A discussion on how the edge is all about generating new value from data and accessing more of that data by getting closer to where it’s created.
It’s a perfect storm for edge computing. There are new connectivity services like 5G, a move to loosely coupled and distributed application architectures, and small, durable, inexpensive, and high-performance compute devices. Combined, these elements are ideal for creating the innovative data-driven edge applications businesses require today.
RTInsights recently sat down with Bill Pfeifer, Edge Messaging Director at Dell Technologies, to talk about what’s driving interest in edge, use cases, and how new technologies are helping advance the field. Here is a summary of our conversation.
RTInsights: Why all the attention on edge computing and why now?
Pfeifer: The edge is where data is acted on near its point of creation to generate immediate essential value. So, success at the edge is all about the new value that we can generate from data and accessing more of that data by getting closer to where it’s created.
Data centers and clouds, where we keep our technology today, came about because data was relatively small and easy to move, while compute was big, expensive, and fragile. We protected it in these centralized deployments where we could give it all the care and feeding that it needs.
But now we’ve got compute that’s small, durable, inexpensive, and easy to install. And we’ve got data that’s very large or generated very fast and whose value declines very fast – often less than a second. There are privacy and security concerns. There may be autonomy concerns. If you lose your internet connection, can you still continue to function?
So, it makes more sense to decentralize our compute and move it to where the data is at the edge. We’re just rebalancing the technology so that it’s not all clustered in those centralized repositories.
Why do people care about it now? I’ve heard edge called the oldest emerging market on the planet because it’s been around for decades. Factories have had automation and tracking systems. But with current advances, edge is growing, evolving, advancing faster than it ever has before.
It’s like a whole new market that’s already well-established, which is an interesting dichotomy. It’s changing so fast it’s like new, but it’s already well-established, so people understand it. Based on analyst feedback that we’ve gotten, we expect to see most of the growth in deployments happening at the edge for the next couple of years. I’ve seen stats that suggest 90% of new business processes will be at the edge within the next few years.
Existing apps and processes will still stay in their data centers and clouds, where they run fine. There’s no need to move them. But those new efforts, the projects that will advance businesses moving forward, are almost entirely happening at the edge. That’s driving a lot of excitement and a lot of interest.
RTInsights: What new and emerging technologies are enabling edge computing?
Pfeifer: There are a number of things that are all coming together really nicely. They include advances in AI and streaming analytics that let us solve problems that we couldn’t solve before. That’s how we monetize the edge. Why make the investment to move that compute out to the data? Because you can do cool and interesting things with the data.
We’ve also got advances in compute that make it easy and reasonable to embed everywhere. Compute has gotten small, cheap, durable, and much more straightforward to use so we can push it out to the edge. We’ve got this whole move toward cloud-native containerization, serverless, DevOps, and infrastructure as code. All of that supports more dynamic environments. In addition to making data centers and clouds more flexible and responsive, it also lets you move your workloads dynamically from place to place, which is critical for the edge.
You don’t want to have more of a traditional monolithic installation plan for thousands of edge sites. That’s just not going to scale well enough. You can’t scale it up. And you can’t scale it down. The efficiencies that we see from containers and serverless provide a massive scale boost for existing compute deployments.
Another technology enabling edge is 5G. We often talk about that like it’s just a faster version of 4G, but 5G is the first generation of cellular connectivity built to handle both consumer and enterprise traffic. If you look back, 1, 2, and 3G cellular were about connecting people to people. 4G was about connecting people-to-people and people-to-systems. The latter includes things like ride-sharing and shopping apps.
5G adds system-to-system connections. So, we have IoT and AI systems working together and AI systems at the edge working with cloud or data center systems. It’s going to be a game-changer for the edge just having 5G available once we start getting into what that machine-to-machine connectivity can do for us.
There are also things like blockchain and other enablers that are interesting technologies that let us solve interesting problems.
But we also see the rise of as-a-service and consumption-based purchase options as a driver for edge adoption. When we talk about the edge, it’s tens, hundreds, thousands of sites with several devices per site. If you think about how a typical enterprise would afford the CapEx purchase of all that and scale their existing IT staff to install, manage, and troubleshooting those devices, that would slow down edge. It would be much less prevalent, much less interesting without that ability to just have hardware show up on-site ready to go, where the customer can just operate it. They don’t have to manage, install, and troubleshoot it.
So, there’s a whole raft of things that have come together, and they’re providing this great synergy that I think we’ll start to see blending at the edge very soon.
RTInsights: What are some common applications of edge computing?
Pfeifer: One of the most interesting things that I’ve seen with edge computing is that it’s highly verticalized. There are vertical-based industry solutions. We’ve always wanted that with PC, data center, and cloud apps, but they’re fairly horizontal. You do roughly the same things in terms of how you manage and operate them.
But from edge, you’re embedding something into the core of your business, and what that is depends hugely on the type of company you are. If you’re in manufacturing, it’s out on your floor, which can be a harsh environment. Maybe it’s a clean environment, depends on the manufacturer. If you’re in retail, it’s on your sales floor, which is a relatively clean, relatively structured environment but can be noise-sensitive. For farming, you’re out in the fields, definitely a harsh environment. Devices are exposed to the weather. They are rained on, overheated, and under heated.
It’s really interesting to see how much difference we see between industries, even within an industry like manufacturing. The main focus areas of manufacturing edge applications are to decrease the cost of goods sold, increase overall equipment effectiveness, and increase worker safety.
For instance, if you’re using robots, you can have computer vision, making sure that people aren’t walking into a space where a robot is swinging around heavy objects. Or visually making sure your employees are properly wearing their safety gear. Or we’re seeing lots of quality assurance-type things. An example is computer vision watching parts coming down the assembly line to ensure that there aren’t defects visible on those parts and automatically and almost instantly identifying where there are problems. Once you have systems like that in place, you can start to identify if there’s an increasing number of those problems. You can track it over time to say, “Hey, this equipment needs to be serviced.” So, you can start to get much more intelligence about your operations.
Within retail, many of the applications are for theft reduction, watching foot traffic, and watching behavior. Where do people linger? Do they need help? You can identify what sorts of behaviors you see across your sales floor so that you can start to customize interactions.
There are theme parks that are adjusting the music based on the weather and the time of day. And they adjust what stock they have out and available in their small retail shops based on what people are likely to want to buy. In that way, they are not wasting shelf space or running out of merchandise.
In farming, we’re seeing many applications that try to minimize what chemicals you have to use by identifying common types of rot and diseases of plants and dryness and things like that. They might use sensors in the soil and computer vision to look at plants to detect conditions before problems can spread across a field. Once you have that intelligence, you can start to spray just exactly what you need, where you need it. So you’re not applying massive doses. You can identify specifically who needs what, and that ends up saving money and reducing the chemicals that are in all of those products.
RTInsights: Where do Dell Technologies and Intel fit in, and how do you help organizations with their edge computing projects?
Pfeifer: What we’re seeing out at the edge today is that our customers are coming up with lots of interesting use cases. They test out an idea and deploy that solution at scale across production. Then they find another use case, and they solve it. They deploy that at scale across production.
Often, those solutions aren’t intended to work together. They’re not planned. Maybe they’re solved in different business units, by different vendors, or across different cloud providers who don’t like to play well together. We’ve seen cases where customers have 10, 20, 30, or more individual devices solving individual use cases at each of their edge sites.
Dell Technologies has been focused for years on modernizing, consolidating, simplifying our customer’s data centers, working with our customers to simplify their data centers and streamline their operations so they can get more out of the technology they have. We’re doing that at the edge as well.
We’re starting to look across those many use cases to find where we can consolidate hardware, data stacks, operational processes, and services and make that all one big functional edge that’s easier to manage and scale.
We’re also seeing two main types of edge deployments. There are the harsh environments and the controlled environments. For harsh environments, we have our OEM business, which has been making custom hardware based on specific demands for the military, industrial purposes, hardware resellers, and software resellers that need a specific piece of hardware in a specific place.
For that, we’ve got a multi-billion-dollar business that’s been around for about 20 years. Dell Technologies is the number one OEM business in the world today. In controlled environments, we often end up at the edge in small, or mini, or micro data centers.
We’re seeing deployments with something on the order of three to five servers per edge location. Once things scale up a little bit, that’s a tiny little data center. Sometimes it’s just a hardened enclosure to help keep the devices safe or a small rack. So that’s just a small decentralized data center deployment.
Now, Dell Technologies has been building data centers forever, including big data centers, small data centers, centralized, distributed, all of those different permutations of data centers. We’re very good at that and helping our customers scale those data center deployments as much as they need to.
We’re great at handling harsh environments. We’re great at helping our customers simplify and consolidate their deployments. We’re great at building both centralized and distributed deployments, and it’s particularly exciting to see the edge grow into the thing that we’ve been aiming at for so long.
RTInsights: With edge, it seems like there are so many players involved. Are people looking for something like a Dell Technologies that offers something across all the environments, including cloud, data center, and edge?
Pfeifer: It depends where they are with their edge journey. We’ve heard from customers that are running multiple factories. You can imagine that’s a relatively harsh environment, space-constrained, and power-constrained. They don’t want to keep running more and more and more gear there. Otherwise, it becomes a giant data center, not a factory. That’s not what the property is designed to do.
They’re using edge solutions to monitor for safety and run their robots and production facilities. Each thing solves one particular use case. It is typically brought in by a vendor and managed by the vendor. There’s the massive proliferation, and they’re starting to ask how do we make these operations more consistent so that we can reduce the hardware? Can we reduce the number of devices we have? Can we build in better redundancy and security?
That’s playing into our strengths where some of the newer customers who are just starting to explore the edge are picking their first use case. They’re not seeing the potential for sprawl. They’re looking at one use case. They haven’t scaled it out yet. They’re just testing. They will grow into that conversation about sprawl and the number of different groups playing at the edge in very different ways.
We have seen customers that have many different vendors out at the edge that don’t play well together. We can help consolidate those operations in many different cases in a number of different ways so that you don’t have these massive silos and still work across multiple clouds.
And overall, whether those customers are just getting started at the edge or have well-established operations at the edge, Dell Technologies stands ready to help our customers simplify their edge so they can generate more business value where it matters most. We can help them grow their business, serve their customers in new and different ways, and get their business ready for the future – whatever comes next!