Date of Award

1998

Degree Type

Thesis

Degree Name

Bachelor of Science Honours

Faculty

Faculty of Science, Technology and Engineering

First Advisor

Jean Hall

Abstract

Relational database systems use join queries to retrieve data from two relations. Several join methods can be used to execute these queries. This study investigated the effect of varying join selectivity factors on the performance of the join methods. Experiments using the ORACLE environment were set up to measure the performance of three join methods: nested loop join, sort merge join and hash join. The performance was measured in terms of total elapsed time, CPU time and the number of I/O reads. The study found that the hash join performs better than the nested loop and the sort merge under all varying conditions. The nested loop competes with the hash join at low join selectivity factor. The results also showed that the sort merge join method performs better than the nested loop when a predicate is applied to the inner table.

Share

 
COinS