Edge computing is raising as the next paradigm in IT and it aims to provide cloud-like functionalities closer to end-users. This shift has the potential to drastically improve the Quality of Experience of the users as well as to enable unprecedented use-cases (e.g., Connected Factories, Connected Cars, Augmented Reality, etc.).
Edge infrastructure is expected to be very distributed and heterogeneous comprising a wide set of devices, spanning from COTS servers to Single Board Computers, Industrial PC, IoT gateway, and sensors. This presents a significant change from Cloud computing where the infrastructure is mainly composed of centralised and homogenous data centres.
As of today, the main approach to Edge computing has been to depart from the existing Cloud technologies and to stretch and adapt them. While this approach is feasible for a first realisation of Edge computing (i.e., regional or metropolitan data centres), an ever-increasing level of decentralisation may pose several challenges. Can we keep stretching Cloud software forever? Does cloud software really fit with the Edge dynamics?
In this presentation we aim to (1) explain the differences between Cloud and Edge infrastructure, (2) expose the challenges on the management of an Edge infrastructure in contrast to a Cloud data centre, (3) present how current cloud solutions perform and scale in an edge environment and (4) illustrate the design choices of the Eclipse fog05 project (https://fog05.io - https://github.com/eclipse-fog05) for coping with an Edge environment. Eclipse fog05 is developed within the Eclipse Edge Native Working Group.