The challenges of key-value stores

Abstract

The escalation of flexibility, scalability, and elasticity demands for data storage solutions has implored the quest of finding an alternative to the traditional relational databases, which in turn, lead to the popularization of NoSQL databases. This paper takes a deeper look into the key-value stores, and accentuates their strengths and weaknesses. The performance of Redis, LevelDB, and Oracle NoSQL is compared to that of the PostgreSQL database. The results affirm the inability of key-value stores to achieve comparable performance on queries that are typically performed on relational databases, such as inserting and deleting records, sorting and querying that involves several join operations. The results indicate that LevelDB outperforms Redis and Oracle NoSQL in most scenarios. The evaluation of the modeling capabilities of each database reveals additional challenges that one needs to be aware of when choosing which key-value store is best suited for their requirements.