Design and development of modern data warehouses and BI applications based on Oracle technologies

OData support
Kovács Ferenc
Department of Automation and Applied Informatics

In the field of database management there is still a significant distinction between systems

that process millions of transactions and those that extract information from enormous

volumes of data in order to support business analysts. These decision support systems not

only need to operate on vast numbers of data, but also have to satisfy the enterprise requirements

of changeability, auditablity or scalability. Nowadays the remedy for these business

needs –apart from Bill Inmon’s always-deployable normalized concept– tends to be the

dimensional modeling (star schema) toolkit of Ralph Kimball. However, both approaches have

certain drawbacks which a certain Daniel Linstedt tried to eliminate –while maintaining their

respective advantages– by devising and publishing a methodology of his: the Data Vault Modeling.

After its initial theoretical success the Data Vault Modeling, however, has not gained

much popularity in practice.

Beyond the familiarization of the methodology the chief goal of this thesis was to work

out a solution by the support of which relational databases can simply and intuitively be

transformed into Data Vaults. The solution involved two tasks: firstly, a general implementation

independent methodology was established to summarize the functions which are due in

case of transforming relational structures (e.g.: converting schemas, loading data or transforming

queries). Secondly, a prototype application –based on the aforementioned methodology–

was realized using Oracle PL/SQL and Java. The application delegates the transformation

logic into a low-level server-side component thus facilitating portability and the advantages

of loose coupling. The user interface is provided by a Java application that invokes the

transformation functionality through a connector library.

The prototype application provides means of efficiently transforming either legacy systems

or new –but orthodoxically designed– databases/data warehouses into Data Vaults.

Moreover, it enables the qualitative and quantitative comparison of different types of architectures.

Therefore the application can serve as a bridge between conceptions and offers a

basis for familiarization and further analyses of the Data Vault Modeling methodology.


Please sign in to download the files of this thesis.