The ongoing evolution of intelligent edge computing

Since information technology was first widely embraced by organisations back in the 1970s, development cycles have evolved in a way many people liken to a swinging pendulum.

Initially, compute resources were centralised and based on large-scale mainframe computers. Then, the pendulum swung the other way, and resources were shifted closer to users in the form of personal computers and standalone servers.

Since then, the pendulum has swung back again, with resources being transferred to the cloud, where they reside in large-scale data centres and are accessed remotely as required.

Interestingly, this process of change has not stopped. However, the next phase is unlikely to involve another wholesale migration of IT resources away from centralised facilities. Rather, it is likely to be a combination of cloud platforms and increasingly intelligent edge devices.

The rise of the edge
In a growing number of cases, this rise in edge computing is being driven by the rapid evolution of artificial intelligence. While large-scale, centralised IT infrastructure is needed to train AI models on masses of data, those models can then be run on much smaller, less compute-intensive devices at the edge.

One good example is a smart surveillance camera. The AI model needed for it to recognise objects and people will require large amounts of data and resources to learn, but it can then run on the more limited resources within the camera when it applies that knowledge.

Another is language translation. While significant resources are needed to train a model that can accurately deliver such a service, the model itself can then run on a much smaller device.

This rapid rise of intelligent edge computing is being driven by two key factors. One is the ongoing miniaturisation of IT. If you consider the capabilities of today's smartphones, having access to that amount of computing resources 20 years ago would have required a room full of servers.

The second factor is the ongoing evolution of networking technology. Edge devices will often be located in places where traditional high-speed connections are not available.

For this reason, the data collected by the device needs to be processed on that device, and the results fed back to a centralised datacentre via whatever network links are available. In many cases these are likely to be Wi-Fi, 4G or 5G wireless connections.

An ecosystem of players
For an intelligent edge infrastructure to function successfully, it requires a number of different parties to work together. The list includes the hardware vendors that design and build edge devices that interact with and collect data from the physical world, which can include anything from movement to rainfall. These devices also act as the input points into the edge system and provide the data that enables AI to make decisions. 

Intelligent edge also requires centralised computing and storage resources, typically provided by a cloud vendor. This is where all the incoming data is stored and processed.

The third required group is the connectivity players. This could be a traditional networking vendor or a telecommunications carrier offering data service connections. This group is particularly vital, as without that connectivity, the entire system is unable to function. 

Finally, intelligent edge computing requires software developers who can build applications that take advantage of the vast trove of collected data. This is what will allow organisations to extract as much useful insight from their edge resources as possible.

Ongoing evolution
The rise of intelligent edge computing is by no means the end of the evolutionary process. Indeed, many industry observers believe that the next step will be to actually blur the distinction between centralised cloud computing and edge computing altogether.

From a business user's perspective, where their application is running is less important than the outcomes they are achieving from it. If the best location is in the cloud, it can run there; however, if higher performance can be achieved at the edge, then that will become the chosen location.

Use cases for intelligent edge computing will continue to evolve in the coming years. As the capability of edge devices improves and the AI algorithms used to analyse the collected data become more powerful, the ways in which they can deliver business value will explode in number.

Just as distributed computing brought exciting new opportunities in the 1980s and 1990s, and cloud platforms change the game in the 2000s, intelligent edge computing will bring significant opportunities for development and growth in the coming years.