In this paper we consider the problem of efficiently checking integrity constraints. Since parallelism is a promising strategy for improving efficiency in database systems we apply parallelism to integrity constraint checking. In a first step we compile the constraints into an extended relational algebra. The extensions of the algebra are due to the complexity of the integrity constraints which come from a real application. The algebraic expressions are input to a parallel constraint engine Pace. We use dataflow-driven, set-oriented constraint execution. Horizontal and vertical parallelism is supported through data distribution and pipelining between extended relational algebra operators. Pace has been implemented on a KSR1-8 MIMD-multiprocessor system with virtual shared-memory. Finally, measurements performed with a parallel main-memory database demonstrate the processing costs that can be achieved by parallel integrity constraint evaluation.
@INPROCEEDINGS{Her95c,
AUTHOR = {U. Herzog and R. Schaarschmidt},
TITLE = {Parallel Execution of Integrity Constraint Checks},
BOOKTITLE = {Proceedings of the 4th Int.\ Conference on
Information and Knowledge Management (CIKM)},
PAGES = {82-89},
ADDRESS = {Baltimore, MD},
MONTH = nov,
YEAR = {1995}
}