5G and edge computing hold tremendous promise for delivering ubiquitous connectivity and new applications. Given this, it’s no surprise that 85% of technology leaders interviewed in the Lopez Research benchmark survey said they will be developing and updating an edge strategy in 2022. With a myriad of wireless networking technologies entering the landscape, it’s never been a better time to upgrade network technology, nor a confusing time.
While there's no one size fits approach to designing networks, companies have access to a range of solutions that provide the ability to create more flexible, adaptive networks. Leading organizations will build wireless strategies to match the right computing type to the right workload in the correct location. The adoption of multiple wireless solutions will create an enhanced distributed computing landscape.
To understand what enterprise buyers should know about the new world of wireless communications and computing, I interviewed Asha Keddy, the Corporate Vice President, General Manager at Intel Corporation, for a Fierce Wireless webinar. We covered many topics, but several themes stood out for me. The first topic was a discussion about the move to distributed intelligence.
The Rise of Distributed Intelligence
Just as there are various cloud computing models, there are also multiple types of distributed computing, with edge computing among them. Edge computing, one of the more recently developed distributed compute architectures, is essentially the organized amalgamation of multiple disciplines, including cloud computing, pervasive connectivity, mobility, the IoT and powerful analytics. Edge computing, powered by 5G and advanced Wi-Fi, provides resources such as compute, storage, and networking closer to where the data gets created, enabling a host of new services and applications that would not be possible if that data had to travel to a data center.
Keddy called this a new world of distributed intelligence. She spoke of how distributed intelligence uses artificial intelligence (AI) to deliver actionable information by analyzing data closer to its creation. Intelligent systems can understand workload requirements and match network capabilities to these requirements. It’s intelligent because it focuses on the analytics and the outcome. Keddy said,“There's more emphasis on personalization and customization for verticals and specific use cases. For example, email can always remain in the cloud, but for a factory use case, you may need to analyze data on-site and act on it during that moment.”
Five years ago, we couldn't say that we had the right technology to build distributed, intelligent computing. Today, almost every new device is equipped with embedded sensors, processors, and wireless connectivity. These connected products allow companies to gather and analyze data in real-time. For example, automakers such as Audi and BMW are building connected factories and appliance companies like Whirlpool and Bosch are selling dishwashers with sensors and connectivity. These industries are delivering on the connected intelligence concept that Keddy described.
Distributed, intelligent computing enables new use cases
Next, we discussed the need for reliable, ubiquitous high-speed networks to support hybrid work and transform businesses regardless of industry. The industry often discusses the capabilities of wireless technology within various spectrum bands. Keddy says the industry needs to focus on providing enterprises with the freedom of choice to know how and where to use the data.
She provided examples of setting policies like service level agreements (SLAs) for privacy and security and defining what data stays on the premises. In the past, we had high-speed connectivity, but many use cases such as traffic management, visual inspection in manufacturing and remote equipment controls require ultra-reliability and low latency. Network slicing will also provide the ability to customize functions such as bandwidth and latency to address various workloads.
Keddy said technology providers must focus on making wireless networks scalable and cost-effective. Keddy also spoke about the importance of software and AI accelerators that support the developer community by abstracting complexity. She noted that Intel provides software-defined, virtualized solutions to help developers create a world of distributed intelligence in addition to the hardware. In the near future, Keddy described the need to move toward automated and even autonomous systems using locally licensed spectrum or working with a communications or cloud service provider.
We also discussed the role of deterministic networks in creating these new responsive systems. Wikipedia defines Deterministic Networking as an effort by the IETF DetNet Working Group to study the implementation of deterministic data paths for real-time applications with extremely low data loss rates, packet delay variation, and bounded latency, such as audio and video streaming, industrial automation, and vehicle control. Keddy spoke about determinism as a different way of looking at SLAs for particular use cases or requests to be fulfilled. For example, a robotic arm may have real-time determinist requirements with sub-milliseconds latency, which means the network must be at the edge.
It also means that less latency-sensitive applications, such as email, run in a centralized cloud computing facility. Soon, businesses will specify the service level agreements they need at a specific cost. A service provider will fulfill that network request using the concepts such as network slicing.
IT, OT and communications networks should become one.
Breaking down internal network siloes offers another opportunity for an enterprise to improve its services. Keddy said, “The way we work together in an enterprise also has to change. We can't have like an IT department separate from an OT (operational technology) department and the communications team.” She described how certain businesses used to have three kinds of networks (OT, IT and Communications), but combining them will provide better business outcomes.
I agree. For years, the industry has discussed IT and OT coming together to build better operational systems. However, it's been separate for a long time, and this change won’t happen overnight. Many companies are asking questions about how to create these new networks? Who makes the buying decisions, and how do you converge? We still need to address these challenges in the market to deliver fluid, integrated systems. The good news? Leading companies are reimagining business processes that will lay the foundation for OT and IT to come together because they are now building new workflows. They're building new concepts of how they want to run and operate the business.
Making distributed computing actionable
In distributed computing, computing lives around you. It’s virtualized, programmable, and supports modern application development with cloud-resident and cloud-aware development tools. With ubiquitous computing availability, companies have options to match the network to the workload’s requirements. Keddy believes the industry needs to stop talking about speeds and feeds such as bits per second or frequency per megahertz. Instead, technology vendors and companies need to define the transformation that AI and distributed computing will deliver. She said leaders, such as Audi, started by solving specific problems by using a predictive quality-control solution to boost weld inspections by 100 times.
She sees enterprises moving from a world of generic proof of concepts to define what’s possible to drive more practical and cost-oriented use cases like supply chain savings. Another example she provided was the concept of using augmented and virtual reality in a warehouse or the ability to create better parking solutions. Keddy described the significant innovations that are happening in robotics across the globe. For example, wireless and AI have enabled Japan and Korea to create companion robots to help offload tasks for an aging population. One of the best developments is that this technology is becoming easier to deploy and manage for an enterprise. For example, Amazon Web Services announced the availability of Private 5G as a service during its Re: Invent conference. Meanwhile, Bosch and others are looking to provide more customized, sector or use-case-specific solutions. There will be public 5G solutions with network slicing, private wireless solutions with licensed and unlicensed spectrum, and a range of solutions in-between.
Conclusion
After discussing intelligent systems for decades, we finally have access to the foundational technology to create innovative, distributed computing systems. I’ve learned from the event that there is a wide range of options, and enterprises should take a portfolio approach to create the best network services with suitable characteristics at the right price.
This article was written by Maribel Lopez of Lopez Research in collaboration with Intel Corporation.
Lopez Research tracks trends for a wide range of technology topics including artificial intelligence, cloud computing, connected devices and A.I. Its clients range from start-ups to global firms, including 10 of the Fortune 30.