Integrated Method for Efficient Design of Distributed Databases

摘要

The database design process is essential for an efficient implementation ofdistributed database systems. Known properties such as 'Completeness', 'Reconstructability' and 'Disjointness' are used to characterize the correctness of the data fragmentation. This report presents some experimental results on the quality of horizontal and vertical fragmentation methods. The results obtained lead to an improved method for horizontal fragmentation which integrates the data fragmentation and data allocation. It is shown how the generated fragmentation schema can be represented to support an efficient processing of queries and updates. Algorithms for the fragment localization and the tuple insertion are given.