ARCHITECTURAL APPROACHES TO SELECTING A TECHNOLOGY STACK FOR IOT DATA STORAGE IN LOGISTICS MANAGEMENT TASKS
DOI:
https://doi.org/10.35546/kntu2078-4481.2025.4.3.22Keywords:
Internet of Things, IoT, Time-Series Database, logistics, InfluxDB, TimescaleDB, Redis, PostgreSQL, high-load systemsAbstract
The rapid development of the Logistics 4.0 concept and the widespread implementation of Internet of Things (IoT) technologies in supply chains have led to an exponential increase in telemetry data volumes. Modern logistics management information systems require processing data streams from GPS trackers, temperature sensors, and fuel level sensors in real-time. The specific nature of this data, which belongs to the Time-Series category, creates a significant load on storage subsystems, which traditional relational database management systems (RDBMS) often fail to handle effectively. The purpose of this paper is to determine the optimal technology stack capable of ensuring both high data ingestion rates and low latency when executing analytical queries, which is critical for the further application of machine learning methods and blockchain technology. Methodology. To achieve this goal, an experimental testbed based on Docker containerization was developed, simulating the load from a fleet of vehicles (up to 1000 trucks) with real-time data generation. A comparative performance analysis of four classes of DBMS was conducted: relational (PostgreSQL), hybrid (TimescaleDB), specialized NoSQL (InfluxDB), and In-Memory (Redis). The study involved two stages: an Ingestion Benchmark (measuring write throughput) and a Query Benchmark (measuring latency for operational and analytical queries). Findings. It was established that InfluxDB demonstrates the highest write speed (~121,800 messages/sec), outperforming PostgreSQL by almost three times. This confirms the efficiency of LSM-tree structures for write-heavy workloads. At the same time, Redis was identified as the leader for operational monitoring tasks with a response time of less than 1 ms. It was revealed that TimescaleDB, having a lower write speed (~14,500 messages/sec) due to partitioning overheads, provides a compromise between performance and SQL analytics capabilities. Scientific Novelty. The limits of applicability for hybrid SQL-based time-series systems in logistics have been experimentally defined. It is proven that for loads exceeding 15,000 events per second, specialized NoSQL solutions are required. An architectural limitation of Redis in scenarios of massive writing of heterogeneous metrics due to its single-threaded model was identified. Practical Value. Based on the obtained data, a hybrid two-tier architecture is proposed: Redis as a "hot" layer for dispatching and real-time tracking, and TimescaleDB/InfluxDB as a "cold" layer for deep analytics, ML modeling, and audit trails. This approach allows leveling the shortcomings of each individual technology.
References
Gubbi J., Buyya R., Marusic S., Palaniswami M. Internet of Things (IoT): A vision, architectural elements, and future directions. Future Generation Computer Systems. 2013. Vol. 29, No. 7. P. 1645–1660. URL: https://doi.org/10.1016/j.future.2013.01.010
Ramaswamy L., Lawson B., Gogineni S. V. Towards a quality-centric big data architecture for federated sensor services. Proceedings of the 2013 IEEE International Congress on Big Data (BigData Congress). IEEE, 2013. P. 86–93. URL: https://doi.org/10.1109/BigData.Congress.2013.21
Zaslavsky A., Perera C., Georgakopoulos D. Sensing as a service and big data. arXiv preprint, arXiv:1301.0159, 2013. URL: https://doi.org/10.48550/arXiv.1301.0159
Vongsingthong S., Smanchat S. A review of data management in Internet of Things. KKU Research Journal. 2015. Vol. 23. P. 215–240. URL: https://doi.nrct.go.th/ListDoi/listDetail?Resolve_DOI=10.14456/kkurj.2015.18
Ma M., Wang P., Chu C.-H. Data management for Internet of Things: Challenges and opportunities. Proceedings of the 2013 IEEE International Conference on Green Computing and Communications and IEEE Internet of Things. IEEE, 2013. P. 1144–1151. URL: https://doi.org/10.1109/GreenCom-iThings-CPSCom.2013.199
Zhu S. Creating a NoSQL database for the Internet of Things: Key-value store on the Sensei-Uring platform : Doctoral dissertation. Mid Sweden University, 2015. URL: https://www.diva-portal.org/smash/get/diva2:841604/FULLTEXT01.pdf
Bader A., Kopp O., Falkenthal M. Survey and comparison of open-source time series databases. Datenbanksysteme für Business, Technologie und Web (BTW 2017) – Workshopband. Bonn : Gesellschaft für Informatik, 2017. P. 249–268. URL: https://www.researchgate.net/publication/315838456_Survey_and_Comparison_of_Open_Source_Time_Series_Databases
Kiefer R. TimescaleDB vs. PostgreSQL for time-series: 20x higher inserts, 2000x faster deletes, 1.2x–14,000x faster queries. Timescale Blog, 2020. URL: https://blog.timescale.com
Fadhel M., Sekerinski E., Yao S. A comparison of time series databases for storing water quality data. Auer M. E., Tsiatsos T. (eds.) Mobile Technologies and Applications for the Internet of Things (IMCL 2018). Cham : Springer, 2019. P. 302–313. URL: https://doi.org/10.1007/978-3-030-11434-3_33
Grzesik P., Mrozek D. Comparative analysis of time series databases in the context of edge computing for low power sensor networks. Computational Science – ICCS 2020. Lecture Notes in Computer Science, Vol. 12141. Cham : Springer, 2020. P. 311–324. URL: https://doi.org/10.1007/978-3-030-50426-7_28
Dias L. B., Holanda M., Huacarpuma R. C., de Sousa Jr. R. T. NoSQL database performance tuning for IoT data: Cassandra case study. Proceedings of the 3rd International Conference on Internet of Things, Big Data and Security (IoTBDS 2018). 2018. P. 277–284. URL: https://www.scitepress.org/papers/2018/67827/67827.pdf
