Fog Computing Architectures: a Reference for Practitioners

TL;DR

This paper reviews the current state of Fog Computing, an extension of Cloud Computing to network edges, highlighting its importance in IoT and its market growth, with insights from standardization and open-source initiatives.

Contribution

It provides a comprehensive overview of Fog Computing architectures, standardization efforts, and commercial and open-source developments, serving as a practical reference for practitioners.

Findings

Fog Computing extends Cloud to network edges, enabling IoT applications.

The market for Fog Computing is projected to reach $18 billion by 2022.

Nearly 45% of data will be processed at the network edge by 2025.

Abstract

Soon after realizing that Cloud Computing could indeed help several industries overcome classical product-centric approaches in favor of more affordable service-oriented business models, we are witnessing the rise of a new disruptive computing paradigm, namely Fog Computing. Essentially, Fog Computing can be considered as an evolution of Cloud Computing, in the sense that the former extends the latter to the edge of the network (that is, where the connected devices -- the things -- are) without discontinuity, realizing the so-called "cloud-to-thing continuum". Since its infancy, Fog Computing has been considered as a necessity within several Internet of Things (IoT) domains (one for all: Industrial IoT) and, more generally, wherever embedded artificial intelligence and/or more advanced distributed capabilities were required. Fog Computing cannot be considered only a fancy buzzword: according to separate, authoritative analyses its global market will reach $18 billion by 2022, while nearly 45% of the world's data will be moved to the network edge by 2025. In this paper, we take stock of the situation, summarizing the most modern and mature Fog Computing initiatives from standardization, commercial, and open-source communities perspectives.