Architectural Backbone Evaluation for Data Stream Processing within the WINNER DataLab - A Project Focused Point of View

Smart platforms for the integration of sensor and actor networks require collecting, analysing and evaluating data. Our research project WINNER aims to integrate systems which cover electromobility, energy consumption within residential areas, local energy production, e.g., with photovoltaic systems, and storages for local smart grids. WINNER wants to use such a platform to optimise the energy consumption. Every actor within a residential area has to be considered, and integration into a centralised data stream process is necessary. As a non-hard real-time system, the platform has to solve enterprise application integration problems, looking at complex event processing and knowledge discovery in data. This paper targets to analyse possible architectural backbone technologies. Out of a wide range of potential technologies Node-RED, Apache NiFi and Apache Camel are selected and compared. Those technologies with a diverse field of application are used to implement a comparable test setup. Furthermore, they are analysed through their characteristics of processing, execution, usability and simplicity. As measured, Node-RED, Apache Camel and Apache NiFi indicate stable and fast message processing, especially in the case of raising message throughput. Node-RED surprises with constant memory and CPU loads and seems to be an exciting option in rapid prototyping.

Vollständige Veröffentlichung