Comparison of full-text search performance in relational and non-relational database systems

Abstract

In this work, we aim to compare and report the performance between relational and non-relational database systems in text retrieval, and more specifically in full-text search. This comparative study is focused on the characteristics of some of the most popular database management systems, regarding the overall performance when querying text documents. We also pose and highlight the features, the technical specifications and the differences of the database systems, through several comparative tests and theoretical exploration. For the evaluation of the performance in full-text search, both relational and nonrelational database systems have been selected to be compared in this study. Our research comprises of studying PostgreSQL on the one hand, as a representative system from the relational database systems family, which supports a wide variety of full-text search capabilities and advanced features, while on the other hand, we study non-relational databases like MongoDB, Apache Solr and Elasticsearch. Our findings indicate that non-relational and text search database systems provide a trustworthy alternative in full-text search, regardless the amount of data they operate with, where on the other hand, the relational models are sensitive to their parameters as it can operate well under specific conditions.