GLT: Edge gateway ELT for data-driven intelligence placement.

In this paper, we introduce the notion of Data Lagoons at the Edge, for the dynamic deployment and migration of analytics tasks at runtime. We present a novel, modular architecture that decouples data ingestion from data processing at the Edge, facilitating the rapid instantiation of data processing components or their offloading to the cloud. To this end, we analyze the requirements for the use cases of Internet Of Things (IoT) Platforms and Multimedia Content Distribution Applications and we present for the former a prototype architecture, based on open source components. Our approach can support the realization of reasoning mechanisms for trade-off analysis and adaptive task relocation, based on real-time monitored measures such as delay and throughput.