Data centers are expanding to the network edge to satisfy demand by AI and other applications requiring fast response times not available from traditional data center architectures.
The problem with traditional architecture is its centralized framework. Data often travel many miles from the sting to computers, then back again. That’s fine when you’re handling email, Google, Facebook, and other applications delivered via the cloud. Human brains are slow computers, unable to register the lag time between, say, clicking on an email message during a browser and therefore the message opening.
But AI and other emerging applications — Internet of things (IoT), cloud-based gaming, computer game — require much faster network response times, otherwise referred to as “latency.” meaning data center processing must move to the network edge. Edge computing can happen in small data centers, roughly the dimensions of shipping containers, instead of the warehouse-sized edifices that currently power the cloud.
Startups like EdgeMicro and Vapor.io are deploying these “mini data centers.”
Datacenter operators can still use their traditional structures, with fast networks and other hardware and software required to make sure the speedy response times needed for edge applications.
And edge data centers can reside on enterprise premises, or in exotic locations like mines, ships, and oilfields.
“The favorite thing [driving edge computing] is that the amount of knowledge being created outside the info center,” said Patrick Moorhead, president and principal analyst of Moor Insights & Strategy. The number of connected sensors will reach 1 trillion by 2024, primarily driven by smart cities and video.
Latency isn’t the sole problem requiring edge computing. “It’s cost — that’s really driving edge computing,” Moorhead said. “Every time you bring data in [to a knowledge center] you pay someone money. the web isn’t free.” Internet providers charge for bandwidth, and cloud providers, like Amazon Web Services (AWS), have “egress charges” for moving data in and out of their clouds.
Organizations need computing at the sting during applications where they can’t access connectivity: in a ship, down a mine shaft or on an oilfield. Meanwhile, a growing list of privacy regulations requires on-site processing in some applications, especially healthcare.
“If you’re a hospital you’re flat-out not allowed to send any data into the cloud,” Moorhead said. And albeit it was allowed, bandwidth costs would make moving much of that data, particularly diagnostic images, prohibitively expensive.
Operators look to the sting
Digital Realty is one of the world’s largest data center operators, differentiating themselves via a worldwide platform and infrastructure starting from massive multi-megawatt facilities to individual cages and racks. the corporate has 267 data centers worldwide, in 20 countries.
Edge locations require a replacement quite infrastructure. “By no means does it appear as if a standard data center,” Digital Realty CTO Chris Sharp said. “The size is far smaller, with workloads requiring tons of power density and interconnection density.” These mini data centers got to be lights-out, with no operations staff on-site. and multitenant support. Dense fiber connectivity back to the core cloud infrastructure is also a requirement.
Digital Realty is in the early stages of deploying mini data centers, with prototypes in Chicago, Atlanta, and Dallas, in partnership with startup Vapor IO.
Mini data centers aren’t the sole option. In many locations, edge applications can run inside a standard data center and still achieve the 5-msec latency needed for rapid application response times, said Russell Shriver, Digital Realty’s director global service innovation. “For tons of enterprises that are trying to find a foothold in major metro areas, that’s getting to be quite sufficient for his or her needs,” he said.
The mini data center remains an emerging market. Indeed, Digital Realty competitor Equinix sees its existing facilities as serving edge needs for its service provider and enterprise customers, said Jim Poole, Equinix vice chairman of worldwide business development. “Equinix, because it exists, is that the edge,” Poole asserted.
Equinix has quite 190 data centers in additional than 44 major metropolitan areas worldwide. Much of the U.S. is already a10-msec round-trip time interval over fiber from applications sitting in clusters of Equinix data centers. That latency covers 80 percent of the U.S. population. Before building new, mini data centers, companies, and repair providers are looking to maximize the deployment of edge applications within that existing infrastructure.
Wireless creates bottleneck
While Equinix is able to do low latency over fiber, edge applications require wireless also, and wireless remains a bottleneck for AI and other emerging edge applications. Current 4G wireless latency is 40 msec at the best, and therefore the average is between 60 to 120 msec, Poole said.
5G promises to slash latency. Hence, service providers are partnering with hyper-scale cloud service providers to require leverage improved performance. AWS and Verizon, for instance, are teaming to attach an AWS data center in downtown l. a. to Verizon’s radio access network (RAN) tower complex within the city. The project demonstrates they will create a sub-10 msec latency zone around the metro area, Poole said. That, in turn, could generate demand for mini data centers. “But until we fix this particular problem, nobody’s getting to spend the incremental capital,” he said.
Additional, 5G “network slicing” capabilities will make it possible to deploy private, wireless networks for added control and security, Poole added.
For now, mini data centers are a promising technology lacking scale. consistent with Poole, “The reason you don’t see people running around making big announcements of deploying hundreds and many these mini data centers is that folks don’t see the business case yet.”
Nonetheless, Equinix does see use cases for modular data centers — not necessarily on the sting, but as a way to enter emerging markets where, for now, it makes little sense to create a $100 million data center.
AI is generating big demand for edge computing, consistent with Kaladhar Voruganti, an Equinix senior fellow,
AI applications include two primary workloads: training and inferencing. Training is what it seems like — teaching an AI model the way to solve a drag. This process often involves organizing petabytes of knowledge.
“Usually you would like tons of computing,” Voruganti said. Training runs on power-hungry GPUs, with each fully loaded rack consuming up to 30 to 40 kilowatts. Training generally must run during a big data center to satisfy power requirements, also as privacy and regulatory concerns in some applications.
Digital Realty has partnered with Nvidia to supply the hardware vendor’s GPUs in colocation servers.
Once models are trained, the subsequent step is inference, a process where the model applies what it’s learned in training and puts it to figure during a production application. Inference requires much less data crunching and may run during a rapidly deployed Docker or another software container at the network edge — during a smartphone, a Tesla, or mini- or metro data centers.
“You might train it within the big cloud, and run the appliance and do the inference right the factory floor, or Walmart, or the gasoline station,” analyst Patrick Moorhead said
These kinds of AI applications are often utilized in a spread of cases. for instance, an airline company might use “digital twins” for predictive maintenance. Or, because the economy exposes from the Covid-19 pandemic, a business could use AI to run heat-mapping and face recognition to spot people entering a facility who could be infected.
Other applications requiring edge compute (and frequently using AI) include gaming, IoT, smart factories, shipping, and logistics. Additionally, retail technologies require edge computing to deliver needed responsiveness.
Moorhead sees a particular demand for edge data centers in retail. A “store of the future” like Amazon Go has many cameras, and likewise, Walmart uses video to trace customers. “They’re driving the heck out of the necessity for this,” he said.
Smart city planners are looking to use AI and other edge applications to market health and safety, track infrastructure maintenance needs, and manage traffic.
Other demands will come from transportation — including much-hyped self-driving cars — alongside advanced manufacturing and visual inspection of products. The energy industry is additionally driving demand, particularly for remote inspection.
Special hardware requirements
Edge AI applications typically use flash storage for top performance, said Equinix’s Voruganti. Those applications also require a high degree of networking connectivity, both from the device to the sting and to the info center. Links are also required between application components which may be running in several locations on the network, Poole said. “They got to have low latency between components and domains,” he added.
Edge computing also must be rugged, for deployment in locations like elevators, transportation system turnstiles and mining equipment, Moorhead said. Shipboard computers must be salt-water resistant.
Edge computing also presents physical security challenges. Conventional hyper-scale data centers have near-military-grade security, but a foothold data center during a country, unguarded, is vulnerable to break-ins, or maybe an attacker carrying a whole remote data center off during a truck.
And the winners are….
Hyperclouds, enterprise vendors, telecom providers, and data center operators all appear as if winners at the sting. “AWS is the big mothership. It’s slowly but surely fielding a reputable edge offering,” Moorhead said.
The public cloud giant unveiled AWS Outposts, a hardware rack running its infrastructure software — and equivalent infrastructure run in an AWS data center. Outposts can run on-premise, on the sting, or during a data center. AWS Snowball, a foothold computer, provides computing, memory, and storage for remote environments such as ships. Another Amazon offering called Wavelength is a foothold device aimed toward carriers, putting computers closer to the sting for 5G deployments.
On the software side, alternatives include AWS IoT Greengrass, an OS that connects IoT devices to the cloud. Meanwhile, public cloud rival Microsoft provides its Azure cloud IoT-for-edge services while VMware also provides edge services. Moorhead said VMware is “surprisingly competitive during this space.”
Google, the opposite major public cloud vendor, has been a touch of a laggard but is stepping up with its Anthos services for distributed cloud applications.
Meanwhile, IT infrastructure giants like Dell, Hewlett Packard Enterprise (edge servers), and Cisco (IoT networking) have a plus since edge computing requires an enormous ecosystem, with the tissue connecting on-premises infrastructure and therefore the cloud, Moorhead said.
Emerging edge data centers vendors like Vapor IO even have a chance to redefine old technology, Moorhead reckons. “There are data centers on the sting for 50 years. Any Walmart features a raised floor and a knowledge center. If you enter a gasoline station or McDonald’s they need a server on the wall,” he said. “Where Vapor IO is basically leaning-in is adding compute on the brink of the network, specifically the 5G network.”
Telco central offices can also be repurposed as mini data centers, creating opportunities for carriers. “A typical neighborhood features a cement bunker with analog lines and a bunch of racks in it,” Moorhead said. “They’re almost empty now. they need tons of power. They’re industrial strength — literally a cement bunker that might be hard to interrupt into — and that they have the facility and cooling.”
Equinix and Digital Reality assert they’re well-positioned at the sting given their strengths as global data center and network operators. “You can’t defy physics,” said Poole. “The telcos will have best — they will create access on the local level. there are no thanks to getting around that.”
He adds, “Datacenter companies like Equinix that have a highly distributed footprint will have best because we are where the applications are today.”
Adds Digital Realty’s Sharp: “You need a worldwide platform otherwise you will have a tough nonce successful. Customers are very cautious about doing deals with point providers in single markets. If you’re not truly invested and have the time to support a worldwide environment, you’re not getting to win.