Cloud Computing at the Edges

Abstract

Currently, most cloud computing deployments are generally supported through the use of large scale data centres. There is a common perception that by developing scalable computation, storage, network, and by energy-acquisition at preferential prices, data centres are able to provide more efficient, reliable and cost effective hosting environments for user applications. However, although the network capacity within and in the proximity of such a data centre may be high – the connectivity of a user to their first hop network may not be. Understanding how a distributed cloud can be provisioned, enabling capability to be made available “closer” to a user (geographically or based on network metrics, such as number of hops or latency), remains an important challenge – aiming to provide the same benefits as a centralised cloud, but with better Quality of Service for mobile users. With increasing proliferation of mobile devices and sensor-based deployments, understanding how data from such devices can be processed in closer proximity to the device (ranging from capability directly available on the device or through first-hop network nodes from the device) also forms an important requirement of such distributed clouds. We review a number of technologies that could be useful enablers of distributed clouds – outlining common themes across them and identifying potential business models.