Why enterprise computing is living on the edge

Like client/server and cloud before it, edge computing will be a critical component in any enterprise IT strategy. Here why you need an edge strategy to be competitive.

People talk about cloud as the way to scale out infrastructure needs. And indeed, cloud and hybrid cloud infrastructure provides an attractive way for enterprises to deliver compute where and when needed at an attractive price. But with an ever-increasing amount of “things” being deployed, from sensors to mobile devices to intelligent assets of all types, the capability to process information locally and react quickly is becoming a necessity.

This movement of compute resources into proximity to the input/output, and which are often analog resources, is what’s driving the rapid rise of edge computing.

Why edge computing is needed?

In many cases, sending data all the way up to the cloud is not an effective or efficient use of compute. Major amounts of generated data needs to be processed and consolidated/aggregated locally for privacy, network transmission savings, latency and better utilization of compute resources. Examples of such requirements include the following:

• Security Cameras generate vast amounts of data, but the majority of such data is not meaningful. What’s needed is a method for data compression with preprocessing only sending anomalous data for further processing. This requirement is met by either edge computing built into the camera, or as edge servers positioned locally to multiple cameras.
• Safety systems needing immediate reaction (e.g., power tools) require extremely low latency if they are to react to emergency situations and prevent machine damage or human injury. Local processing of data in close proximity is required to minimize latency.
• Health monitoring has significant privacy concerns and need to minimize the amount of personalized data sent to computers. Having a protected local resource at the edge minimizes the privacy and security potential for data disclosures, while at the same time also improving any necessary responses due to latency issues that results in sending data up to a distant cloud.
• Autonomous vehicles with collision avoidance and related safety requirements must minimize latency to milliseconds at most, and sending data in a circuitous route to a cloud over a potentially slow or clogged network will not work. Local edge compute is a requirement in this scenario.

What’s empowering the edge?

Companies often struggle with defining their needs in edge computing. Indeed, as can be seen from the examples above, there are many definitions of what the edge is. Carriers are placing increasing emphasis on edge computing as they scale out their 5G networks and create computing platforms at their towers and intermediate control sites that reflect the move to network function virtualization (NFV). They, of course, hope to sell services as edge needs proliferate.

Cloud companies, such as AWS, Google IBM and Microsoft define Edge by how scalable (downwardly) their cloud offerings are. Indeed, Microsoft, for example, makes their Azure Stack components that can run on small server-based systems at a multitude of local sites, and have defined their Azure Sphere as the way to empower individual “things” to take best advantage of edge in a most secure fashion, and has gained Qualcomm and MediaTek as chip partners.

And it’s not just the big cloud providers that are focused on edge. Intel recently announced it is acquiring SmartEdge technology from Pivotal to enable its data center products to better serve the growing needs of edge-based computing. Siemens is enhancing its Industrial Edge portfolio by adding edge runtime and device management software options.

Dell provides its PowerEdge servers for use in edge computing as a way to create scalable servers that can be deployed in many different enterprise situations. Even device makers are getting into the act, and things like smartphones and network access points, now powered by significant computing resources, can be employed as edge servers. And these are only a few examples of companies jumping on the edge bandwagon.

How should the enterprise respond to the edge?

While there are many ways to implement edge computing in the enterprise, here are a few concrete criteria that companies can look at to know if edge is really needed.

• Are there several data-generating sources that provide large amounts of data, but with most data being uninteresting when needing to take further actions?
• Is the amount of time necessary to send and process data, and return it to the point of action large enough to cause inefficiencies in command, control or actionable intelligence needs?
• Is it cost effective to send large amounts of data to a central cloud processing facility or would it be more cost effective to do localized processing?
• Can localized edge computing be incorporated directly into the devices that are generating the data (e.g., security camera, machine tool, medical equipment)?
• Can your preferred cloud (either public, private or hybrid) be segmented in such a way to provide an instance of the cloud capability on a localized server?
• Can your applications and other software assets be segmented to run across an edge-enhanced infrastructure?

The above are starting points and not the final word on whether or not edge computing will be of benefit. Each enterprise must look at how edge can help it create a more efficient operation and/or enable new ways to do business unavailable without edge.

Bottom Line: Edge computing is an important and often necessary step in achieving maximum performance and minimum cost for a variety of enterprise computing needs. If you are not already looking at edge as an important component of your enterprise computing strategy, you are not going to be competitive in the future. Just like client/server and cloud before it, edge will be a critical component in any enterprise computing strategy.