Join algorithms for parallel computers for relations based on interpolation based grid file

Abstract

The recent advances in parallel and distributed processing and its applications to database operations such as join have initiated extensive research in this field. Investigations on hash based join algorithms compared to methods such as join-index join and merge-sort have given encouraging results. However, they involve a costly data partitioning phase prior to the join. This costly partitioning phase can be avoided if file structures that keep data already partitioned in the secondary storage are used. Interpolation Based Grid File (IBGF) is such a file structure. In this thesis new joint algorithms for parallel computers for relations based on IBGF are investigated. Different algorithms are used for uniform relations and nonuniform relations. The efficiencies of these algorithms based on relation and architecture, have been studied using simulation. However the comparison of different techniques is not within the scope of this work.