Performance Analysis of NoSQL Database Technologies for AI-Driven Decision Support Systems in Cloud-Based Architectures
Uttam Kotadiya, Amandeep Singh Arora, Thulasiram Yachamaneni · International Journal of Emerging Research in Engineering and Technology · 2022
With the emergence of increasingly intricate, data-driven platforms for Decision Support Systems (DSSs), as Artificial Intelligence (AI) applications increasingly are applied to a wide variety of industries, related to decision making, there is a need to carry out deeper analysis of the concept of decision support system and to assist in making sound decisions. The systems demand good, scalable and effective data management systems. The volume, variety, and velocity of data created in cloud-based AI platforms pose a challenge to traditional Relational Database Systems (RDBMSs). NoSQL databases have become an attractive alternative due to their schema-less approach, horizontal scalability, and high-performance data manipulation capabilities. In this paper, a detailed performance study of the leading NoSQL databases has been developed: MongoDB, Cassandra, Redis, and Couchbase from the perspective of decision support systems based on AI in cloud structures. We analyse all the databases using a set of parameters, including read/write latency, throughput, scalability, fault tolerance, consistency, and ease of integration with AI frameworks. Proposed methodology comprises the implementation of each of the NoSQL technologies under a simulation cloud environment with the help of benchmarking devices like YCSB (Yahoo! Cloud Serving Benchmark) and applications designed after the instructions. In addition, we suggest a decision matrix and multi-criteria analysis framework to guide architects in selecting the appropriate databases in accordance with specific AI-based DSS requirements. The findings have shown that MongoDB performed better in terms of flexibility and integration, Cassandra in write-intensive workloads and fault tolerance, Redis in low-latency operations, and Couchbase in balanced operational performance. The results of this research provide important references to the recommendations concerning the appropriateness of NoSQL solutions that can be used to improve AI-based DSS in distributed cloud environments. It educates both the researchers and practitioners on trade-offs and the best deployment practices