Cardiff University, UK
Prof. Omer Rana
Edge Clouds: Architectures and Deployment Models
An architecture for undertaking data processing across multiple layers of a distributed computing infrastructure is introduced, composed of edge devices (making use of Internet-of-Things (IoT) based protocols), intermediate gateway nodes and large scale data centres. In this way, data processing that is intended to be carried out in the data centre can be pushed to the edges of the network – enabling more efficient use of data centre and in-network resources. Such an approach also suggests the need to re-think existing cloud computing deployment architectures, requiring a “traditional” data centre-based cloud system to stretch to the network edge. We suggest the need for specialist data analysis and management algorithms that are resource-aware, and are able to split computation across these different layers. A coordination mechanism is proposed that is able to combine different types of data processing capability, such as in-transit and in-situ. An application scenario is used to illustrate the concepts, subsequently evaluated through a multi-site deployment.