One of the foundational structures of a business intelligence (BI) solution
is the data warehouse. To understand the most basic schema design for a data
warehouse (i.e., the star schema), you must first understand the relationship
between BI and data warehousing.
BI is a business management term that refers to the applications and technologies
used to gather, provide access to, and analyze data and information about a
company's operations. A data warehouse is a repository for a company's historical
data. Data warehouses can be physical or virtual, and they can be structurally
relational, quasi-relational, summarized, cubes, flat files, or a combination
of styles. Data warehousing is the set of technologies and techniques that you
use to build and manage the data warehouse.
Figure A illustrates the relationship between
BI and data warehousing. The data warehouse gets its data from a variety of
sources, including the extraction, transformation, and loading (ETL) staging
database, the online transaction processing (OLTP) transactional database, or
even directly from external data sources. Then, depending on the data needed
for a BI project, you can spin off multiple OLAP cubes (also called multidimensional
databases) from the data warehouse. (For example, a bank might analyze ATM transactions
for behavior, time of day, or queue information, whereas a retail operation
might perform a basket analysis on point of sale—POS—transactions.)
On top of this underlying architecture, with the ability to tap into any or
all of the data sources, is the BI software tools layer. This layer represents
numerous BI packages that you can use to analyze data, generate reports, and
find information for making business decisions. You can even feed information
into automated activities and other processes for additional analysis.
End of Article
