Shortest Semijoin Schedule for a Local Area Distributed Database System

The semijoin provides a means of reducing the amount of data transmission among sites in a distributed database system. Previously the semijoin has been studied mainly for reducing communication cost in an environment with global public communication networks. In a local area system, however, wide bandwidth is usually available and the communication cost is virtually negligible. In view of this, we adopt a simplified model of a local area network, imposing no constraint on the transmission line capacity and the communication processing capability at each site. For this model, an efficient algorithm for obtaining the shortest semijoin schedule, in the sense of minimizing the total number of semijoin transmissions, is developed. It is based on a schedule diagram newly introduced to represent the semijoin schedule.