Helder Almeida - Fotolia
Millions of devices currently linger on the network edge, receiving and transmitting data wirelessly over the vast...
distances separating them from the cloud. In between the device and data center exists the fog: a nascent, distributed networking and computing layer that enables data processing closer to the device itself.
Fog advocates say as the internet of things (IoT) grows, the traditional cloud computing architecture won't cut it. With a skyrocketing number of endpoint devices trying to communicate with the data center, latency will increase and network performance will spiral. The fog offers an attractive alternative: process data locally, minimizing cloud involvement and enabling a smarter, more self-sufficient space between the data center and the network edge.
"There's growing awareness that 'business as usual' won't work in IoT due to latency, bandwidth and inconsistent network challenges," says Lynne Canavan, executive director of the OpenFog Consortium, an organization that wants to develop an open fog reference architecture.
The terms "fog networking" and "fog computing" are often used interchangeably, and their meanings can vary depending on whom you ask. But the two typically go hand-in-hand, connoting an architecture that pushes intelligence and processing power out of the cloud and toward the edge.
Some IT pros also use the term "edge computing" when discussing fog -- a habit OpenFog board member Brent Hodges discourages.
"We work to make sure fog computing isn't just about the edge," says Hodges, who is also an executive in charge of IoT planning and product strategy at Dell. "It's about the 'and' -- taking elements from the cloud and getting closer to where computing is happening."
The stakes are high. Consulting firm McKinsey & Company estimates the potential economic impact of the internet of things could be $3.9 trillion to $11.1 trillion per year by 2025. But without interoperability between devices and IoT systems -- the kind fog networking could provide -- enterprises could lose as much as 60% of potential value, the firm wrote in a 2015 report.
FogHorn Systems, an edge-intelligence software vendor based in Mountain View, Calif., helps organizations realize that value.
"Fog computing has its greatest value as applied to industrial IoT, which is really just beginning," says David King, FogHorn CEO.
For a wind farm operator in India, for example, FogHorn recently installed intelligent control systems on individual turbines to monitor existing conditions and adjust the pitch of blades accordingly. This allows the operator to optimize power production on the whole farm and automatically update its forecasts to the local electricity regulator as often as every 15 minutes.
For an electric utility customer, FogHorn combined its Lightning software with radio frequency identification receivers to track batches of parts during the manufacturing process. The rate of materials that had to be scrapped was cut in half.
"We're allowing you to use operational technology data and IT data and use it simultaneously in a much more powerful way," King says.
Fog does the math
Fog applications extend far beyond the factory floor. A ferry line in the Oslo Fjord in Norway has decreased fuel use by about 15% thanks to fog networking technology. The company's Route Exchange (REX) sea traffic management software analyzes data from onboard sensors to predict arrival times, and then uses that information to optimize routes and ship speeds.
Faced with tight schedules, ferry captains naturally tend to lean on their accelerators -- better to be early than late. REX lets them know when they can afford to ease off the gas, says Geir Fagerhus, MARSEC Inc. president and CEO.
"With inadequate real-time support tools, captains consequently tend to speed up -- with a negative effect on the fuel bill," he says. "With our system dynamically calculating the estimated arrival time within a margin of 15 seconds, the captains get the information they need to fine-tune the operation."
Using fog networking, REX allows ships to share real-time route and navigation information with each other and traffic monitoring centers based onshore, which could help avoid at-sea collisions.
This same kind of connected navigation could also one day translate on the ground, saving time for people driving connected cars to work. And in the air, the Federal Aviation Administration estimates that new air traffic control systems using fog-enabled GPS rather than radar to track planes could reduce delays, save fuel and cut carbon dioxide emissions.
Brent HodgesOpenFog Consortium board member
Schneider Electric, which specializes in energy management and automation products, has customers using fog computing in vineyards in California and Oregon. Sensors in the soil measure temperature, waterfall, fertilizer content and other data points, and then conduct analytics on the spot to automatically control irrigation and other systems.
Steven Carlini, senior director of data center solutions at Schneider, says the company's banking and retail customers are focused on using fog computing to process data closer to users and bring more computing capacity into stores. High-end retailers are using facial recognition to spot their best customers and personalize the in-store experience. At the other end of the spectrum, big box stores are experimenting with augmented reality to display coupons and product information as shoppers walk down the aisles.
A rising tide of data
As an increasing number of smart devices get online, they will generate a growing sea of valuable data. But most of the information that IoT devices collect today goes unused. Industrial applications collecting sensor data on wind turbines, for example, or measuring temperature and stress on a bridge, generate 10 TB of data or more per day, too much to transfer to the cloud.
McKinsey estimates that only 1% of data from an oil rig with 30,000 sensors is actually examined. And even then, the data is not used for optimization and prediction purposes, which analysts say offer the greatest business value.
Processing that data near the edge, however -- close to the oil rig sensors themselves -- would avoid flooding the cloud network with more information than it can handle.
In another example, Schneider Electric's Carlini says security staff at large stadiums and museums -- such as the Sagrada Família in Barcelona -- are turning to facial recognition technology applications that use fog networking and computing.
"Because of the granularity of cameras they have to use to make facial recognition work, there's a lot of data," Carlini says. The fog processes that data locally, comparing visitor scans with terrorist databases without having to involve the cloud.
New fog products and platforms are emerging. One startup, NGD Systems, based in Irvine, Calif., is developing intelligent solid-state drives that perform computation within the storage device itself, rather than moving the data to an external drive or the cloud for processing.
Founder and CEO Nader Salessi says this approach reduces latency and power consumption.
"We are combining storage and edge computing together," he says. "Combined into a single solution, nobody has that."
How fast fog networking catches on is anyone's guess, but there is no doubting its potential.
"The availability of widespread broadband and narrowband network infrastructure and cheaper sensors will make it much more interesting for businesses to exploit these technologies to enhance product features and improve business processes in the years ahead," says Dan Bieler, principal analyst at Forrester Research.
IoT sensors require a new kind of network
Minimize network stress with edge analytics
Edge computing holds key to faster IoT