As technology extends deeper into every aspect of business, the tip of the spear is often some device at the outer edge of the network, whether a connected industrial controller, a soil moisture sensor, a smartphone, or a security cam.
This ballooning internet of things is already collecting petabytes of data, some of it processed for analysis and some of it immediately actionable. So an architectural problem arises: You don’t want to connect all those devices and stream all that data directly to some centralized cloud or company data center. The latency and data transfer costs are too high.
That’s where edge computing comes in. It provides the “intermediating infrastructure and critical services between core datacenters and intelligent endpoints,” as the research firm IDC puts it. In other words, edge computing provides a vital layer of compute and storage physically close to IoT endpoints, so that control devices can respond with low latency – and edge analytics processing can reduce the amount of data that needs to be transferred to the core.
In “Proving the value of analytics on the edge,” CIO contributor Bob Violino offers three case studies that illustrate the benefits of edge architecture. Two involve transportation: One centers on the collection and processing of telematics from fleets of freight vehicles to improve safety; the other focuses on real-time collection of traffic data in Las Vegas to improve the city’s traffic control. The third is an epic edge case: Adding analytics processing to satellites that capture geospatial imagery, cutting the amount of data transferred to the ground.
Edge architecture is also shaking up one of the original IoT areas, medical devices. Processing medical IoT data at the edge at scale is a relatively new idea, explains Computerworld contributor Mary K. Pratt in “The cutting edge of healthcare: How edge computing will transform medicine.” With the healthcare industry faciing a fresh wave of data emanating from wearable health monitors, allocating edge compute power to process those petabytes will become increasingly imperative.
InfoWorld’s Martin Heller takes a different tack in “How to choose a cloud IoT platform.” All the major clouds offer platforms for IoT asset management — cataloging devices, monitoring them, updating them, etc. Also, they provide edge “zones,” appliances, and various on-prem cloud choices that can serve as edge computing nodes. And of course, the big clouds offer all the analytics options you could want for processing IoT data.
Unfortunately, you can’t escape the fact that the more you physically distribute your compute and storage, the more you increase your attack surface area. That’s one concern examined in “Securing the edge: 4 trends to watch” by CSO contributor Jaikumar Vijayan. Another trend is even more obvious: Escalating alarm over the inherent vulnerabilities of IoT devices themselves, which together raise the ante for edge security. One positive development Vijayan identifies is the accelerated shift to SASE (secure access service edge), which integrates SD-WAN and security into a single edge solution (see the guide “Who’s seling SASE and what do you get?“).
Security is only one of the liabilities raised in Network World’s “Edge computing: 5 potential pitfalls.” Complexity is the leading villain — there are so many choices of technologies and providers that enterprises often turn to partners for planning and implementation.
But that’s true of many emerging areas of technology. Edge computing is exciting because it signals a shift in the way enterprises view the IT estate: If we’re really going to transform the enterprise, then appropriate technology must be deployed in every corner the business, with streaming data feeding continuous optimization. Edge computiing provides a framework for that vision.
