A qualitative comparison of NoSQL data stores

Abstract

Due to the proliferation of big data with large volume, velocity, complexity, and distribution among remote servers, it became obvious that traditional relational databases are unsuitable for meeting the requirements of such data. This led to the emergence of a novel technology among organizations and business enterprises; NoSQL datastores. Today such datastores have become popular alternatives to traditional relational databases, since their schema-less data models can manipulate and handle a huge amount of structured, semistructured and unstructured data, with high speed and immense distribution. Those data stores are of four basic types, and numerous instances have been developed under each type. This implies the need to understand the differences among them and how to select the most suitable one for any given data. Unfortunately, research efforts in the literature either consider differences from a theoretical point of view (without real use cases), or address performance issues such as speed and storage, which is insufficient to give researchers deep insight into the mapping of a given data structure to a given NoSQL datastore type. Hence, this paper provides a qualitative comparison among three popular datastores of different types (Redis, Neo4j, and MongoDB) using a real use case of each type, translated to the others. It thus highlights the inherent differences among them, and hence what data structures each of them suits most.