Analysis and benchmarking of state-of-the-art database management systems for IoT

Details

Supervisors

Abreu Araujo, Flavio

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] en sciences informatiques, à finalité spécialisée

Abstract

The increasing adoption of Internet of Things (IoT) systems, which connect sensors and devices to collect and exchange data, has led to the generation of massive amounts of time-stamped data. This growing data volume demands robust and efficient database solutions capable of handling the unique requirements of IoT workloads, such as high write frequencies and complex queries over time-series data. This thesis investigates the performance of various database systems, including relational and time-series databases, in managing IoT workloads. The analysis focuses on key metrics such as query execution times, scalability, and data retrieval efficiency under diverse scenarios. To achieve this, a custom benchmarking framework was developed using a Flask API, simulating real-world IoT environments with different data loads and access patterns. Key findings reveal that time-series databases like TimescaleDB and QuestDB exhibit superior performance for specific workloads, particularly when handling large-scale data with time-series characteristics. However, the results also highlight the limitations of object-relational mapping methods and the potential benefits of raw queries for optimizing performance. Based on the results, recommendations are provided for database selection and schema optimization. Additionally, this thesis outlines best practices for integrating database solutions into IoT systems, emphasizing the trade-offs between scalability, query performance, and system complexity. The findings aim to guide users in selecting and configuring database systems to meet the demands of modern IoT applications.