Effect of schema size on fragmentation design in multirelational databases

作者:

Highlights:

摘要

Vertical fragmentation in database design is aimed at improving database performance. In shared databases, referential integral constraints are maintained to ensure the consistency of database. The database fragment design differences and operating cost differences between large and small databases, in which differential levels of integral constraints exist, are presented. Experiments are conducted with database schema size ranging from 4 to 20 relations with six transaction environments and four fragment design methods. A total of 150 database design problems were solved. Statistically significant optimum design differences between small and large databases were observed. It is shown that the MRGB design is recommended in small and large database fragmentations. The other designs recommended are: unfragmented design in larger databases, trivially fragmented design in smaller databases and Hammer and Niamir's design in large databases with low updates and in smaller databases with medium updates.

论文关键词:Database management,physical design,data storage representation,database performance,database design,fragmentation,referential integrity

论文评审过程:Received 4 April 1988, Revised 26 May 1989, Available online 17 June 2003.

论文官网地址:https://doi.org/10.1016/0306-4379(90)90002-7