Comparative Analysis of Three IoT Data Storage System Architectures on AWS Cloud

Internet of Things (IoT) devices generate large amounts of data, creating the challenge of designing efficient IoT cloud storage solutions. This study focuses on an IoT application managing air quality measurement data, which requires frequent retrieval of recent data for near-real-time monitoring and access to long segments of time-series data. The study performs a comparative analysis of three IoT storage system architecture options on Amazon Web Services (AWS) cloud. In contrast to existing literature, the options take into account data access and retrieval patterns specific to the use cases of the IoT application under study, where only data produced by a particular device is accessed at a time. The first option utilizes a NoSQL database, the second employs data streams and an object storage, and the third uses a time-series database. Each option was tested by performing retrievals of different lengths of time segments, focusing on retrieval times and query capabilities. Results show that a time-series database is suitable for applications benefiting from query options or time-series functions, such as interpolation. A Key-Value NoSQL database is suitable for applications retrieving short time segments without extensive queries. Conversely, an object storage excels in storing and retrieving large volumes of raw data, while data streams are unsuitable as a direct source of data for near-real-time applications. The study provides a framework for comparing alternative architectures to analyze their suitability, guiding decisions in selecting the most appropriate storage solution based on specific IoT application requirements.