There’s an evolution occurring that blends the move toward hosted services in data centers with our always-connected devices, morphing from ‘the Cloud’ to ‘the Fog’ as our devices start to interconnect rather than to route all interactions through networks to large data centers. This more distributed model is deja vu in terms of past network configurations of thick clients (desktop PCs), but makes a great deal of sense given today’s data deluge and the increasing computing capacity and expanding storage of our mobile devices. Networks of devices battle the bandwidth issues on the edge of the network where less data is needed from the cloud due to connectivity of proximate devices and sensors.
This evolution will increasingly become important as the Internet of Things takes hold and we begin seeing more solutions from networked sensors. The cloud describes an amorphous entity that’s always present above us somewhere, whereas the fog brings that technology down to Earth with a harnessing of the terrestrial devices and sensors around us. Both the cloud and the fog are getting thicker, with repercussions on delivery of services as these networks get more complex. Perhaps the key differentiator of ‘the fog’ is the geographical distribution of devices, and how location provides an important input.
The fog certainly won’t replace the cloud, but there are distinct advantages for making more use of the computing power and storage on devices rather than relying on Internet connections and hosted Web Services for all applications. These advantages include :
The Internet just isn’t ready for the huge amounts of data that are coming from sensors. Cisco’s plan is to store some data in routers themselves, and to facilitate more localized computing. This approach is a whole new paradigm for Internet-based computing, with the addition of a great many nodes that combine intelligence as well as aid capacity, gaining a node-based understanding and intelligence about their surroundings.
The localized fog platform will achieve much faster connections that approach real-time thanks to the power of the distributed nodes that will work even when network Internet connections are lost. The local processing and filtering of incoming data locally provides a situational awareness that is most beneficial at the point of collection, pushing out just reports and the condensed understanding to others through the cloud, and uncluttering the network from a deluge of unactionable pings from sensors.
With location-aware nodes that are informed by the local sensor network, the intelligent network nodes could be the answer to meaningful augmented reality and a host of other applications. The current issues with today’s augmented reality include the need for real-time understanding and precise positioning. The speed advantages of the fog have been discussed, and robust local nodes could also provide an added input for greater location through triangulation of local signals from multiple nodes. The fog could indeed be the key to breaking down barriers that have made augmented reality more of a dream than an operational approach.
Other applications that will benefit from the fog are deployments of networked sensors such as smart grids or smart water, and intelligent transportation along roads and rail. The higher-quality signals along these networks from the distributed deployment of local nodes will present a new means for adding robustness to these networks, potentially offering closed loops dedicated to their monitoring with bi-directional flows that improve understanding of behavior without the latency issues typical of such large networks.
The new architecture of the fog is a development that fills many needs, while building upon the advancements that the cloud have provided. Geographically distributed intelligent nodes could be the key ingredient to realizing more intelligent and predictive network performance, where location is one of the key differentiators from the cloud.