SQL and NoSQL database comparison: from performance perspective in supporting semi-structured data

Abstract

In this digital era, social media web applications have churned out huge amount of unstructured data each day. These social media data may be processed into meaningful data through text analytics. With the rapid growth of the volume of unstructured data produced daily, NoSQL database is increasingly popular that it has become the chosen database to store data. However, little research is done on the comparison of SQL and NoSQL in terms of indexing, performance tuning, and amount of records supported. This paper aims to provide a thorough comparative evaluation of MongoDB and MySQL, a tool for SQL and NoSQL databases, respectively, in terms of their performance in populating and retrieving big data after performance tuning. The findings presented in this paper give a new insight from the aspect of how these databases support semi-structured social media data by considering the options of performance tuning. The methodology for this research consists of four performance measurements, namely, insert, select, update, and delete up to 1 million Twitter data stored, to evaluate SQL and NoSQL databases. Our result findings indicate that MongoDB does perform faster for all the four operations. However, there are more performance tuning options provided by MySQL for more flexible performance optimization.