A study of nested-relational joins in mediator-based distributed environments

This thesis reports on the results of a study of the nested join operator in a mediator-based setting. As the rate at which data is accumulated increases, the need for integration of di erent (physical) Database Management Systems (DBMS) becomes apparent. The most fundamental and arguably important operation that is used in order to achieve this is the join-operator. The nested-relational model is a exible data model and can be used to describe data on practically all de facto data models such as relational, object-relational and XML-based DBMS. Although joins are not new, very little work has been done in the area of the nested-relational join operator. Some authors [15, 18, 19] have proposed nested relational join operators, however an implementation has not been reported. Information systems companies in their IT-landscape are often of a di erent make. This leads to di culties concerning data integration. To solve these di culties in an e cient way a mediation service, also known as a mediator, is introduced to their IT-landscape. The mediator facilitates communication between di erent databases and the end-user. The focus of this study is the design, implemention and evaluation of a nestedrelational join operator in a mediator-based context. This study presents a generalized version of the join introduced by Garani & Johnson [15]. This generalized join is implemented in a mediator-based context, which poses some additional constraints on the implementation. Subsequently the implementation is evaluated and the results are presented.