Connected devices are generating vast amounts of data, which organizations across all industries can use to make better business decisions and more quickly respond to dynamic work environments and customer engagements.
However, to do that, they can't rely solely on cloud computing to handle all the data, despite spending much of the past decade moving compute resources en masse to private and public clouds.
Organizations are finding that the volume of data is too vast to efficiently and cost-effectively move from those endpoint devices to the cloud for analysis, only to send the processed information from the cloud back to those devices.
Estimates put the problem in perspective: IDC, the tech research and advisory firm, calculated that 55.9 billion connected devices will generate more than 79 zettabytes of data by 2025 -- that is up from the 13 ZB of data generated in 2019.
This article is part of
This has led organizations to turn to edge computing, a technology paradigm that puts compute resources physically close to endpoint devices, including purpose-built gateways and servers and even on-premises virtualized data centers.
The goal of edge computing is the same regardless of the actual device deployed in any given use case: to collect and analyze data from the endpoints and then use that information to drive follow-up actions by those endpoints. This is new territory for most organizations, experts said, but they expect to see a growing number of edge computing trends for years to come.
"Edge computing solves a problem that didn't exist before the growth of IoT," said Dan Miklovic, founder and principal analyst at Lean Manufacturing Research LLC and a member of The Analyst Syndicate.
"When we were only accessing a little bit of information, moving it upstream [to a private or public cloud] and processing it and then moving results downstream, that action was possible. But, now, we're getting gigabytes a second vs. daily or weekly, so the idea of trying to transmit all that data somewhere, process it and send the results back is impossible to do," he said. "Edge gets you closer to the source of that data and lets you turn it into information, lessening the burden on the communication challenge that has a finite capability."
Teresa TungManaging director, Accenture Labs
That, in turn, is opening numerous opportunities for organizations. Technology experts and IT executives expect edge computing to improve existing processes, as well as spur new operating models and new activities. These are, and will be, powered in large part by the following four edge computing trends.
1. Edge computing speed
In many cases, data can technically move from endpoint devices to the cloud, where applications could process the data and then send that information back to instruct those endpoints on the proper actions to take, which can happen in seconds.
However, that might not be fast enough for some use cases, said Teresa Tung, managing director at Accenture Labs. This is where edge computing has an advantage, with its locational proximity delivering the additional speed that some actions require.
"You don't want an action to be delayed by the round-trip loop to the cloud; you really need the edge," Tung said. "It could be milliseconds faster, but for some decisions, that speed is needed."
She noted that 5G does, in fact, move data faster than 4G and LTE networks -- but 5G won't fully eliminate the advantage that edge computing has when it comes to speed, especially as organizations put more traffic and stress on 5G networks as its availability expands.
2. AI and machine learning
Edge computing won't fully replace the use of cloud resources, but enterprises will be putting more AI and machine learning capabilities as close as they can to endpoint devices to ensure those smart processes can work at the speed and reliability required.
For example, consider how using AI along the edge could aid in manufacturing, Tung said. Sensors deployed along the assembly line can visually inspect the manufactured product for defects and then send that inspection data to edge devices running algorithms that identify whether a product contains a defect, determine the type of defect, select the best way to remedy the problem and then finally direct the manufacturing systems to take corrective action.
Edge enables those machines to rapidly make intelligent decisions on their own, Miklovic said.
He cited self-braking cars as intelligence on the edge in action. According to Miklovic, AI embedded in the vehicle's edge computing devices processes locally generated data -- whether there's an object in the vehicle's path, whether it's stationary or moving, the vehicle's own speed of travel, whether it's decelerating and, if so, at what rate -- and then uses the analyzed information to decide when to stop the vehicle.
AI embedded in edge devices in industrial and consumer settings works similarly, Miklovic said, adding that the intelligent system would learn over time to become increasingly more efficient and effective in analyzing data to determine which actions to take.
For instance, an AI-based home security system can learn to distinguish between a homeowner and a stranger setting off a home motion detector, while an AI-based monitoring system on heavy equipment can learn to distinguish between temporary stress on the system that requires no action and stress that could lead to catastrophic failure if the system isn't immediately shut down.
The decentralized nature of edge computing has significant implications for enterprise cybersecurity, as the concentration of both data and compute power shifts from its centralized location in the core data center -- whether that's a public or private cloud -- to nodes throughout the organization's entire network.
That shift requires IT security leaders to broaden their defenses in a corresponding fashion to ensure that enterprise defenses extend from central data centers through the network to edge devices and the endpoints themselves.
However, edge computing does bring some benefits to the security program, said Yannis Kalfoglou, AI and blockchain expert with PA Consulting.
According to Kalfoglou, edge computing enables increased security and resiliency because its decentralized nature eliminates a single central point of failure. Security teams can cut off endpoints and edge computing devices that are attacked, breached or compromised in any way.
"In an edge paradigm, you just cut off one branch -- you don't lose the whole tree," he said. "Breaches will still happen, but if there's a breach, you can mitigate it by cutting off the edge device or devices."
Kalfoglou added that organizations might further boost their security posture as they adopt a true edge paradigm, where they keep more data on the edge and only send limited information back to the core. By doing so, they are limiting the potential value of successfully hacking into the central data centers.
"In a distributed environment, there's no such thing as a honeycomb," he said, noting that organizations can keep data securely on the edge through encryption, hashing and other measures -- a move that can also help organizations meet data privacy regulations requiring certain data to stay within specific geographic locales.
However, organizations that want to reap such security improvements with their edge deployments must have a strong data governance program in place, one that identifies what data is being generated and where that data should be processed, transmitted and stored. But Kalfoglou said such work can bring dividends.
"I actually see security as one of the motivations to move into edge devices," he added.
4. Edge computing and the cloud
Organizations are already keeping more data at the edge, and that will only increase as they devise more use cases where they can use the technology.
"In technology, we see these pendulums swings. Over the past 10 years, we swung toward cloud, but now, we're swinging back toward edge," Tung said.
But the edge computing trend doesn't eliminate the need for cloud computing -- experts expect organizations will continue to rely on public and private clouds for some capabilities.
"No edge device will be fully autonomous without having a periodic check-in with the cloud," Kalfoglou said.
Organizations will likely architect their IT environment to use cloud resources for heavy-duty applications, while using edge computing for lightweight applications. It will contain mesh networks with nodes on the edge connecting to each other, sharing information and instructions that, when combined with AI and machine learning, as well as more automation, will power autonomous work processes.
"Edge computing and cloud have a symbiotic relationship. While they both address different problem sets, they complement each other nicely," said David Williams, managing principal at Ahead, a provider of enterprise cloud solutions.
"We are beginning to see solutions that have both edge and cloud components -- loosely coupled -- where edge environment leverages the strengths of the other," he added. "It's also important that they both have autonomy such that edge and cloud applications and services can run and operate independently without tight dependencies. We anticipate that cloud and edge computing will continue to evolve together, with practical applications blurring the line between the two."
What the future holds for edge computing
More products specifically designed for edge computing will reach the market in the next few years, in addition to the products already available for edge computing deployments -- the networking, storage and compute appliances -- that were derived from typical data center and cloud infrastructure.
In its August 2019 report, research firm MarketsandMarkets projected that the global edge computing market will grow from $2.8 billion in 2019 to $9 billion by 2024.
Vendors are bringing to market more form factors set up specifically for edge computing, including processors that can better withstand environmental elements, such as high temperatures and vibrations, said Abhijit Sunil, a Forrester analyst serving infrastructure and operations professionals.
Experts also expect municipal governments to make investments into edge computing as part of smart city initiatives, smart vehicles to use edge for automated functions and private sector companies of all kinds to pilot projects. "I can't think of any industries that shouldn't be thinking about edge computing -- it's just some will be moving faster than others," said Proteus Duxbury, transformation expert at PA Consulting. "The next two years we'll be in this try-it-out mode."