Scalability and Performance Evaluation of NewSQL and Relational Databases: A Comparative Benchmark Study
Maryam Abbasi,Rui Santos,J. Silva,Pedro Martins,Paulo Váz,Bruno Nascimento
DOI: https://doi.org/10.1109/SmartTechCon57526.2023.10391567
2023-08-18
Abstract:The purpose of this article is to test and compare the scalability and performance of NewSQL databases (CockroachDB and SingleStoreDB) versus standard relational databases (MySQL and PostgreSQL). The dataset chosen for this benchmark was an TPC-H generated database, making it a suitable dataset for testing database scalability and efficiency in handling complex queries and big amounts of data.To execute the benchmarking, we will construct assessment criteria based on 3 main categories: Process, Performance and Strategy. In this way, it will be possible to evaluate the behavior of each solution, in different contexts and how they use the respective functionalities to obtain the best result. A Benchmarking tool is essential for obtaining sustainable, reliable, and community-accepted indicators, as it uses standard metrics to assess and classify behavior and performance. Based on these paradigms and the mechanism used for data generation, the benchmark tool chosen for this paper is the TPC-H Benchmark, which explores the data organization and complexity.The results of this benchmarking will give significant insights into the strengths and drawbacks of NewSQL and RDBMS databases, as well as whether the database is better suited for certain use cases. Furthermore, using a dataset like TPC-H will deliver more accurate and useful results, allowing developers and database administrators to make more educated database selection selections. Another point to check is the behavior of technologies in the face of a large volume of data, which is something natural and increasingly common nowadays.
Computer Science