Databases Reference
In-Depth Information
into the data warehouse. The content and purpose of these ETL data stores is
similar to those described for the bottom-up approach. The characteristics of
the data warehouse are:
1. What data will be stored here (reference and/or transaction data)? His-
torical data, needed to satisfy business requirements, will be stored in the
data warehouse. This includes all historical transactions and history of
all reference data that can be captured within the scope of what has been
addressed to date.
2. What is the primary purpose of keeping the data here? The purpose is
to facilitate the integration of data across the enterprise and to provide
a single platform to store the data. The data should support analytical
and reporting needs of the organization. The data warehouse serves as
the central component of the enterprise solution. All of the data is loaded
here and then distributed as needed.
3. How will the data be structured? The data structures follow traditional,
sound data design principles. The data is organized into subject areas. It is
organized in a traditional third normal form. The modeling technique for
this data organization is called entity-relationship modeling (see Chapter
7). This data represents the pure data relationships, and when done
properly eliminates redundancy.
4. What is the persistence of the data or how much history will be stored?
In theory, all of the history that may be needed by the business is to
be retained in the data warehouse. All of the historical transactions
are retained here. The data warehouse also includes all changes to any
reference data for the entire period of time that the transaction data is
needed. Data may be archived or deleted from the data warehouse when
it is no longer needed by the business.
5. Who will be able to use the data here? Theoretically, business users may
access the data in the data warehouse. However, the complexity of the
data structures can make it difficult to write queries, and the data is not
organized for optimal query performance. It is not recommended that
large numbers of users access the data warehouse structures directly.
Data that is needed by many people should be published, as described
in the next layer. There are probably a handful of highly sophisticated,
data-savvy users with strong technical skills who are capable of accessing
the data warehouse directly. These are often market researchers, actuaries,
or statisticians.
6. What type of data access will they have? The primary access to the data
warehouse is through programs and utilities used to distribute data to
the data marts. Limited user access to these tables is possible, but not
recommended for ad hoc querying.
