Databases in Grid Applications: Locality and Distribution

This paper focuses on two areas that experience in building databaseoriented e-science applications has shown to be important. Firstly, methods of promoting data locality are vital due to the high cost of moving data in servicebased distributed systems. Databases provide an excellent basis for achieving this due to their potential for moving computation to data. The paper also describes a new infrastructure that further promotes locality by enabling servicebased computations to migrate to data. Secondly, the ability to combine information from a set of distributed databases has proved invaluable in many applications. The paper describes the design of an adaptive distributed query processing system that is able to exploit facilities offered by an underlying grid infrastructure. In addressing these two areas, the paper gives an overview of some of the generic components that have been designed to simplify the integration of databases into e-science applications.