Edge computing and 5G give business apps a boost
Edge computing requires specialized software with a capable, flexible controller platform to ensure that data is processed as quickly as possible. The following article explains why and how edge computing is likely to be embraced by most enterprises in the years to come.
Over the past decade, advances in cloud computing have driven a centralized approach to system administration and operations, while the growth of mobile computing, SaaS, and the internet of things (IoT) have driven computing toward a distributed architecture. With the rollout of 5G and edge computing technologies, companies are now looking to take advantage of both approaches while boosting performance for their applications.
While much of the hype around 5G and edge tend to focus on innovative, cutting-edge applications in areas such as robotics, augmented or virtual reality (AR/VR), and autonomous vehicles, experts say the benefits of edge computing go beyond these apps to provide IT professionals an array of opportunities.
How edge computing tackles latency
Enterprises have benefited from cloud computing during the past decade by centralizing resources at data centers owned by cloud providers – saving money on management costs and avoiding capital expenditures needed for internal data centers. But centralization has led to performance issues when dealing with endpoints on the internet’s “edge,” such as IoT devices/sensors and mobile devices.
While today’s smartphones are essentially intelligent computers that fit in your pocket, they still require a vast amount of processing done in the cloud. “Why can’t you put all the intelligence at the end? In other words, why can’t your smartphone just do it?” Asked Mahadev Satyanarayanan, a professor of computer science at Carnegie Mellon University.
“The answer is to do the kind of compute that you want done, you need far more computing resources than you would carry with you on your smartphone,” he said. “If you think about the video camera on your smartphone, it’s extremely light. But if you wanted to do real-time video analytics on it, you couldn’t do it with the computer on the phone today – you would ship [the data] to the cloud, and that’s where the problem begins.”
The solution, as outlined in an influential 2009 IEEE Pervasive Computing article co-authored by Satyanarayanan, is to use virtual machine-based “cloudlets” in mobile computing – in other words, placing mini data centers at the network’s edge close to where their processing power is needed.
On average, Satyanarayanan explained, the round-trip time between a smartphone and cell tower is about 12 to 15 milliseconds over a 4G LTE network, and can be longer depending on legacy systems and other factors. However, when you ping the data center from your smartphone, this could take anywhere between 100 milliseconds to 500 milliseconds, even up to a full second in some cases. Satyanarayanan calls this lag the “tail of distribution,” which is problematic for low-latency applications.
“Human users in applications like augmented reality are extremely sensitive to the tail,” Satyanarayanan said. “If I give you half an hour of an augmented reality experience, you may have 25 minutes of a superb experience. But what you will remember is five minutes of a horrible experience.”
Reducing the tail of distribution down to the edge is what makes edge computing appealing.
The 5G connection
The concept of moving intelligence to the edge didn’t really catch on until three or four years ago, when telecommunications companies began making plans for 5G wireless – and realized that 5G’s speeds only help in the last mile.
Remember that data travel time to and from a cell tower of 12 to 15 milliseconds over 4G? With 5G, vendors are touting latency levels of just 2 to 3 milliseconds – but the trip to and from a distant data center can still take 100 to 500 milliseconds or more. “If you have to go all the way back to a data center across the country or other end of the world, what difference does it make, even if it’s zero milliseconds on the last hop?” Satyanarayanan said.
Dave McCarthy, Research Director for Edge Strategies at IDC, agreed.
“By itself, 5G reduces the network latency between the endpoint and the mobile tower, but it does not address the distance to a data center, which can be problematic for latency-sensitive applications,” he said. “By deploying edge computing into the 5G network, it minimizes this physical distance, greatly improving response times.”
That makes edge computing crucial for the rollout of 5G networks and new mobile edge computing (MEC) services, he added.
Experts say it’s important to realize that edge computing and 5G are not connected at the hip. While 5G networks absolutely require edge computing technologies in order to succeed, edge computing can operate on different networks, including 4G LTE, Wi-Fi, and other network types.
How edge and 5G can boost business apps
When you combine the speed of 5G with edge computing’s processing capabilities, it’s only natural to focus on applications that require low latency. This is why early use cases tend to involve AR/VR, artificial intelligence, and robotics, which require split-second decisions from computing resources. But there’s potential for a variety business apps to benefit from both edge and 5G.
“In on-premises edge, there are many applications that already exist which could potentially be ‘moved’ or leverage a mobile edge compute,” said Dalia Adib, principal consultant and practice lead for edge computing at STL Partners. “There is a sweet spot of use cases – for example, those that use video, IoT, and AI.”
Experts cite a range of use cases for edge computing in the enterprise, including:
- Businesses with capital-intensive assets in industries such as manufacturing, oil and gas, and energy using 5G and edge for maintenance and repair activities. This includes AR/VR apps to guide technicians through repair, as well as drones for visual inspections of rail lines, bridges, or buildings using advanced analytics to identify potential defects or items in need of maintenance.
- Real-time process optimization in manufacturing facilities. Data generated from smart, connected equipment can dynamically adjust calibration settings, increasing yield and reducing defects.
- Condition-based monitoring – using IoT sensors to check certain parameters on an asset or machine to ensure it’s working properly.
- Video analytics for surveillance, such as using real-time processing to determine whether a person entering a building is an employee or a visitor and to confirm the identity of employees.
- Video analytics to provide real-time advice for law enforcement decision-makers in emergency situations. (See this clip from 60 Minutes discussing wearable cognitive assistants.)
- Telehealth applications in healthcare – using video and analytics to diagnose a patient, or to conduct remote patient monitoring.
Satyanarayanan foresees the development of edge-native applications that are built to take advantage of edge computing’s strengths, such as low latency and bandwidth scalability. These apps will likely drive demand for 5G networks and edge computing growth, he said.
“Edge-native applications that augment human cognition are potential killer apps for edge computing,” he and his co-authors wrote in a 2019 article, The Seminal Role of Edge-Native Applications. “These apps improve some aspect of human cognition (e.g., task performance, long-term memory, face recognition, etc.) in real time. By leveraging edge computing, the computing resources that can be brought to bear in this task can be far larger, heavier, more energy-hungry and more heat-dissipative than could ever be carried or worn by a human user.”
Further business benefits for edge
Beyond the benefits of low latency, experts said edge computing can provide businesses advantages including bandwidth cost savings, better privacy options and regulatory compliance, and support for situations when network connectivity is inconsistent.
On the bandwidth front, IoT device can process their data on the edge, and then send only essential data back to cloud servers. Consider the bandwidth saved by not sending the data from, say, 100 video cameras covering a building or airport to a central cloud server for facial recognition or other real-time analysis.
Data privacy is another benefit. Storing and processing data at the edge keeps it from being sent to a distant cloud server in a data stream from which personal information could be extracted via machine learning algorithms, Satyanarayanan said.
What’s more, “in some instances, edge computing is a method of achieving compliance with government or industry regulations,” said IDC’s McCarthy. “For example, GDPR in Europe dictates data sovereignty requirements, which limits where data can be transferred and stored. Edge computing gives enterprises more control over where applications are deployed.”
Edge computing also benefits companies whose workers need to use mobile apps in situations where network connectivity is inconsistent.
“This is common in industries where the end point moves in and out of coverage areas, like transportation, mining, and agriculture,” said McCarthy. “By running application logic locally, functionality can be persistent, and the resulting information is uploaded to the cloud or other data center at a later time.”
Additional examples include disruptions that follow a natural disaster, or for military applications where an enemy would take out an internet connection to disrupt communications.
Finally, the flexibility and scalability of edge computing can be a key benefit for enterprises looking to move computing resources off centralized dedicated appliances. “This is what is driving the move towards edge in industries such as manufacturing; logistics and warehouses; retail; and oil, gas and mining,” said STL Partners’ Adib.
Enter the pandemic
Despite all these advantages, businesses may still see 5G and edge computing as on the fringe. But another, more immediate use for edge computing is emerging: it can help support employees who find themselves working at home as a result of COVID-19 lockdowns or work-at-home orders.
Many enterprises continue to use legacy applications or proprietary, customized software that requires the use of virtual desktop infrastructure (VDI) to operate – and many of these solutions need employees to be nearby.
“That doesn’t work very well when you work from home. With VDI infrastructure, you need extremely low latency because you are sending your keystrokes and mouse movements to basically a remote desktop,” Satyanarayanan explained. “Edge computing opens the door to what I call EdgeVDI, where you move the virtual machine from the private data center inside the enterprise to a device at the edge. You will notice that as a result of COVID-19, there’s a huge amount of growth of the VDI business, precisely for this reason.”
In response to changing work patterns, content delivery networks (CDNs) are embracing edge computing, according to IDC’s McCarthy.
“When more employees were in offices, those buildings were connected to the internet with enterprise-grade technology,” he said. “Now, with the shift to work-at-home, cable companies and other multiple-system operators are handling more of the load. A good edge strategy can give businesses the flexibility to move workloads between different types of infrastructure to better serve their employees and customers than a centralized approach can offer.”