Transcript Introduction to Data Warehousing

DATA WAREHOUSING
A Curriculum on designing a
Data Warehouse
What is a data warehouse?
A database filled with
large volumes of
cross-indexed
historical business
information that users
can access with
various query tools.
The warehouse usually
resides on its own
server and is separate
from the transactionprocessing or “runthe-business” systems.
Benefits of a data warehouse
Intended to provide an architectural model
for the flow of data from operational
systems to decision support systems
DW involves a many record analysis, during
which all data has to be locked
Used to discover trends and patterns
Present opportunities
Identify problems
Design considerations
Database structure
Normalized design vs. star
schema
Tradeoff between flexibility
and performance
Metadata
Keeps data warehouse
uniform and updates
possible
International company?
Integrating the catalog
into back-office functions
Simplifying the
maintenance that
merchants must manage in
regard to both presentation
and back-office
integration
Making an interface easy
to navigate for new
visitors
Cio.com (designing principles)
Strategic Use
Strategic planning
provides
Direction
Focus
Perseverance
Flexibility
Strategic planning is used
to create Sustainable
Competitive Advantage
Data Warehousing
provides competitive
advantage in the battle
against time
Past - learning from
mistakes
Present – allows for
maneuverability
Future – leaders can act on
foresight
Data warehouse Development Lifecycle
Iterative process (Rapid application
Development)
Produces a working prototype that is
retained and improved upon, rather than
discarded
Kelly, DW in action
Keys to success
Create data model
Know the user’s
desires/requirements
Build small and smart
with iterations
Common data definitions
Patience!
Cio.com (blueprint for success)
Performance measurement
Database familiarity
Tools selection
Developer’s angle
Cost/benefit of data warehouses
New insights into
Customer habits
Developing new products
Selling more products
Cost savings and revenue
increases
Cross-selling of products
Less mainframe computer
storage
Identify and target most
profitable customers
Capital outlay and
development/training time
can be extraordinary.
In 1997, companies spent
an average of $1.9 million
on DW projects
Quality of system output
Levels of risk
Intangibles
Cio.com (middle ground)
How to avoid failure:
30-50% failure rate!
Lack of executive
support
Poor management
Unrealistic
expectations
Lack of user demand
Lack of business
objective
Unstructured user
interviews
Poor data model
Where is source data?