Edge Computing or Fog Computing?

Though "fog computing" is not a term I use much, those who do sometimes have to explain how "fog" is different from "edge." 

It is fair to note that "fog computing" is a proprietary term used by Cisco to describe cloud computing extended close to the enterprise edge. 

In that context,  fog computing is best understand as a broad term for many types of edge computing--but not on the device--that happens close to where data is generated, and not at a remote data center. 

Fog computing sometimes is described as a framework for edge computing, excluding computing on a device itself. Others might say there are device, premises and infrastructure edge (including aggregation nodes) components to edge computing.

For me, the important new concept is not computing directly on a device or at a remote data center, but the emergence of a new ability to compute close to the source of data generation. And the key new value is ultra-low latency computing response.

About the only sort of computing fog does not include is traditional computing at a hyperscale data center or some other remotely-located data center.

For me, that works best. In cases where a former central office becomes an “edge computing center,” that is computing within the fog architecture. But so is edge computing at some other intermediate location between a single end user device or appliance and the place where the wide area network is encountered.

The phrase local area network almost always occurs when “fog” is defined. And that is where some confusion can occur. In a traditional sense, the local area network is a privately-owned, indoor or campus-wide network separate from the public “access” network.

But in a more general sense, some might refer to the “local” area network as some intermediate point in the access, feeder or distribution portions of a public network (downstream of a central office, for example).

Some might say the fog concept involves computing close to where data is generated, but not on the sensor or device itself, or remotely across the wide area network. Others might not use the term at all, and simply categorize computing as on the device, on the premises or at some other location in the local access network.

Computing at the base of a cell tower is one analogy for infrastructure edge computing, which implies something about the owner of the facilities (a connectivity provider).