Fog Networking/Computing

The Future Is Foggy – HR Report

As nebulous as its name suggests, fog computing looks set to attract a lot of attention this year as one of several technologies that could support new business models and applications in the connectivity game.

For mobile operators, fog computing is unlikely to be a short-term revenue opportunity, according to a new study from Heavy Reading , the market-research arm of Light Reading. And while more progressive operators could emerge as important "enablers" of fog computing, it will be up to other players to take the lead in developing use cases, says Gabriel Brown, a senior analyst with Heavy Reading who co-authored the report.

In large part, it seems, that's because of the considerable uncertainty that surrounds fog computing -- which still evades any straightforward definition -- and its commercial proposition.

Many observers see parallels with the concept of edge computing, whereby IT resources are moved out of centralized facilities and deployed at basestations or basestation aggregation points, much closer to the end user. Proponents believe edge computing could help to reduce operating costs and create new service opportunities.

Want to know more about the Internet of Things? Check out our dedicated IoT content channel here on Light Reading.

Fog computing appears to go a step further, however. Instead of putting IT resources at an edge network facility, companies could introduce them into end-user equipment. While he does not use the fog computing expression, Peter Levine, a venture capitalist with Andreessen Horowitz, says this kind of move could support applications such as real-time image processing by car-mounted cameras. Indeed, it is easy to imagine fog computing playing a key role in the broader Internet of Things (IoT) market.

While many telcos may regard fog computing as a part of the IoT, some of the concept's chief supporters insist that it covers a multitude of deployment scenarios. "While MEC [mobile edge computing] is focused on the RAN [radio access network], fog includes access, edge and the devices themselves," said Steve Vandris, a board member for the OpenFog Consortium, during a conference in Germany last September. "It distributes computation across all these nodes." With a goal of creating an open fog computing architecture to support interoperability, the OpenFog Consortium counts ARM Ltd. , Cisco Systems Inc. (Nasdaq: CSCO), Dell, Intel Corp. (Nasdaq: INTC) and Microsoft Corp. (Nasdaq: MSFT) as founding members. (See Will ETSI Lose Its Edge as Fog Rolls In?)

Table 1: MEC Vs OpenFog, According to OpenFog

MEC OpenFog
Focus on RAN [radio access network] cellular network access Access node/network connection agnostic, fully supporting cellular, unlicensed radio, wireline and optical interconnects
Focus on the RAN edge, adds computing in a single layer of nodes in the RAN Horizontal architecture for distributed computing, storage and networking services across cloud-to-thing continuum
Focus on mobile service providers, uses mobile carrier models Physical and logical hierarchy of fog nodes with north-south, east-west and diagonal connectivity: covers edge but also access and things in IoT intermediate layers between edge and cloud
MEC focuses on moving applications into the RAN, which requires many of the functions the OpenFog RA [reference architecuture] provides (e.g., distributing software/apps to edge nodes, orchestrating resources to support applications, managing the lifecycle of the software/apps distributed to the edge, and securing these distributed systems) Multi-tenant service models, virtualization, orchestration, management, uses enterprise and web-scale models
MEC can right-size the OpenFog RA to work inside RANs and for its newly expanded multi-access scope to support the fog-related functions MEC requires OpenFog is an SIG [special interest group] that partners with multiple SDOs [standards development organizations] to cover fog computing standardization needs across multiple disciplines, can leverage MEC technologies and APIs [application programming interfaces]
Source: OpenFog.

As grandiose as this sounds, it may do little to entice mobile operators to be answerable to returns-hungry investors. What's more, if intelligence moves from the network into devices, then players other than telcos -- including manufacturers and software developers -- may stand to be the main beneficiaries of fog computing.

Operators' wariness about the relatively new concept of fog computing is perhaps unsurprising given the lack of major commitments to edge computing, despite years of publicity. Rummaging through old magazine clippings this week, Light Reading's own Carol Wilson came across stories from the early 90s about "distributing intelligence to the edge of the network." Yet most respondents to a recent Light Reading survey were unconvinced that edge computing is as strategically important as other next-generation technologies. (See The More Things Change..., Edge Computing Not Yet a Telco Priority – Poll and graphic below.)

Source: Light Reading.
Source: Light Reading.

This does not mean operators are going to be oblivious to what happens in fog computing, though. "The positive spin is that there will be a lot of innovation at the edge over the next ten years," says Brown. "Operators will be supporting that development, providing connectivity and hosting and being enablers, but it's not going to be their role to come up with new revenue models."

Instead, network architecture will evolve accordingly as emerging technologies, including 5G, force telcos to adapt to new service requirements. "Over the longer term, we'll see progressive developments in network architecture that will mesh nicely with some of the emerging edge applications," says Brown.

Ultimately, everything is going to get a lot foggier.

— Iain Morris, Circle me on Google+ Follow me on TwitterVisit my LinkedIn profile, News Editor, Light Reading

mendyk 1/6/2017 | 11:23:06 AM
Re: Telcos rejecting edge intelligence Seven -- good points, and this is exactly why "telcos" need to at least understand fog. Rather than sell off COs and PoPs, these strategic facilities can be transformed (OK, I hate that word as much as anyone else, but still) into a central nervous system for true next-gen networks.
brooks7 1/5/2017 | 4:59:40 PM
Re: Telcos rejecting edge intelligence Dennis....and holy cow a Dave Burstein sighting...Hi Dave..its Jim Sackman...

The telcos lost their R&D when AT&T was broken up in 1984.  The old way this worked was...

Bell Labs did Research and Architecture Design

Bellcore (aka Telcordia) did Specification and Testing

Western Electric (aka Lucent) made the Products

Baby Bells (aka RBOCs etc) deployed the Products.


In those days, people who wanted to sell to the telcos built products that matched the architectures and specifications.  But no carrier had R&D to determine architecture and design of how things were going to be built.  We have ended up in this hodge podge of things where we have pretty much stuck to the architecture of the network pre-1984.  We have replaced many of the elements but they essentially have the same spots in the network that we have always had.  Yes, there is not 100% overlaps but think about how router deployment mirrors Class 4/Class 5 deployment for example.

Nobody has said anything like...Hey if we ignored our current network, How would we partition the jobs between equipment types?  That Bell Labs job is just gone.  We have had things like the DSL Forum or the Broadband Forum or the MEF try to handle pieces of it at a time and generally within a specific tier or function within the network.

Think about it.  When we were doing POTS, I could walk up and test a POTS card for voice transmission capability.  I could do this independent of the entire network.  Why?  Because a group in Bell Labs decided that this is the loss plan that a POTS card should have to meet a MOS score when the entire network was in place.  They had lots of choices but picked one that worked.  NOBODY is doing that kind of work today.  

So why not FOG or Edge or Node to Node....how the heck does that work?  And that is why many of these new technologies (I am looking at you SDN and NFV) are struggling.  If I were building an NFV based network, I would build essentially massively connected data centers NOT the way our network is built today.


mendyk 1/5/2017 | 11:30:43 AM
Re: Telcos rejecting edge intelligence At this point, telcos don't seem to have a grip on fog, either -- which is not surprising given the sad state of their R&D efforts over the past decade. No shocker that the Web crowd looks like it's going to dictate the future of networking.
daveburstein 1/4/2017 | 7:06:24 PM
Re: Telcos rejecting edge intelligence Mendy

I didn't say the telcos were right, just reporting what they tell me they are doing. 

I don't understand FOG well enough to have an opinion.

mendyk 1/4/2017 | 6:11:41 PM
Re: Telcos rejecting edge intelligence Dave -- The telcos as we've known them are being steadily marginalized when it comes to determining how networks are built and run. The underlying concepts for fog make a lot of sense in the context of where networks are heading (IoT, 5G), whether or not telcos are happy about that. It's not surprising that telcos aren't going to move quickly on this -- and, clearly, this isn't something that will happen immediately. But I wonder if the attitude would be different if telcos hadn't turned away from R&D that focused on long-term possibilities.
daveburstein 1/4/2017 | 5:06:39 PM
Telcos rejecting edge intelligence Iain, Carol

FOG is interesting, and essential if 5G wants 1 ms latency. But putting intelligence at the edge in 5G would add $billions to deployment costs. 

AT&T today specifically said 5 ms, not 1 ms. bit.ly/ATTGIG2016

Verizon and Softbank have confirmed Cloud Ran rather than edge intelligence. Both imply no FOG intelligence at the edge. Neither Ericsson nor Nokia could point me to any telco committed to 1 ms & edge intelligence.

It doesn't look like this will fly at telcos, unfortunately. 
Sign In