Reduction filters for minimizing data transfers in distributed query optimization

It has long been recognized that query optimization in distributed database systems is an important research issue. The challenge is to determine a sequence of operations which will process the query while minimizing the chosen cost function. Finding the optimal optimization for a general query is an NP-hard problem so, in general, heuristics are employed to find a cost-effective and efficient processing method. We present a novel approach to the problem, which uses reduction filters, with the objective of minimizing the total volume of data transferred in the network. We assume a distributed relational database management system and select-project-join queries. This means that we have a number of relations, each located at a different site in the network, which must be joined and the result made available at some distinct query site. Our technique is to reduce the relations, before shipment to the query site, using reduction filters and thereby significantly reduce the total communication cost.