Service Providers Will Adapt to IoT by Embracing Edge-Fog Architectures Featured

Service Providers Will Adapt to IoT by Embracing Edge-Fog Architectures Image Credit:

The definition of a service provider is expanding. Today’s service providers include not only traditional telcos but also mobile network operators, cable MSOs and cloud platform operators. And as service providers expand, so does the growth of IoT data on their networks. In fact, it’s increasing at staggering rates.

You’ll find IoT on smartphones, smart homes, voice assistants, medical implants, autonomous cars, smart cities, air-to-sea, public safety, and on and on.This represents a huge business opportunity for customers of service providers, across a wide range of verticals.

But IoT data is a different animal compared to IT data. For example, it needs to be constantly available, tracked and analyzed. IoT data is often time or latency sensitive. Additionally, many IoT services and applications are mission or life-critical. IoT customers are very demanding, and for a good reason.

As more and more IoT data and applications are pushed across networks, there is mounting pressure on the service providers. Fortunately, they are responding by upping their game. By evolving with these trends, service providers will usher in the next era of digital automation. They will improve the every-day efficiencies of industries and workforces across verticals, economies and our society.

One critical proficiency that will help service providers deliver on the promise of the IoT will be their ability to embrace edge and fog architectures in their network roadmaps.

The value of Edge-Fog

Let’s investigate what edge-fog capabilities can do for IoT service providers. Edge-fog moves the computational, storage and networking capabilities of the cloud and traditional data networks much deeper into IoT networks. That means these capabilities are logically and physically closer to - or even residing on - the sensors, actuators and smart endpoints in IoT systems.

This enables advantages that are difficult, if not impossible, to achieve without edge-fog networks. For example, in the most ideal traditional network environment, the round-trip latency from a sensor through a cloud-based application and back to an actuator may take hundreds of milliseconds. That’s much too long for many classes of critical IoT applications.

Edge-fog deployments can close that loop in less than a millisecond. Carrying the full bandwidth of a data-rich sensor stream (a UHD camera feed, for example) all the way back to the cloud can eat up massive amounts of network bandwidth. This causes unnecessary and unacceptable delays and network overloads, and it’s often cost prohibitive.

Considering the rapid growth of near real-time applications rising from the unrelenting growth of IoT-connected devices, the only way to ingest, process and act on mission or business-critical data is at or near the source. Importantly, edge-fog nodes can run local analytics algorithms that can make immediate (near real-time) decisions, perform serious data reduction and greatly improve network efficiency.

For IoT applications that need bullet-proof security, edge-fog nodes can perform local encryption much stronger than many simple IoT endpoints are capable of supporting. The reduction of attack vectors, strong control over cryptography algorithms and keys, and the ability to recognize and quickly address threats makes local edge-fog nodes safer.

Less transport of unencrypted or lightly encrypted traffic across networks means less exposure to risk. For systems requiring high reliability, local edge-fog processors have fewer failure modes caused by cable cuts, wireless interference, cloud server outages or distributed denial of service attacks. This gives them a better shot at achieving the Five 9s availability many critical IoT applications require.

How to Get Started

So, how can service providers enter the edge-fog market to achieve these advantages for their IoT customers? In general, building out a network of physical resources including remotely located edge-fog nodes will probably be required. For the operators who already have access networks (wired, radio, cable or fiber), this means identifying locations in their outside distribution plant where all the elements needed to install edge-fog nodes exist. These elements include:

Power, i.e.,several hundred to several thousand watts, ideally with a backup

Good network connectivity, i.e., copper, fiber or radio backhaul to a fast Internet peering point

Good environment, i.e.,cooling, physical protection and access for maintenance; and

Proximity to likely IoT end users’ networks of things

Basically, the idea is to take a device that was already out there supporting the access network (such as a Wi-Fi Access Point, DSLAM, DOCSIS fiber node, or 4G cell site) and add a significant compute/storage server to it. This server provides the local edge-fog functionality and could be pretty powerful. For example, it could house dozens of CPU cores, terabytes of storage and even local computation accelerators like Graphics Processing Units (GPUs) for images or Tensor Processing Units (TPUs) to accelerate machine learning.

The edge-fog architecture will be multi-tenant, sharing resources. Service providers will have multiple independent subscribers that need to share the fog-edge nodes, but are protected from resource inequities or data breaches from other subscribers in the same way cloud servers protect their users from each other.

Significantly, existing cloud operators may have a slightly different view of what it means to begin providing edge-fog services.They will partition their existing cloud offers into increasingly fine-grained micro data centers, which are then interconnected with a network. This could be either a new network that the cloud operators construct, or use of bandwidth they lease from more traditional network operators.

Those local data centers can potentially be installed in lots of places, such as distribution warehouses or brick-and-mortar retail locations a cloud provider may own or lease space in(i.e. libraries, schools, water towers, traffic signal cabinets).

The First-Mover Advantage

Basically, to provide the advantages of edge-fog networks described here, service providers will need to depart from the model of a handful of huge data centers that serve an entire continent (sometimes with 1,000 miles of spacing) and move to thousands of even millions of small data centers, on perhaps the order of 10 miles of spacing.

The actual edge-fog nodes could be pretty small and innocuous. A good model would be something about the size of a shoe box, microwave oven, or perhaps a small refrigerator. A device this size can be tucked into an existing rack space, a closet or a corner of a room, and can run largely unattended for long periods.

Service providers who deploy edge-fog nodes for IoT can greatly magnify their existing offers and grow their business. They can create new revenue streams, improve customer satisfaction and attract new customers, and even improve their own operations, systems and customer service.

There is a first-mover advantage, however. A build-out like this is going to take planning, deal-making, engineering and significant time and money to construct. The service provider that gets to critical mass first within the largest footprint is positioned to run away with a market potentially valued at tens of billions of dollars. Those who hesitate will find it nearly impossible to catch up.

Chuck Byers is the CTO of the OpenFog Consortium, which joined forces with the Industrial Internet Consortium on January 31. The combined association is creating a powerhouse of architectural vision, use case creationand testbed designs for IIoT, including fog-edge computing. It is a great starting resource for service providers interested in deploying fog-edge services. 


Winning with Customers in the Age of On-Demand Entertainment


Rethinking Data Management in the Face of 5G

Latest Videos