Databases Reference
In-Depth Information
Listing Data Elements
Because a data warehouse is for storing data, business groups may put
together a list of data elements that they think they need. They may, for
example, compile a list of the key performance indicators that are needed to
manage the business. These are often developed independently of any actual
data source research.
The problem with using this list as the basis for developing a data warehouse
is that it may involve pulling data from many sources and limits the work
to just the elements on the list. Once these are delivered through the data
warehouse, change begins. The first five requested elements might not turn
out to be what was expected. The second five data elements may be exactly
what was requested but turn out to not be helpful to the business. This leaves
only the last five data elements, which are part of what was wanted. Now,
work must begin to go in search of the next suite of data elements that are
of interest to the business. This requires starting back at the underlying data
sources.
Developing Functional Specifications
The purpose of developing functional specifications is to define what a com-
puter application will do, how it looks, and what user interactions are needed.
This is an important step in the design of an operational system. However, it
does not have the same value for data warehouse design. The use of a data
warehouse is quite fluid and can change from person to person and from
week to week. Data warehouse design is much more focused on the optimal
organization of the data needed for analysis. However, once the database has
been designed, it can be beneficial to develop functional specifications for the
extract, transform and load processes to populate the data warehouse.
Moving Beyond Immediate
Many people are unable to get beyond the immediate data or report frus-
trations. These frustrations may also be more operational in nature and not
anything that can be addressed with a data warehouse anyway. Listen care-
fully; you may identify an opportunity to address a problem in an existing
system. This helps the overall organization, but it does not provide data
warehouse requirements. If this happens during an interview, encourage the
group to let off some steam, and then direct the discussion toward the future.
Tell the group to imagine that all of these issues have been resolved. All the
data is clean, integrated, and at their fingertips. Let this thought sink in a
moment, and then ask the group what they would do now. This usually leads
the interview session into exploring the types of business requirements that
are needed to take the organization forward.
