The edge takes shape: The 5G telco cloud that would compete with Amazon
Micro data centers at the edge of the wireless network were first considered a way for telcos to earn some extra revenue. Now they’re being leveraged to solve a deeper, darker problem: managing backhaul.
The edge, as telecommunications providers (telcos) perceive it, is the part of their networks that resides closest to their customers in the supply chain. The closer that content stores and databases may be brought to their consumers, the easier they become not only to access, but also to maintain. On the web, multimedia content (e.g., Netflix, videos from CNET) is brought closer to the customer through the use of data centers with exclusive, high-bandwidth connectivity. These content delivery networks (CDNs) prove their value proposition, ironically, by bypassing the internet, tunneling around its public routes and extending themselves as close to their end-users as possible.
Telcos are hopeful they can earn desperately needed revenue from 5G by offering cloud computing and data center hosting services that compete, respectively, with the likes of Amazon and Equinix. Building cloud data centers that can run telco base stations off-site, while at the same time offering cloud services to their own customers, was the idea that inspired 5G in the first place. The problem is, telcos may need some of those services up and running right away, well before true 5G wireless service gets underway. They need an urgent reason to build services at the edge today.
A long drive past what most communications providers would consider "the edge," at the intersection of US Highway 49W and State Road 8, you'll come across a little town of 2,700 people about dead-center of the Yazoo-Mississippi Delta, called Ruleville, in Sunflower County, Mississippi. It's known for being the home of civil rights activist Fannie Lou Hamer, who in the 1960s fought her way back from being unwillingly sterilized by her own hospital and beaten by her jailors, to co-found a black wing of the state's Democratic Party, catalyzing a voters' rights movement that continues to this day.
But right where she is, Ms. Annie receives perhaps the best care in the state. Since 2014, she's been part of a telecommunications experiment — one that has already revealed kinks in the supply chain for 4G wireless, and points to a radical new solution with 5G.
On paper, it's called the Diabetes Telehealth Network. It enables Ms. Annie to use a blood glucose monitor connected via Bluetooth to an iPad, to send data about her glucose levels daily to experts from larger medical facilities throughout the network, including the University of Mississippi. She and her caregivers receive same-day analysis, advising her as to how she should be eating, and to what extent she should be exercising.
Since 2014, Ms. Annie has been receiving semi-regular visitors from up north, who you might not expect would be making the trip: commissioners from the FCC. When her service was first established, she met then-Commissioner Mignon Clyburn. Last June, it was Commissioner Brendan Carr's turn.
"The tablet chimes every morning as a reminder," Carr told a House subcommittee last July [PDF]. "Ms. Annie then pricks her finger and her A1C level is displayed on screen. Based on that, the app suggests appropriate actions: from a particular food or exercise, to watching a relevant video. If she forgets to enter her numbers that day, she'll get a phone call from a nurse. With this technology, Ms. Annie's A1C levels have gone down and she says she's never felt better."
It probably isn't the iPad or the glucose monitor that has captured the commissioners' attention. Ms. Annie gives us a glimpse into a future where millions of Americans, both ailing and healthy, share their vital signs, glucose levels, sinus rhythms, and brain wave patterns with a myriad of servers.
We talk about an Internet of Things as though we can simply attach it, like an add-on pack to a video game, or like another congressional subcommittee, and it will just power itself on and everything will work splendidly. In so doing, we ignore the difficulties we're already facing with the internet we already have, and the things we've already duct-taped onto it. With HIPAA already mandating a circuitous, twisting route for the exchange of electronic health records, a nation full of Ms. Annies cannot be sustained by 4G. It won't work.
The frontline of backhaul
Today, particularly for Ms. Annie in Ruleville, connectivity is not a problem.
Ruleville is covered by a wireless access provider called C Spire, which partnered with healthcare providers in establishing the Telehealth Network. C Spire is part of the reason rural Mississippi has been blanketed by 4G LTE coverage. In April 2017, as part of an effort to extend its signal over longer distances, C Spire began adding more low-band, sub-gigahertz (700 MHz – 800 MHz) spectrum to its transmitter sites -- well before 5G technology providers Ericsson and Qualcomm both began advising all carriers to do the same, to ensure enough bandwidth for all the applications they've promised for 5G, including telehealth.
The issue, in a world where everyone may have a health monitor like Ms. Annie's, is deep in the core of the network. The hardest job of any wireless access provider is to ensure that the network behind that access point is accessible. The name for this job has been borrowed from the shipping industry: backhaul. Everything a wireless network makes available to you from its access point, is the product of backhaul.
Whenever you begin tracking the sources of a network's backhaul, you very literally re-enter the mid-19th century. The nation's right-of-way laws rely upon claims made by railway companies, and much of America's Interstate highways are planned around the routes first taken by trains. Trains often made stops at river ports, or cities built along rivers.
As a direct result, the US' fiber optic cable system closely follows the old railway and highway routes, because their rights-of-way still permit telcos to use narrow strips of land adjacent to these routes. In 2015, a research team led by University of Wisconsin, Madison, computer science professor Paul Barford, released the results of a four-year effort to plot the locations of long-haul fiber in the continental US. That effort involved reproducing data provided by tier-1 internet service and cable providers, but also included diving into printed, paper records in municipal offices.
In the map at left, where the white pin represents Ruleville, the colored pins mark the approximate locations of physical internet fiber conduits. Any effort to bring Ruleville closer to "the edge" will involve either utilizing existing fiber lines between a nearby base station and one of these conduit locations, or laying new lines.
The architecture of 4G LTE was not about altering the internet, but making the existing system accessible. The costs of accomplishing this while leaving the internet alone are already unsustainable for some telcos.
Richard Webb, who directs research into mobile backhaul and small cells for analysis firm IHS, recently remarked: "One thing is certain: If we continue to build backhaul as we currently build it, for the 5G environment, it will not cope."
For applications that require sub-millisecond latency, 5G architects are working to build mmWave. During a 2016 IHS webinar, Webb advised that simply expanding backhaul capacity, even throughout the network, will not be enough to ensure maximum connectivity. 5G network traffic will not only have high volume, but tremendous complexity. "For backhaul, this means it's not just about how much capacity mobile operators add when deploying 5G," he told his audience, "but where in the network they add it, what technologies underpin it, and how the network is going to be managed to cope efficiently with the volume and complexity of that traffic."
"I live in Austin, Texas. And I get backhauled to St. Louis, Missouri, before I do anything on the internet," stated Cole Crawford, the CEO of edge computing infrastructure provider Vapor IO.
Vapor IO is one company defining architecture for a micro data center -- for instance, the Vapor Chamber cylinder small enough to be hauled on a highway. Crawford was speaking to me through a commercial online communications link. I was in Indianapolis, about 460 miles away. My backhaul, before I do anything on the internet, comes from St. Louis as well.
"Because my provider doesn't have a central office in Austin, they have to take long-haul fiber back [from St. Louis]," remarked Crawford. "Think about the size of that pipe! Think about how much fiber is needed to terminate all the people on my network, onto a network that allows them to watch Netflix. That's a lot of fiber, and it's a lot of money."
For the experiment that is 5G wireless to be successful, several very audacious projects have to go exactly right. Certainly 5G has no bolder goal than the establishment of compact micro data centers in close proximity with 5G wireless transmitters at the edge.
Building micro data centers such as this one in Chicago would enable telcos to relocate their base station processors from the transmitters themselves -- whose cooling costs have already become unsustainable in some parts of the world -- to ground-based, centrally cooled facilities. Telcos could recoup their costs for these facilities -- and might even earn a profit from them -- if they permitted the third-party firms charged with managing their transmitters today, to operate commercial cloud services from those micro data center locations. These firms would be competing with Amazon, Microsoft Azure, and Google Cloud, at least for some services, and they would be sharing their profits with their adjacent telcos.
You might compare the size difference between a hyperscale cloud data center and a micro data center on a scale similar to the one between a football stadium and a football. But with the software infrastructure being planned for these edge facilities, that might not matter.
The Kinetic Edge
For the telecommunications industry to want to build micro data center facilities now, instead of waiting until all the 5G transmitters are built (or until the "5G Evolution" phase has run out), there needs to be a near-term payoff -- some source of revenue that telcos can put to use right away, even before 5G service truly comes to fruition. Vapor IO's proposed solution is a network of micro data centers assembled through an organization of partners announced last February 13, called the Kinetic Edge Alliance. The premier service KEA would enable is private wireless 4G LTE network access for telcos' customers, using unlicensed spectrum that is not exclusive to anyone.
A Kinetic Edge station would serve, Crawford explained, as a peering site where customers may cross-connect directly into a wireless provider's network packet core -- which for 4G is Evolved Packet Core (EPC), and for 5G is expected to be replaced with what's called 5G Core. This way, the network hand-off between the content provider and the network provider could happen instantaneously. This would be a boon for certain streaming multimedia providers who want instantaneous access, in place of the CDN arrangements they currently have.
It might also be fertile ground, the CEO added, for a telehealth provider. Custom health monitoring devices could access this wireless network on frequencies completely separated from wireless phones. In fact, one of the new Kinetic Edge Alliance's principal new members is Federated Wireless, which helps businesses establish new and custom wireless applications using the unlicensed Citizens Broadband Radio Service (CBRS) spectrum — about 150 MHz set aside in the 3.5 GHz band.
"We've taken our software to enable spectrum and core network functionality, and integrated it into the Packet Bare Metal platform," explained Iyad Tarazi, Federated Wireless' CEO, speaking with ZDNet Scale. Packet, as you've seen in ZDNet, offers a self-service provisioning system for enterprise customers similar to how a cloud offers virtual machines, except with physical ("bare metal") servers. By way of Packet -- another charter Alliance member -- a customer such as a telehealth provider may be able to establish a wireless access service with customers wearing custom devices, with all the data center power that service will require, on bare metal servers, in an interval of time that Packet clocks at about eight minutes.
"We've done a three-way integration with the Vapor team to bring that all the way into the locations that Vapor's deploying," continued Tarazi. "So as it's rolled out, all a customer has to do, once they activate their service, is deploy access points with Vapor that are pre-integrated with our platform, and they can be up and running with a CBRS, private LTE, and, in the future, a private 5G network, at the edge or at their location."
In the data center colocation industry, a plum selling feature for any site provider is direct connectivity with carrier access lines, usually connected by optical fiber to base stations. The most desirable, and often the largest, colo sites are the ones with these "carrier hotels."
The KEA's plan would turn this entire strategy inside out, by relocating data centers to within a few dozen feet of carriers' core networks -- making base stations into "data hotels." Then by moving the applications servers into the same facility, the network that will be required by law for a HIPAA-compliant operation would be radically simplified.
Throughout the year, the KEA will launch charter service sites in six select markets: Atlanta, Chicago, Dallas, Los Angeles, Pittsburgh, and Seattle. Two Chicago sites are online now. The Alliance's goal -- probably by the time 5G service is finally switched on -- is to provide micro data center accessibility in as many as 30 cities, reachable by half the US population.
That level of accessibility may still be a few years away for Ms. Annie in Ruleville, Mississippi. But something tells me she'll be ready and waiting.