A Containerized Edge Cloud Architecture for Data Stream Processing

Internet of Things (IoT) devices produce large volumes of data, which creates challenges for the supporting, often centralised cloud infrastructure that needs to process and store the data. We consider here an alternative, more centralised approach, based on the edge cloud computing model. Here, filtering and processing of data happens locally before transferring it to a central cloud infrastructure. In our work, we use a low-power and low-cost cluster of single board computers (SBC) to apply common models and technologies from the big data domain. The benefit is reducing the volume of data that is transferred. We implement the system using a cluster of Raspberry Pis and Docker to containerize and deploy an Apache Hadoop and Apache Spark data streaming processing cluster. We evaluate the performance, but of trust support of the system, showing that by using containerization increased fault tolerance and ease of maintenance can be achieved. The analysis of the performance takes into account the resource usage of the proposed solution with regards to the constraints imposed by the devices. Our trust management solution relies on blockchain technologies.