Join optimization in distributed databases on broadcast network

Abstract

Problem of joining two fragmented relations on a broadcast network is dealt with. Replicated copies of a fragment are considered to improve parallelism, transmission cost, and reliability. Our approach for optimizing a two-way join consists of the following three phases: (1) Construct join graph, (2) Minimize transmission cost, and (3) Final processing. In the first phase, the empty fragment joins that cannot have matching tuples are determined by considering the join predicate and qualifications of fragments. The problem of determining empty fragment joins is transformed into a rectangle intersection problem, and an efficient algorithm is proposed for the rectangle intersection problem. In the second phase, the fragments and join results that is to be transmitted are determined. In the last phase, a pipelined, reliable join processing is performed.