Big data analytics using SQL: Quo vadis?

Abstract

Big Data processing and analytics are dominated by tools other than SQL based relational databases, which have lost their numero uno status. In a world deluged by data, the general perception is that SQL databases play a marginal role even for analyzing structured Big Data despite their inherent strengths in processing such data. Focusing on the most important aspect of Big Data processing, namely analytics for data mining, we examine the validity of this perception through a study of competing technologies, published results on SQL implementations of data mining algorithms, the impact of cloud platforms and the raging debate on SQL vs NoSQL vs NewSQL. Contrary to the general belief, it appears that SQL databases in their parallel, columnar deployments on cloud with UDF support do solve some, if not all, Big Data problems and are not likely to become dinosaurs in Big Data era.