A Quality Attribute-based Evaluation of Time-series Databases for Edge-centric Architectures.

Edge computing is unlocking the potential of Industrial Internet of Things (IIoT) to deliver business value and impact. Reliable data processing and guarantee of services within the IIoT sphere mandates the use of local storage within the device. The choice of database is also crucial for the performance of applications deployed on this resource-constrained hardware. Yet, there has not been any defined quality-attribute based approach for selecting databases for edge deployments. Much of the work done in the past focus on evaluating databases for applications deployed on high computing servers and cloud-first architectures. In this study, we address this gap by identifying software quality attributes for evaluating time-series databases for edge-centric deployments. We then conduct detailed experiments on four databases -- InfluxDB, MariaDB, Redis and TimescaleDB using these defined attributes. The paper concludes by highlighting how the evaluation methodology has aided in the selection of databases for use-cases from two different domains decentralized traffic management and condition monitoring of distributed transformers.