Databases Reference
In-Depth Information
Corporate information factory
departmental
data marts
DSS
applications
accounting
finance
sales
marketing
ERP
(rpt)
staging
area
CRM
ETL
eComm
edw
Bus Int
changed
data
capture
exploration
warehouse/
data mining
global
ODS
oper
mart
cross media
storage manager
ERP
corporate
applications
granularity
manager
firewalls
near line
storage
local
ODS
session
analysis
dialog
manager
cookie
cognition
web log
tapes
by Bill Inmon and
Claudia Imhoff
copyright 2001, all rights reserved
Internet
preformatted
dialogs
©
Web environment
FIGURE 6.4
CIF architecture (reproduced here with explicit permission of authors).
its closest form to the source, and subsequent layers are added to the architecture to support
analytics, reporting, and other requirements.
2.
BUS architecture, also known as the Kimball architecture, is based on a set of tightly integrated
datamarts that are based on a dimensional data model approach. The data model allows business
to define and build datamarts for each line of business and then link the diverse datamarts by
aligning the common set of dimensions.
In the information factory architecture shown in
Figure 6.4
, the data warehouse is built in a top-
down model, starting with multiple source systems across the enterprise sending data to the central-
ized data warehouse, where a staging area collects the data, and the data quality and cleansing rules
are applied. The preprocessed data is finally transformed and loaded to the data warehouse. The third
normal form of the data model enables storing the data with minimal transformations in to the data
