Security and performance comparison of NoSQL database systems

The advent of Big Data has led the development of several key technologies and platforms to answer the ever-growing nowadays data. NoSQL was one of these technologies. Although NoSQL databases began to develop before the appearance of big data, the adoption of NoSQL databases did not occur until big data created the need. However, many versions of NoSQL databases appeared each with different architectural design, its query language, and its own set of solutions for scalability, compression, security, clustering, which render it to become difficult to decide which NoSQL database provides the solution to a given problem or issue. Therefore, studies should be dedicated to testing these databases concerning several factors such as query processing speed, security of data, and its readiness to be used in the scalable environment. In this study, we propose an investigation in two-folds: first, examining the performance of two NoSQL databases, MongoDB (3.6.3) and Cassandra (3.11.1) in single node and multi-node (cluster) configurations using the recent version of Yahoo Cloud Serving Benchmark (YCSB-0.12.0). A testing environment is set up for each workload, and the responses for each database management system used in the study are examined for each workload. This study will help IT decision makers to determine which database is better according to its deployment requirements. The second fold of this study is to investigate the security of both databases. The first step is a comparative study of both databases' security features according to ten selected features from the literature. The second step of our security investigation is data encryption overhead. Overhead is the time spent by the database engine to encrypt all incoming data streams plus the time spent to decrypt the data to answer queries. The results of this study have shown that the performance of each system differs due to the differences in its respective data storing mechanisms. The results have also revealed that the performance of MongoDB is better than Cassandra in a single node test, whereas in proposed multi-node test Cassandra performed better in one workload and took the lead in two other workloads when testing bigger record size. Security comparative investigation has shown that both databases have improved significantly concerning previous studies. However, Cassandra took the lead as it has better supported and the implementation of the selected security features is presented in this study. Lastly, MongoDB encryption performance regarding runtime and throughput was between nearly 2x and 2.5x faster than Cassandra which makes it more suitable to be used in the environments when encryption is a requirement.