Databases Reference
In-Depth Information
important, where do you want to be? The data warehouse data architecture
provides the vision for the organization's data.
How data is handled and structured in the data architecture is independent
of your technology. This is why you need to pay such careful attention to your
data architecture. Sound data architecture gives you flexibility and avenues
for growth, and gracefully accommodates technical changes over time.
Although many organizations are getting started with data warehousing,
the early adopters have been doing this for over two decades. In some cases, the
original data design has been moved to a different database platform twice
since the initial implementation. You don't go in thinking that you will change
out your database, but if you have ten to fifteen years to reflect on, you probably
have already done so. Who can predict what technology might be available
in fifteen years? The bottom line is that you need to expect success and plan
accordingly. This means ensuring that your data fits together properly across
the enterprise.
Revisiting DW Goals
There are many different opinions about, and approaches to, handling and
structuring data in a data warehouse. Each different approach is trying to
address a common set of objectives. Regardless of the data architecture, the
goals are as follows:
Extract, clean, integrate, and make the data available to the business in a
timely manner.
Be able to adapt to changes at any point in the process.
Address enterprise integration requirements.
While these goals can be succinctly stated, they are actually complex and
have broad implications — implications that are specific to your organization
and the current state of the data. Examples of specific challenges that need to
be addressed for the first goal include the following:
The need to bring together data from disparate claims handling systems.
The primary source system does not have a sound database design, which
results in data being scattered throughout the database; duplicates; and
inconsistencies.
Extreme data volumes make it difficult to meet performance expectations
for loading and/or querying data.
Source systems do not retain history.
There is no common definition of a product (this is common in financial
services companies).
