Using semantic knowledge of partitioning and allocation of data in distributed databases

Designing distributed databases involves determining how data sets are to be partitioned and spread over multiple sites in a network in order to achieve good performance. This paper presents a methodology for partitioning and allocating data while designing distributed databases. In a relational database environment, the knowledge available in the database schema such as identifier domains (primary keys, foreign keys) defines semantic relationships within and between the relations in the schema. This semantic knowledge is used along with use knowledge such as user views to form the design alternatives. These can subsequently be evaluated using performance measures like response time or total processing cost.