中文 English

For The Edge, It’s All About Location, Location, Location

Thinking outside the data center to bring computing closer to users.

popularity

They are centrally located, are connected to power grids and water systems, and are rapidly thinning out. And you can probably get a new cell phone case or a corn dog in the atrium.

Could shopping malls become a future home for the edge?

Edge computing has transformed over the last few years from being a vaguely defined concept to a fundamental part of the future data infrastructure. Bandwidth costs, the emergence of ultra-low latency applications like intelligent robotics, and the growing pervasiveness of cloud native software among other factors are prompting a shift to bring computing closer to where it gets created, consumed, and analyzed.

An estimated 75% of data will be processed outside traditional centralized data centers by 2025 with around 1/3 of server shipments for telcos ending up in the edge hinterlands. What the “edge” means, of course, remains a subject of debate. For the sake of argument, let’s say it’s a server or data center that can keep the round trip of data to 50 milliseconds or less or within 100 miles and often less to the user.

Let’s go back to the mall. Coresight Research estimates that 25% of the 1,000 of U.S. malls left standing will empty out over the next three to five years. They generally are located within or next to residential and business districts, reducing latency and costs. Water, power, and often fiber links exist already, although they may not be adequate for the data loads the edge will require.

Or think of downtown business corridors. Even after Covid-19 recedes, working from home, either full or part time, will remain a reality for many. That means centrally-located floor space outfitted—to some degree—for edge operations becoming vacant at a time when companies and people are clamoring for extra bandwidth for better videoconference.

Now hit the road. Many states have abandoned industrial sites and it can be easier to convert these into modular data centers instead of hipster condos. As an added bonus, many industrial sites also come complete with dirty canals and wastewater pond: it’s the ideal raw material to fuel the growing interest in liquid cooling and to keep the Water Use Effectiveness (WUE) rating low.

A sophisticated edge will ultimately also relieve some of the pressure on cloud providers in getting more hyperscale centers online to accommodate growing demand. Amsterdam, Beijing, and other cities have blocked or placed strict requirements on data center construction because of their power and water requirements. Diesel generators for back-up power must meet increasingly stringent emissions and runtime limitations. (A 3MW backup diesel generator can emit 1.6 million tons of greenhouse gases per year and cost over $1 million to operate, according to statistics from Cummins.)  As a result, approximately 80% of potential sites for traditional data centers have to be abandoned, according to Aggreko.

Intelligence at the edge

More importantly, edge will improve the computing experience by reducing latency and cost while improving resiliency and quality of service. In the mall, edge capacity would provide the needed performance to run AR simulations to show shoppers what they might look like in a particular jacket or how it could be paired with some new pants at another store. Real-time foot traffic analysis could optimize advertising buys. The available local capacity could also be used to fuel entertainment venues and speed up connectivity at nearby apartments. Suddenly, the dying mall becomes a destination again.

Or think of work. The massive amount of data that can be collected by lights, motion sensors, HVAC and other systems could be translated into anonymized, multidimensional heat maps to optimize space and pave the way for hot desking that works. Processing data at the edge would effectively give facilities managers the green light to track more data—because it costs less to track and could be more easily directed to the people who could take advantage of it—and ultimately develop more applications.

The impacts will be both unpredictable and wide ranging. The high cost of pharmaceuticals is a worldwide controversy with manufacturers pitted against consumers and regulators. Edge computing is creating a path for modular manufacturing where smaller, more cost-effective facilities can be managed with the same necessary real-time precision as traditional facilities. Personalized medicine—where individualized batches of medicines can be produced with precision in liter qualities—will follow as edge becomes more pervasive and powerful.

So what’s the catch?

For this ubiquitous, invisible, high-performance edge to take shape we will have to improve the underlying technology. Edge equipment will have to be far more energy efficient than traditional equipment. Power and water will be constrained. Waste heat—one of the biggest issues for data centers—can aggravate maintenance problems and equipment failures. New processors, storage systems, server designs and networking that deliver more work with fewer watts will be required.  Edge equipment will have to be extremely robust. It will also have to be reprogrammable, unlike much of the customer premise networking devices in use today.

We will also need broadly shared standards around device design, security, over-the-air upgrades, and software conventions to eliminate the friction that can come with deployment and maintenance. Today’s data center designs were honed over 40 years. We won’t have that luxury of time with the edge.

Standards will also be critical for keeping costs within reasonable parameters. While 5G and edge services promise to dramatically increase the number of connections, many of these new devices, however, will likely be linked to skinny service packages. An open hardware strategy that allows providers to mix and match equipment for 5G networks could lower the TCO substantially, according to analyst Caroline Gabriel of Analysis Mason. Reducing waste heat and other maintenance issues, meanwhile, should help keep truck rolls to a minimum.

The edge will be absolutely necessary to realize the vision that’s been sketched out for digital transformation. The challenge over the next few years will be figuring out the mix of technology and real-world assets to make it happen.



Leave a Reply


(Note: This name will be displayed publicly)