
layer according to the SLAs. The second layer,
called “data staging area,” contains data deliveries
from different source systems that are temporally
kept in its original format or in database, such
as flat files or relational tables. The data staging
area involves data extraction, transformation, and
Search WWH ::




Custom Search