Edge Computing vs. Cloud Computing: What’s the Difference?

The term cloud computing is now as firmly lodged in our technical lexicon as email and Internet, and the concept has taken firm hold in business as well. By 2020, Gartner estimates that a “no cloud” policy will be as prevalent in business as a “no Internet” policy. Which is to say no one who wants to stay in business will be without one.

You are likely hearing a new term now, edge computing. One of the problems with technology is terms tend to come before the definition. Technologists (and the press, let’s be honest) tend to throw a word around before it is well-defined, and in that vacuum come a variety of guessed definitions, of varying accuracy.

Edge computing is a term you are going to hear more of in the coming years because it precedes another term you will be hearing a lot, the Internet of Things (IoT). You see, the formally adopted definition of edge computing is a form of technology that is necessary to make the IoT work.

Tech research firm IDC defines edge computing is a “mesh network of micro data centers that process or store critical data locally and push all received data to a central data center or cloud storage repository, in a footprint of less than 100 square feet.”

It is typically used in IoT use cases, where edge devices collect data from IoT devices and do the processing there, or send it back to a data center or the cloud for processing. Edge computing takes some of the load off the central data center, reducing or even eliminating the processing work at the central location.

IoT Explosion in the Cloud Era

To understand the need for edge computing you must understand the explosive growth in IoT in the coming years, and it is coming on big. There have been a number of estimates of the growth in devices, and while they all vary, they are all in the billions of devices.

* Gartner estimates there were 6.4 billion connected devices in 2016 will it reach 20.8 billion by 2020. It estimates that in 2016, 5.5 million new “things” were connected every day.

* IDC predicts global IoT revenue will grow from $2.71 billion in 2015 to $7.065 billion by 2020, with the total installed base of devices reaching 28.1 billion in 2020.

* IHS Markit forecasts that the IoT market will grow from an installed base of 15.4 billion devices in 2015 to 30.7 billion devices by 2020 and 75.4 billion in 2025.

* McKinsey estimates the total IoT market size was about $900  million in 2015 and will grow to $3.7 billion by 2020.

This is taking place in a number of areas, most notably cars and industrial equipment. Cars are becoming increasingly more computerized and more intelligent. Gone are the days when the “Check engine” warning light came on and you had to guess what was wrong. Now it tells you which component is failing.

The industrial sector is a broad one and includes sensors, RFID, industrial robotics, 3D printing, condition monitoring, smart meters, guidance, and more. This sector is sometimes called the Industrial Internet of Things (IIoT) and the overall market is expected to grow from $93.9 billion in 2014 to $151.01 billion by 2020.

All of these sensors are taking in data but they are not processing it. Your car does some of the processing of sensor data but much of it has to be sent in to a data center for computation, monitoring and logging.

The problem is that this would overload networks and data centers. Imaging the millions of cars on the road sending in data to data centers around the country. The 4G network would be overwhelmed, as would the data centers. And if you are in California and the car maker’s data center is in Texas, that’s a long round trip.

Cloud Computing, Meet Edge Computing

Processing data at the edge of the network — where it is taken in – has a number of benefits, starting with reducing the latency and makes connected applications more responsive and robust. Some applications might need immediate response, such as a sensor for failing equipment or for detecting a break-in.

It also takes the computation load off the data center if data can be processed and reacted upon at the point of origin rather than making the round trip to and from the data center. So it reduces the burden on both the data center and the network.

One company specializing in this is Vapor IO, a startup that puts mini data centers called Vapor Edge Computing containers at cell towers. The containers are smaller than a car but contain redundant racks of computing systems that use special software for load balancing. The load is balanced both at each container and between containers scattered at cell towers around a city.

A special software stack for managing a group of locations makes the containers in an area function and appear as a single data center. It has all of the standard data center features, such as load balancing and automated site-to-site failover.

Vapor IO is a part of what are known as micro data centers, self-contained systems in ruggedized containers to withstand the elements that provide all the essential components of a traditional data center but in a small footprint. Vapor is not alone, although it is a startup dedicated specifically to the micro data center.

Some very big names in data center technology are also experimenting with micro data centers. Schneider Electric, the European power and cooling giant, has a line of micro data center modules, and Vertiv (formerly Emerson Network Power) has its own line of outdoor enclosures.

It looks to be a growing market as well. The research firm Markets and Markets believes that the micro data center sector could be worth a staggering $32 billion over the next two years.

You may hear edge computing referred to by other names than micro data centers. They include fog computing and cloudlets. Fog computing, or “fogging,” is a term used to described a decentralized computing infrastructure that extends the cloud to the edge of the network.

Cloudlets are mobility-enhanced micro data centers located at the edge of a network and serve the mobile or smart device portion of the network. They are designed to handle resource-intensive mobile apps and take the load off both the network and the central data center and keep computing close to the point of origin.

The cloud is a great place for centralized computing, but not every computing task needs to run on a centralized system. If your car is getting real-time traffic and GPS updates from the surrounding area, there’s no reason to send data back and forth to a data center five states and a thousand miles away. So as the IoT grows, expect edge computing to grow right along with it. They will never be an either/or choice for data center providers, the two will always work in tandem.