Edge computing: 5 potential pitfalls

Edge computing is gaining steam as an enterprise IT strategy with organizations looking to push storage and analytics closer to where data is gathered, as in IoT networks. But it’s got its challenges.

Its potential upsides are undeniable, including improved latency as well as reduced WAN bandwidth and transmission costs. As a result, enterprises are embracing it. Revenues in the edge-computing market were $4.68 billion in 2020 and are expected to reach $61.14 billion by 2028, according to a May 2021 report by Grand View Research.

But the edge is also fraught with potential difficulties, and companies need to be prepared to address them if they expect to reap the benefits. Here are some of the hurdles organizations could face when deploying edge computing, some of which might not be obvious.

Choosing the best approach

Because edge computing is still relatively new there’s no established base of success stories or metrics that prove its worth to IT decision makers.

“I think the one of the biggest challenges for edge decisions right now that no one really talks about is that there is very little real-world performance data available to help drive edge-deployment decisions,” says Jennifer Cooke, research director, edge strategies, at IDC.

Plus, there are many options to pick from—on-prem, hosted on-prem, managed by ISPs or cloud providers—and sorting through them may be beyond the capabilities of many enterprises.

“There’s a myriad of choices out there, but it’s often overwhelming for organizations to navigate,” Cooke says. “The reality is that edge solutions require a lot of coordination across different providers—from the database and applications to the infrastructure and then the connectivity. For this reason, many organizations are turning to partners to assemble the ecosystem for them.”

As part of the process, these organizations should seek edge-integration partners that can quantify the increased performance and cost reductions that vendors tout.

“I’ve also observed a recent shift from a [do-it-yourself] attitude on edge deployments to accepting that it’s often better to lean on partners to manage edge resources,” Cooke says. “The pandemic actually accelerated this trend, showing organizations that remote monitoring and leaning on partners for management actually worked pretty well.”

Security risks and expertise

As with anything else related to IT, the edge has its own set of security threats and vulnerabilities.

“How does an organization fully consider the many layers and subsegments and achieve a zero-trust environment?” says Matt Kimball, senior analyst, data center, at Moor Insights & Strategy.

“This has to include infrastructure, networking, the full software stack, and the integration of all these different elements with each other, the cloud, and management and monitoring consoles,” Kimball says. “This market segment is so layered and in some ways so nichey that CISOs and IT executives need to invest heavily in the people that are tasked with designing and implementing a holistic strategy. And those people are hard to find.”

The security risks that organizations should keep in mind before formulating an edge strategy include the potentially enormous number of IoT devices and supporting infrastructure the edge requires, as well as the massive volumes of data they generate—all of which need to be defended to protect data and the network.

Some vendors provide tools to help bolster edge security, Kimball says. “But again, the challenge is finding the people that can make sense of all of these challenges, competing solutions, and design a zero-trust environment—one that is fully integrated from the device to the edge to the cloud to the data center,” he says.

Supporting data management and analytics

Being able to analyze data at the edge and draw insights from the analytics is one of the appealing aspects of this environment, but the process is not easy for enterprises.

“The edge is basically a large-scale distributed data-management problem statement,” says Vijoy Pandey, vice president of engineering and CTO, emerging technologies and incubation, at Cisco.

Data management and data science are as important as security to the business success of organizations, Kimball says. “It should be no surprise that the company that can more quickly derive nuggets of intelligence out of the mounds of data being generated every second is the company best positioned to win,” he says.

Like security, data science is hard and practitioners scarce, Kimball says. “And I believe this is equal parts technical and art,” he says. “The tools must be in place to glean the best insights. But a good data scientist understands the nuances of what data [is] most important to the business. And like security, the folks that are good at this are in high demand.”

This is an area where organizations can benefit from an outside perspective, Kimball says. “There are consulting organizations, VARs, and specialists that get this space very well and have had success in deploying the data edge,” he says. “As an ex-IT executive, I can tell you that I despised having to go outside my organization to drive IT initiatives. But I also knew that relying on outside help for uncharted waters always paid off.”

Prepping the IT infrastructure

Creating a network to support edge computing takes time, money, and knowledge—resources that not every organization has enough of.

“IT is used to shipping a few servers and implementing the backup necessary for home-office connectivity,” Kimball says. “Look at any chain [retailer], and you’ll see a couple of servers that are used to connect to the home office and run operations locally in the event of a break in connectivity.”

But with IoT deployments and the need for more complex storage and data processing at the edge, now those environments have to do more than simply run the back office, Kimball says. “As a result, deploying and managing these environments is more critical and challenging than ever,” he says.

In general, a good practice is to keep things simple when it comes to infrastructure and IT operations, Kimball says. “I believe most of the IT-solutions vendors have viable and solid offerings from a hardware and software infrastructure perspective,” he says.

Because of this, he suggests that when IT executives are looking to invest in edge infrastructure, they should start by considering the vendors they already know.

“If you’re an IT shop that has standardized on Dell, it is probably your best starting point for edge infrastructure,” Kimball says. “Likewise HPE, Lenovo, Cisco, Supermicro, etc.”

Duos Technologies, which provides automated edge systems for railway operators, faces challenges providing connectivity and power when it deploys those systems in remote areas.

“But these are relatively straightforward obstacles to scale in most cases,” says Scott Carns, Duos’ chief commercial officer. More challenging is finding ruggedized servers that can operate in the same environment. “Most servers are designed to be installed in data centers and IT environments that are brick and mortar, with perfect power and environmental control systems.”

Edge computing for Duos Technologies’ applications requires robust, well-designed hardware. While its servers are mounted in traditional cabinets or racks, “the operating environment does not have the same level of continuity and protection a data center would provide,” Carns says.

Scaling without creating complexity

The edge has the potential to be extraordinarily complex, given the number of systems, devices, and applications involved.

“Most edge investments [have been] driven by the need to solve a specific business problem by a non-IT business owner,” says Gil Shneorson, senior vice president, edge portfolio, at Dell.

As a result, organizations may have multiple individual edge devices performing specific tasks and operating off their own infrastructure. Each solution was bought, deployed, managed, and secured independently over time, leading to inefficient infrastructure sprawl at the edge, Shneorson says.

“We’re seeing a transition in the industry where IT is called in earlier in the process so they can apply IT best practices and strategic thinking to the edge environment across multiple use cases,” he says.

Being called in earlier is a good idea, but for many IT teams, that means having to architect a single, flexible, efficient infrastructure to support more edge tasks. IT needs to consolidate even as the edge expands, “by modernizing their edge technology foundation and data pipelines with consistent hybrid-cloud architecture, operations, and management, so they can derive the most value from data across use cases, sites, and clouds,” Shneorson says. Meeting that challenge is a major step forward.