Edge data centers are being proposed for various applications. They represent a decentralization of data center functions as a way to improve latency, real-time performance, and serve an emerging range of applications in artificial intelligence (AI), autonomous driving, telecom, and many others. The balance is gradually tipping from the efficiencies and economies of scale enjoyed by large, centralized data centers toward smaller nodes located exactly where they need to be.

Here are some of the key use cases that are likely to take shape.  

Miniaturized Data Centers in a Box

The edge data center is going through a miniaturization process – and one that will continue for the foreseeable future. Traditionally, they were thought only to be slimmed-down replicas of main data centers that had to compromise on physical security, power, and cooling. But their functionality is more and more being considered to match that of large data centers. This places a greater emphasis on efficiency. Innovations in technology such as NVMe drives, the latest processors, containerized infrastructure, and cloud-native applications can combine to bring sufficient density that the footprints of edge data centers can be small yet pack a punch.

“The net effect is an architecture that supports a data center anywhere mentality through software-defined infrastructure,” said Satish Ramakrishnan, a senior executive at MinIO.

Mobile Edge Boxes

Mobility is another feature of these mini data centers. Just as data centers can import mobile power generation packages that fit into a shipping container from vendors such as Siemens Energy, GE Gas Power, and Mitsubishi Aero, edge technology facilitates packaging aimed at mobility and reuse.

What we are likely to see are edge data centers in packages that can be moved around as needs dictate. Perhaps due to seasonal demand, or a sharp increase in customer demand, a unit can be moved from A to B to C. Thus, they must be rugged and versatile. They can’t be too large that shipping becomes a challenge. They must be able to survive the rigors of transportation and being positioned in areas that may be subject to weather extremes. And they must be able to support different backhauls options depending on what is available on site.  

Autonomous Vehicle/Traffic Management

One of the big use cases driving innovation and adoption of edge data centers is autonomous and semi-autonomous vehicles. Think about it for a moment. These vehicles utilize sensors, processors, radar systems, and cameras that transmit massive amounts of data. That data needs processed, coordinated, and acted upon in real time or near-real time to avoid collisions, maneuver through traffic, comply to road signs, remain within lanes, and arrive at the destination. Braking, for instance, requires an enormous amount of judgment in the blink of an eye. There is no time to take inputs from hundreds of sensors and systems, transmit it all to a central data center, process it, and then receive instructions. The numbers have to be crunched, fed to AI-based processors, and decision made almost instantly.

The rest of the data center industry will benefit from the deep pockets of the automotive space as they vie for self-driving dominance. According to Omdia, the automotive AI market will amount to around $27 billion globally by 2025. They are spending big in technologies inside vehicles and in surrounding infrastructure to ensure safety, performance, and comfort.

“Surveillance and monitoring channels include traffic cameras, ground pavement sensors, traffic control lights and signals, warning indicators for on the road, as well as applications spanning the tracking, positioning, and telemetric measurement to gain awareness into traffic management,” said Greg Schulz, an analyst at StorageIO Group.

Specialized Edge Boxes

Another market that will play a role in advancing edge data center innovation is telecommunications. With 5G representing a high-value service, edge data centers are needed to facilitate the rollout of these new services and telecom applications.

“5G needs include density, sustainability, low maintenance, and low complexity,” said Hao Zhong, co-founder and CEO of ScaleFlux. “We will have to do a lot more with a lot less; much higher storage density; working in tighter operating envelopes; reduced power budgets and cooling; running for longer life cycles with less maintenance; and analyzing more data with more compute stuffed into fewer nodes.”

Automotive and telecom edge data centers may be in the vanguard, but other specialized units are happening, too. Some users want high-capacity storage nearby, while others need compute power or bandwidth. Expect to see units designed to meet these requirements. They will include boxes that offer double-digit petabytes with all internal infrastructure streamlined to serve a specific niche. MinIO, for example, provides high-volume object storage for the edge running on commodity hardware to serve containerized applications. Other units will be created to being edge data centers closer to application workloads such as the Industrial Internet of Things, (IIOT), gaming applications, analytics, and AI.