Cloud computing alone is insufficient to power the next-generation of apps for self-driving cars and augmented reality. These technologies need secure compute power at the edge. The solutions are just around the corner that will accelerate the pace of change. The only question about edge computing remains: Are you ready?
-It’s 2025, and you’re asleep at the wheel, driving down a highway at 70 miles an hour. A deer appears on the side of the road and scampers across four lanes. In under 10 milliseconds, your autonomous car slows and changes lanes to avoid an accident. You keep right on sleeping.
This is the promise of self-driving cars, where machine reaction times to road hazards are significantly faster than human reaction times. The same almost instantaneous response time is required for the commercialization of a whole range of new applications, from augmented reality (AR) to robotics. Without it, virtual images are out of sync with the user’s view of the real world, which is disorienting. Consider a doctor in a hospital in Los Angeles performing remote surgery on a patient in Afghanistan. Any latency in the images the doctor sees could be fatal.
Now consider other yet-to-be-commercialized apps, like Google’s Project Soli, which detects precise gestures such as fingers turning a virtual dial that is visible only to the wearer of AR glasses. Or Magic Leap’s “neurologically accurate” images that go beyond AR glasses by superimposing virtual images on a user’s eyeballs to create the overlay effect.
The technology to enable these and other yet-to-be-developed apps is available today and will become a reality starting in 2019 with the rollout of the fifth-generation mobile communications system (5G) and the installation of multi-access edge computing (MEC) infrastructures. Compared to 4G, 5G promises a 10X decrease in latency, a 10X increase in throughput, a 100X improvement in traffic capacity and a 100X improvement in network efficiency, according to ABI Research.
If you’re not boarding the low-latency train already, you risk being left at the station.
Challenges at the edge
MEC delivers access to local compute power, software, and algorithms that sense, think and act almost instantaneously by bringing the datacenter closer to the point of consumption. Edge computing technology lets cloud farms cooperate with cell towers, regional datacenters, autonomous driving systems and ultra-thin smart glasses. MEC pushes apps, data, analytics to the logical extremes of the network.
But operating at the edge introduces new challenges for carriers, network providers, industrial companies, enterprises and IoT networks. Most of these concerns revolve around ensuring the resilience of the service and the privacy, reliability, and security of the data, algorithms, and apps that comprise the MEC.
They are similar to the challenges companies faced when they transitioned from operating their own monolithic datacenters to cloud-based computing. Today, this means moving server-less applications, open-source microservices, and multi-vendor machine-learning frameworks to the edge of the network. Distributed systems require a strategy that our MEC engineers call a “small-unit defense” which includes a few core principles:
- Edge autonomy. Security can be enhanced by granting more autonomy to localized assets, whether they are virtual assistants, in-car infotainment systems or industrial equipment.
- Nano segmentation. The “blast zone” of exploitation needs to be reduced and cordoned off around the impacted microservice and serverless function.
- Crowd-sourced intelligence. Expert anti-malware hunters need to be incentivized to log vulnerabilities and decipher bad-actor behavior and threat indicators.
What’s your MEC action plan?
Whatever platform your applications are designed for—augmented reality, autonomous vehicles, public safety, robotics, or smart buildings and cities—the objective of the edge architecture and software solutions are the same: low latency, high bandwidth, high throughput and superior network efficiency. At the same time, all of these services must be secure and resilient.
Here are three principles for delivering secure edge computing services that provide a useful framework for developing MEC capabilities:
- Get the design right. This requires a clear understanding of the requirements of the end user and the enterprise to ensure the application satisfied or better yet, exceeds their requirements. It’s worthwhile and less costly to spend the time up front to design the architecture than to re-architect later in the process.
- Leverage existing reference architectures or, if need be, build customizable reference architectures to reduce your time-to-market.
- Onboard applications quickly but make sure you’re doing it safely and securely.
Edge computing and being 5G ready offers a unique opportunity for innovative companies to define the future with new services that are hard to imagine today. But within the next few years, the market leaders will be established, and the window of opportunity will narrow.
It’s a disruptive period for all companies but living on the edge today will pay dividends tomorrow.