|
|
| |
 |
|
Oracle Tips
by Burleson Consulting
|
The Data Warehouse Development Life Cycle
Warehouse Project Management
When embarking on a data warehousing project, many pitfalls can
cripple the project. Characteristics of successful data warehouse
projects generally include the following aspects:
* Clear business justification for the project--Measurable
benefits must be defined for a warehouse project (e.g., sales will
increase by 10 percent, customer retention will increase by 15
percent). Warehouses are expensive, and the project must be able to
measure the benefits.
* Staff is properly trained--Warehousing involves many new
technologies, including SMP, MPP, and MDDB. The staff must be
trained and comfortable with the new tools.
* Insuring data quality and consistency--Warehouses deal with
historical data from a variety of sources, so care must be taken to
create a metadata manager that ensures common data definitions and
records changes of historical data definitions.
* Insuring subject privacy--Gathering data from many sources
can lead to privacy violations. A good example of privacy violation
is the hotel chain that targeted frequent hotel customers and sent a
frequent-user coupon to their home addresses. Some spouses
intercepted these mailings, leading to numerous divorces.
* Allow the warehouse to start small and evolve--Some
projects fail by defining too broad of a scope for the project.
Successful projects consider their first effort as a prototype and
continue to evolve from that point.
* Ensure intimate end-user involvement--Data warehouses
cannot be developed in a vacuum. The system must be flexible to
address changing end-user requirements, and the end-users must
understand the architecture so they are aware of the limitations of
their warehouse.
* Properly plan the infrastructure--A new infrastructure must
be designed to handle communications among data sources. Parallel
computers must be evaluated and installed, and staff must be
appropriately educated.
* Perform proper data modeling and stress testing--The data
model must be validated and stress tested so that the finished
system performs at acceptable levels. A model that works great at 10
GB may not function as the warehouse grows to 100 GB.
* Choose the right tools--Many projects are led astray
because of vendor hype. Unfortunately, many vendors inappropriately
label their products as “warehouse” applications, or they exaggerate
the functionality of their tools.
 |
Expert Remote DBA
BC is America's oldest and largest Remote DBA Oracle support
provider. Get real Remote DBA experts, call
BC Remote DBA today. |
 |
|
|
|
|
