This WSJ article, in May 2014, was most certainly not advocating the end of the Cloud, but instead was highlighting the challenge of the ever-growing numbers of devices of all types connecting, and interacting, in a wholly different manner. The technology industry has come to realize that the Internet of Things is not just about sensors, it’s about the shift to millions and millions of devices interacting in new ways around events.
We are all too frequently aware of the poor responses caused by insufficient bandwidth, now bandwidth use, (over use), is multiplied by two dimensions; More traffic to centralized cloud based Data Centers for traditional heavy duty IT computational tasks; Plus and ever increasing number of small devices using connected service models.
The centralization of services into a lower number of ever larger cloud data centers might be desirable for the cost/scale needs of IT as we know it, but we need to recognize the rise of Internet of Things, IoT, also changes the game in terms of where, how and what resources are made available. Even more important to event centric nature of these new devices and their services is response times!
In the blog entitled; ‘We know things are changing, but what is making everything change? A summary of change factors in technology and business in 2015 – Part 1’; the root cause seen to be a global move in society, business and technology towards ever increasing de-centralization. Massive Data Center Clouds are doing an admirable job of delivering the centralized tasks of IT in a more efficient, effective manner, but it has become clear something different is required for mass numbers of small active devices. A different technology infrastructure and charging structure is needed to empower new de-centralization business activities; an infrastructure that unites Enterprises with their customers, and their increasing demands for personalization.
Fog Computing is an architecture that uses a collaborative multitude of end-user clients, or near-user, edge devices to carry out data storage, communication, control, and management tasks locally to improve responsiveness. Fog Computing is particularly suited to the kind of localized lightweight interactions around sensing and responding to an event that IoT, in its many forms, introduces.
This is an edited version of the Wikipedia definition of Fog Computing that gives a great deal more depth and correctly notes that the term and concept was first defined by Prof. Salvatore J. Stolfo a Professor of Computing at Columbia University, New York. However the name of Fog Computing is usually linked to Cisco following the better known publication of a white paper under this name by four Cisco Research staff in 2012, today the term Cisco Fog Computing is trademarked leading to a reluctance by some other technology vendors in using the term.
Though there may be reservations over the name Fog Computing due to its link to Cisco Marketing what is not in doubt is the almost total acceptance within the industry for the concept. Edge Clouds is a possible alternative name and at the end of this blog a list of papers on the topic, including some by major Technology, venders is included as a more detailed update.
Edge Computing models are not new, as in every iteration of technology from mainframe onwards has faced the challenge of the amount of connectivity bandwidth available at the time, and the resulting latency affecting response times. However IoT Devices do bring a different dimension to this challenge being less about background computational transaction and more about visible on screen interactions for users. The increased dimension of user and event time significance warrants a different approach to what is meant by, and required as’cloud’ architecture. A more detailed explanation of the concepts of Fog Computing, (still the most popular term), or Edge Cloud Computing a more recent term, both defining the use of small localized clouds can be found in a previous blog in this series published in January 2015 entitled IOT: From the Intranet of Things to the Internet of Everything – Introducing the required solution architecture So why come back to the topic again only nine months later?
The level of engagement and activity related to developing Fog Computing across the entire technology sector has ramped up considerably as it becomes increasingly accepted that an edge-based architecture is an essential part of a mass Device or IOT environment. At the same time its become clear that the distributed ‘interactive’ IoT environment using Fog Computing supports still further innovation in technology capabilities around Data models and event processing.
To grasp the activity level and understand how Fog Computing and Edge Clouds is being developed there is at the end of this blog a list of links to papers and sites. Understanding the basics of the infrastructure architecture leads naturally to grasping two very important consequences that following blogs will outline;
De-centralized Business models require a de-centralized open infrastructure on which to develop their capabilities and innovative competitive propositions; these business models require access to, interaction with, and use of data in a wholly different manner to the conventions of assembling and analyzing data currently in use.
Some resources for Fog Computing or Edge Clouds information;
An HP white paper defining their views on Fog Computing
IBM explains their views on Fog Computing
Microsoft defining how its offerings match different Cloud/Fog requirements
SAP team with Cisco to explain how they see the architecture developing
Central list of resources at Cisco for all Fog Computing matters
Ericsson provide an IOT telecoms view of Cloud Edge Computing
The case for Smart Gateways to Operate Edge Cloud Computing architecture
A good briefing on Fog Computing by a startup product company
A new site campaigning to become the news site for Fog Computing