Hash-semijoin: a new technique for minimizing distributed query time

Abstract

Traditional semijoin and semijoin with multiple hash functions are two important methods to reduce the data transmission cost in distributed DBMS. However, these two methods are sometimes inefficient, due to their complexity and lack of flexibility. To remedy these problems, the authors propose a new technique, named hash-semijoin. They identify the situations in which it performs better than traditional semijoin and semijoin with multiple hash functions. The optimal hash functions used in hash-semijoins are also studied.<>