Software systems are tested for their behavior under different load conditions due to the execution of load tests. In order to obtain meaningful test results, the software must be monitored while it operates in a defined test environment. An important part of this environment are test data. The test data must have correct properties and must be produceable in any desired volume and in a certain amount of time.
However, the generation of realistic test data for a complex software system requires high effort. In addition to a realistic amount and completeness of the data, semantic correctness is another important factor.
This thesis focuses on analysis of different ap- proaches for defining realistic test data. Developed on top of that, procedures will be presented which can be used to automatically generate test data based on existing test data definitions. The presented procedures will be examined and evaluated for their ef- fectiveness. Finally, a selection of products on the market will be evaluated based on the shown properties.
Original title:
Automatische Erzeugung von produktionsnahen Testdaten für Datenbanksysteme

Written by Patrick Dinger, September 2011

Second supervisor: Eduard Tudenhöfner, M.Sc., NovaTec GmbH