Data Warehousing, LANSA's Practical Solution - Part 2
|Product/Release:||LANSA- All Platforms|
|Abstract:||Data Warehousing, LANSA's Practical Solution - Part 2|
|Submitted By:||LANSA Technical Support|
Data warehousing, as it exists today, is an evolution of data analysis techniques that we have been using for years. Data warehousing is a more formalized methodology of these techniques.
For example, many sales analysis systems and executive information systems (EIS) get their data from summary files rather then operational transaction files. The method of using summary files instead of operational data is in essence what data warehousing is all about.
Some data warehousing tools neglect the importance of modeling the structure of the data in the data warehouse and focus on the storage and retrieval of data only.
These tools might have strong analytical facilities, but lack the qualities you need to build and maintain a corporate wide data warehouse. These tools belong on the PC rather than the host.
A corporate wide (or division wide) data warehouse needs to be scalable, secure, open and, above all, suitable for publication.
Scalable means that a data warehouse must be able to handle both a growing volume and variety of data and a growing number of users that can access it.
Secure means that the data warehouse administrator can centrally control who is allowed to access what data and when.
Open means that the data in a data warehouse is open to a wide range of query and other front end tools.
For this reason, a relational data base should be the first choice for a corporate wide data warehouse.
The proprietary data storage structures that are used by some data analysis tools can be fed from this central data warehouse.
Suitable for publication means that the data must be reliable, consistent , complete and well documented. The data must also be well laid out for retrieval by business users.
Business users require easy data navigation and speed of access. The many layers of file join relationships in a production system make data access too complex and also too slow. A more suitable data structure for a data warehouse is a dimensional model.
Scalability, security and openness depend largely on the choice of platform, database and tools. These provide the foundation for building the data warehouse.
To make data suitable for publication a combination business skills and data modeling skills is needed as well as the tools to build the various floors that make up the data warehouse.
Finally on top of all this (in the "penthouse" of the data warehouse building) are the tools to access the data warehouse.
However, no matter how glamorous and smart the query and data analysis tools are, if the underlying structure of a data warehouse is not well laid out, the decision support system (DSS) is doomed to fail.