Although the technology is roughly similar to datacenter or cloud, the unique challenges of edge computing require new approaches to storage, networking, orchestration, deployment, and more. We were kicking off a new season of Utilizing Tech focused on edge computing, featuring Alastair Cooke and Brian Chambers as co-hosts along with Stephen Foskett. One of the key differences for edge computing is the strictly constrained resources there as well as the massive scale. What we now call edge has existed for decades but the proprietary hardware previously used has been largely replaced by commodity and internet-connected systems that inherit technologies like virtualization, containerization, and hyperconvergence. Businesses and consumers expect more interactivity and capability in retail, restaurants, and hospitality environments, and the IoT revolution is strongly affecting manufacturing, military, and industrial settings. But what is the edge really, and how do we deliver services there? That’s the question we hope to answer in this season of Utilizing Edge.
Hosts:
Stephen Foskett, Organizer of the Tech Field Day Event Series, part of The Futurum Group. Find Stephen’s writing at GestaltIT.com, on Twitter at @SFoskett, or on Mastodon at @[email protected].
Alastair Cooke, independent analyst and consultant working with virtualization and data center technologies. Connect with Alastair on LinkedIn and Twitter. Read his articles on his website.
Brian Chambers, Technologist and Chief Architect at Chick-fil-A. Connect with Brian on LinkedIn and Twitter. Read his blog on Substack.
Follow the podcast on Twitter at @UtilizingTech, on Mastodon at @[email protected], or watch the video version on the Gestalt IT YouTube channel.
Episode Transcript
Stephen Foskett: Welcome to Utilizing Tech, the podcast about emerging technology from Gestalt IT. This season of Utilizing Tech focuses on edge computing, which demands a new approach to computing, storage, networking, and more. I’m your host, Stephen Foskett, organizer, tech field, and publisher of Gestalt IT. Joining me today to kick off the season of Utilizing Tech are my co-hosts, Brian Chambers and Alastair Cooke.
Brian Chambers: I’m Brian Chambers. I am the leader of the Enterprise Architecture Group at Chick-fil-A, and we are known for live edge compute work that was done on site at retail locations.
Alastair Cooke: I’m Alastair Cooke, and I write about technology. You can follow me as Demitasse NZ, and the NZ is for New Zealand, where I live. You can also find me at Demitasse.NZ.
Stephen Foskett: And I’m Stephen Foskett, publisher of Gestalt IT. You can find me on the weekly Gestalt IT On-Premise IT Podcast, as well as our weekly Rundown news show, and of course, right here on Utilizing Tech.
Stephen Foskett: Thank you both for joining me this week to kick off the season of Utilizing Tech focused on edge. We recently did our first-ever Edge Field Day event, and both of you were part of that event. It was a lot of fun and really interesting. It also opened up our eyes, I think, about this whole new world of edge. From my perspective, the reason I wanted to do this podcast season on this topic is that my background was in data center IT, regular traditional enterprise IT, which has verticals like networking, storage, and compute. Then I got really involved in enterprise cloud and hyperscaler cloud, which also has compute, storage, and networking, and so on. All of those things are similar but different; they kind of rhyme. The thing that kills me about edge computing is that it’s the same deal. Once again, we’ve got a new environment, we’ve got the same verticals, and in many cases, the same technology. Sometimes even literally the same devices. But just like cloud is different from data center, edge is fundamentally different from cloud and data center as well. And for me, that was the thing that really got me. What do you think of the edge as a different, I guess, horizontal approach to IT?
Alastair Cooke: I think this is it. So, there’s a lot of similarities between enterprise IT and cloud IT and edge IT, and the fun part is finding the differences. So, what are the things that you always do on cloud that just don’t make sense on edge, and what are the things that you might always do in enterprise that again don’t make sense on edge? The unique things you have to do differently on edge, and that’s the new technology understanding, its frameworks, and how it works. How is it different from what you already understand is always interesting to me. I always like to see how you approach solving different problems. There are fundamentally a whole new set of problems at the edge that maybe don’t exist at the village and sometimes don’t.
Brian Chambers: The other word Alastair is saying, I think it’ll be interesting to explore how perhaps edge is maybe a little bit like the data center to cloud-like hybrid model. There are certain reasons you had to keep certain types of workloads on-premise originally, and then people started to move some out, but had interesting interactions where maybe part of an application or experience was in the cloud, and another part was still anchored in a data center.
Stephen Foskett: One of the things that really interests me is that building an application that’s fit for purpose is more important than having some crazy new technology or amazing insight. The unique challenges and constraints at the edge fundamentally transform how technologies are designed, deployed, and built. While it’s increasingly the same stuff that’s used in data centers and clouds, it is still fundamentally different due to the environment.
Brian Chambers: I completely agree. A lot of the technology is the same, but it’s with a new set of constraints. The scale is much more finite at the edge, and we don’t get the benefit of some of the paradigms that people have gotten used to when building applications in the cloud. We need to consider the constraints of the infrastructure, connectivity, and other things that are going to work in the real world.
Alastair Cooke: Legacy applications are also a challenge at the edge, and we need to bring them along while building new applications using containers. Seeing solutions that encompass this idea will be an interesting part of exploring the edge.
Brian Chambers: I agree. One advantage we had in pulling an edge architecture is that everything we used was Greenfield. We were able to start from scratch, build things in containers, and run them on K3s, a lightweight version of Kubernetes. We had the benefit of rebuilding both the infrastructure and all of the connectivity around our solution within the constraints we had in a restaurant environment with unreliable internet connectivity. However, this is a huge advantage that not everyone in the real world will have when thinking about deploying at the edge.
Stephen Foskett: Yeah, and that’s actually a really good point. As I mentioned, I was working on, I guess, edge before it was called edge in a retail environment. So, probably 25 years ago, and frankly, it was horrible because it was just this big stack of proprietary weirdness. I was in a retail gas station, and we were deploying point-of-sale systems, inventory control and management, pricing, all sorts of things. Every one of those systems was bizarre and proprietary and had, for example, their own backhaul. We literally had multiple satellite dishes on top of the gas stations. We had, oh well, this one runs over a modem that connects with a phone line, and this other one uses this like Hughes giant satellite dish, and this one requires this specific piece of hardware, and this one requires a completely different hardware that’s actually literally the same thing, but it has to be deployed differently. And it was just us. This is total nightmare of technical debt and weird legacy. I hope and think that now that edge is a thing, I hope and think that that’s changing. Is that changing? Is there more standardization? And can companies hope to be like Chick-fil-A in that they have a Greenfield, and they’re able to write new applications that just sort of run?
Alastair Cooke: I’m sure there are companies who are in complete agreement. Is that rangefinder, if you’ve got an existing deployed environment, you’re typically going to be building on top of where it’s been deployed. Bron’s benefit was that he deployed the stuff out completely clean Greenfield environment. A little bit range on bus. Those organizations that hamster, they dedicated hardware in a custom hardware on all of us, and let links will be really enjoying a move towards commoditization towards shared hardware and shared links underneath. And probably even enjoying a move to use of the Internet and VPNs, rather than all of those satellite nations. And that’s notarization is why I was saying so much growth of the edge, because the cost, the shared cost of putting in and supporting all of these systems that Stephen experienced with your answer the places that you could put them into had to be much more profitable. And so what we’re saying is with monetization on the use of the Internet, that the options for where you can put in compute can drive north. Any way you can deliver better service through this huge compute, just expanding, expanding was driving always has a growth in each compute.
Brian Chambers: Yeah, to add to that, I think now with the emerging landscape, there will be an opportunity to fix some of the legacy problems. In the past, we had many different solutions for the same things on the same physical premise. The question is, will people be willing to bite off the work to re-architect? That’s a question every company will have to wrestle with. But I think the technology that is emerging at the edge and in the cloud is much more friendly to multi-tenant type solutions and has better security controls. But when we think back to restaurant environments, we were working with a computer system that was for a single purpose, proprietary, and not open to allow other things to run on it. It was just for a point-of-sale system in our case. Now we have an architecture that allows us to run any kind of application as long as it’s container-based. You can imagine other iterations of that with Micro VMs, WASM, or whatever else the case may be. Technology provides the opportunity for people to use a lot of these tools to orchestrate a new kind of platform that allows multi-tenancy with a lower cost of management, less complexity, and more interoperability benefits.
Stephen Foskett: Yeah, and I think the cat’s out of the bag on basically simple, complicated, proprietary solutions in these environments. What we were working on in the ’90s was basically what would now be considered hyper-focused advertisement. The idea was, “Well, this guy usually comes in and buys a Snickers bar, so I’m going to have a custom ad go to him.” In the ’90s, that was considered radical. Nowadays, I think that kind of personalization and customization doesn’t sound so radical. I think people expect environments to be more personalized, more customized, and there are going to be more metrics and consumer tracking happening. People also expect that there will be a lot more self-service actions happening, whether it’s retail or even an industrial situation. Therefore, we have to have commodity containerized or virtualized at least systems that can allow us to deploy this stuff. Otherwise, the application owners and the business itself wouldn’t function. Think about going to a store or a restaurant. You’re going to have a bunch of interactions and applications running in those environments that you might not have expected to encounter 10 years ago, and you can’t put that genie back in the bottle.
Alastair Cooke: It’s not stopping, so that range of experiences and services is going to continue to increase. What strikes me is that each of these edge locations is, in some ways, a micro cloud. It’s delivering a set of software-defined services that are being consumed by multiple application teams. So, there is a job function to be the manager of that micro cloud in a way that if you’re actually using, consuming from a public cloud, it’s somebody else’s problem. AWS runs the cloud, Microsoft runs their cloud well. But in terms of the developer experience of building those new micro applications to deliver at the edge locations, they just want to consume cloud-like IT resources at the edge. Therefore, there’s a real strong place for building that cloud platform at the edge. But as Brian says, that’s completely different scalability. It’s very limited scaling at that edge location.
Stephen Foskett: One of the things that occurs to me as well, and this is one of the questions that I’m gonna have for our guests all throughout the season is that, as I kind of alluded to a minute ago, the edge and the edge, in the edge, there’s different edges here. So, you know, like I said, my experience is in the gas station world and the retail world. Brian obviously knows quite a lot about the quick-service restaurant world. Certainly, there is a whole world of factory and industrial IOT. There are also things I was talking to a guy recently about mining, and in a literally under the ground and above-ground mines and the unique challenges of deploying devices in those kind of environments. We’ve talked to people about cruise ships, about military ships, even I know a guy who worked on satellites, and they approach that as an edge computing environment. There’s a whole world here. Is there really such a thing as edge, or is edge just what we’re calling a bunch of completely desperate things that have nothing to do with each other?
Alastair Cooke: Well, I think one of the challenges here is that, because the term edge is popular, it’s being applied to a lot of different things. I wrote about this just prior to Edge Field Day. I wrote about different types of edge, and when I categorize as the fire up so it’s a location that belongs to the organization but doesn’t have any IT stuff ever visiting it. But there’s also what are characterized as the nearest, which is a location that doesn’t belong to us but is a Donna center, so it might be that we’re using collocation all over the place. And this is sometimes what you see is the edge from the point of view of a telco or from the part of your stomach, make a comment, distribution, or gaming kind of use case where the objective is to get close to their retail subscriber, but that subscriber doesn’t access it over the network, so for me, that’s in the air inches, still highly connected, more scalable, probably looks like a data center, and so you can run data center structure. You can run hardware that needs cooling and these palpitations, whereas what was mostly been talking about at Edge Field Day, so what is what are characterized as the far edge. So there’s no power protection, there’s no environmentals, that you cannot put dollars into equipment in there because it’s simply won’t survive the amount of dust and dirt and that mine. So I don’t think there is a single thing that is the edge. I think we need to have a more sophisticated. If you are at one of the slides that was put up and it feels like I’m one of the vendors showed five different layers of edge. And so I think it is important that we recognize a little bit. The edge is not just a single entity, and that there are different types of infrastructure, different constraints of those different edge locations. I am personally I like that the fire rage. There are some really cool interesting things happening in region for such a mandatory into mobile locations delivery trucks into unattended sites. Those kinds of things are really interesting, but there’s still huge business in that marriage as well where it looks much more like a datacenter much more like a collocation site.
Brian Chambers: Yeah, I agree. There’s a large spectrum of what edge is, from high-end electric ranges down to potentially mobile devices. I’ve come across a lot of recent discussions about where Apple may go in the large language model game. Perhaps they have something that runs on their chips, which could be models living on mobile devices, executing business functions. That’s a type of edge that you’ve got, the floor edge. Then there’s the type of edge that you’d find in a restaurant, where you have local devices running your business stuff. There are also regional add points of presence, like CDN pops, where companies like Amazon are starting to run serverless functions. You can put applications in any of these places to give users the experience they want for your applications. Of course, there’s also the cloud on the other end. I think all of this is coming together to form something that we can’t really call “edge.” It’s more like a complete mess where you run things where they need to be, as close to the cloud as possible, but close enough to the user to meet the requirements of that particular app. When I see all the stuff going on, I think that’s where the industry seems to be headed. It’s going to be a mix of different types of edges that make that happen.
Alastair Cooke: One of the things you brought up that’s really significant is that all of this is about solving business problems. It’s not about an academic definition of what “edge” is. You have to understand that framework of all these different places at the edge can be. When does this product work? What is this product useful for? How does it help me to build out this application? Although the definition doesn’t help us to identify a business problem or solution, these characterizations help us work out what is the right tool to use. So they’re a good way of quickly working out which tool should I be using for this problem. I’m sure Brian doesn’t want to have a whole three-node rack-mount server environment going into each Chick-fil-A location. For some definitions, that is an edge computer. So it is useful to have these terms to make it simpler to identify what products can bring what benefits.
Brian Chambers: Do you guys think that the edge, outside of those more constrained environments, when it’s in the original data center to CDN or when it’s in a telco, like a fiber hub or something like that, do you think that paradigm is going to converge with the cloud? And the tooling and experience for application developers is gonna be the same? And then it gets narrower with the constraints to come to the question you get to the user, or something different with y’all’s opinion on that?
Stephen Foskett: Yeah, I definitely think it will, Brian, and I’m already seeing it. Basically, clouds are getting bigger and bigger and bigger. And for what it’s worth, I mean, I see that the sort of “what is the edge” and, you know, that’s for the academic, like, let’s define this word, kind of thing, as sort of pointless. Because it was just as pointless in the cloud era when we were all like, “Oh, that’s not cloud. This is cloud.” You know, absolutely whatever you want to call it. If you look at things like, for example, Cloudflare workers, or what Akamai is doing, or what’s happening in 5G with distributed applications and distributed containers or serverless, a lot of this stuff is leaking out of the traditional and traditional cloud. Imagine that such a thing as traditional power, we can out of the traditional cloud, and becoming more of, as you say, a mesh of distributed services that are closer and closer to users. So I absolutely think that whatever the cloud is, it’s going to go right out to the edge as far as it can, in a unified fashion, and that you’ll deploy applications that will use workers that happen to be running in regional data centers or in the telco Pop that also is, you know, the 5G, or yes, even running in a retail store or even in the consumer’s maybe in a consumer zone closet. You know, what I definitely think that’s all going to be more and more in a grade over time.
Brian Chambers: Yeah, to your point, where they’re seeing some of that with like it’ll be us wavelength, which is a service that allows you to put some very basic primitive Amazon compete services at 5G Verizon cell towers, but where we’re seeing that the cloud sort of expand further and further towards the user in this house away, so I think interesting pattern to learn more about them and keep an eye on.
Alastair Cooke: Yeah, and from my perspective, one of the things I do as a day job is teaching AWS training, horses, and so wavelength and outposts and local zones are always a delivering those St. Cloud native services closer and closer, and even if you start looking at things like on the latest snow, family the snow cone tiny thing this morning have to go out to why this one each location has Wi-Fi on on his local computer extending those same development paradigms that you haven’t plowed out 2-inch location. I think there’s gonna be a lot of different ways of solving these problems continuing Some more focused on extending the cloud night, Adela for experience, others more into towards an in-house developments or on-premises development, is extending out to the edge, and we’ll see quite a lot of different solutions sets.
Stephen Foskett: But given that, I guess, just like you know, we know the cloud when we see it, we know the edge when we see it. What are the characteristics of the edge that we know that means it’s the edge? Once I get some things that immediately come to mind, as Brian mentioned, is the unique constraints on terms of resources, of connectivity. We didn’t mention people, hands-on operations and things like that, so that’s one thing. Another thing is the scale and the distribution, not maybe which is different than scale, meaning, yeah, you might have 10,000 servers in the cloud, but they’re not in 10,000 locations, right? That’s a unique characteristic and you get a unique element. Another thing that I have that we’ve talked about is the unreliability, is the fundamental lack of anything approaching high availability in these environments. Am I off-target, do you think, are the characteristics that make edge, edge?
Alastair Cooke: I think the diverse range of locations is a really important part of the different organizations that means different scales. So if I’m gonna go to a hundred retail stores, my edge is gonna be a hundred retail stores. Where is if you’ve got 10,000 gas stations, it’s gonna be a little larger on I just definitely that idea that a Scally out some multiple physical locations and that it’s a cookie-cutter between those locations as well. I think that’s one of the important characteristics is that each other location is delivering a specific set of services that is usually likely consistent across those locations. It’s not that we’re delivering a general-purpose compete platform that can be used for something different in each of those 10,000 locations. We are templating it out and the point I’m not sure if that’s just a relation of the fact that we don’t have school hands at the site and so we have to do it from as a population with her. We’ll see over time more customization for each particular site that might be an interesting angle, but in terms of what is the edge, I think that the Stolas this idea that it’s on the outside of your data center or the public cloud. And those two meters is the big partier that makes a difference at the edge. The one part that we’ve discussed in the school that I think this is getting pretty modular on being there just worked on mobile device. I think mobile, the device in your hand is a general-purpose platform for computer and it’s not nearly as specific to a task is what I consider to be in infrastructure. Maybe that comes a little bit from commitment from that that’s on operational technology industrial automation, kind of side where it is very specialized equipment, and we’re trying to ask you so much more generalized way rather than having a general purpose to lock your phone and delivering it to it. So for me, the other things we talked about, and it doesn’t quite in jail for me where is I think Stephen I’m much more interested in the things that a data center near near edge than you are. I think it will be up pretty clear if your edge is much more of what I did I categorize as far edge.
Brian Chambers: Yeah, I think I generally agree with you on this, Stephen. I think you have most of the big things covered. One of the new wants is that, as it seems like most of the edge applications and use cases that I can think of at the moment, are more targeted in nature. So it’s not that you want to lift and shift your cloud computing workloads to the edge because that’s better for some reason. It’s that you have 2345 things, and leaving them in the cloud is insufficient. So, you need to find something that is closer to your user to meet their expectations or works when there’s some other constraint like it works too slow, it’s offline, or privacy requires it, whatever you’re trying to move for a reason. But it’s targeted towards a series of purposes, not just a place to put any near random workload you have. It’s not that we want to run the models of our ERP or cash management from the edge because it’ll be faster. That’s pretty trivial. But running things that are critical to your manufacturing probably makes a ton of sense. So, I think that’s probably the big thing that makes it adverse. Not the edge itself, but rather the slicing off of work that really requires something that you can’t get from the public cloud or your data center nature.
Stephen Foskett: Yeah, that actually came up in a discussion about utilizing AI a couple of seasons ago on this very same podcast, where we were talking about metrics collection and data processing in industrial settings specifically. One of the things that I thought was really awesome and mind-bending was this idea that if you can process data near sensors, then you are able to collect and process more data than you could otherwise. In other words, by moving things to the edge, by having low latency and compute power that’s not subject to the unreliability of networks and backhauls, it opens up new possibilities for smarter and faster applications. That is a hallmark of the edge, and it came up in the context of deploying machine learning tensor processing engines in factories, and doing live video and sensor processing at the edge. I think that’s really relevant to this discussion as well. One of the things I used to always say about cloud computing is that the fundamental nature of cloud computing is that some applications just don’t want to be there. They work better elsewhere, whatever that may be. Email is a classic example – it’s better in the cloud than it is locally. Similarly, many edge applications are better suited for the edge. It may be hard to wrap your head around it, but once you start thinking about it like that, you’ll realize that it definitely makes sense to do some things on-site instead of in the cloud or in the data center.
Alastair Cooke: I think it’s a corollary to Amdahl’s Law. So, Amdahl’s Law says the best I/O is the one you don’t do. In this case, it’s the best network transfer is the one that you don’t do. So, well, Brother Movies, I definitely see Edge as part of the function. The Edge is a doctor, we’re fine right. So, the Edge compute takes in lots of data from the site refined that works at Wadsworth sending back the head office, and I’m actually taking action on. So, somethings are acting very rapidly, and somethings you ain’t on over how much longer-term insights and things in Franklin rapidly in the process clothes. And then the things that a long time, you send why so long term insight. But what we do see if there’s a flow in the other direction as well, there’s that that’s been distributed out to be close to the use it. So, it’s not just that it’s a refinery for data this being generated at the edge. It’s also a place with presentation at those types that they’re not being generated at the edge, but I’ve been displayed at the edge, so this is definitely a two-way flow, which I haven’t thought through prior to Edge Field Day.
Stephen Foskett: I’m really looking forward to seeing where this goes. I think that we’ve hit on a bunch of things here during this discussion, and hopefully that this discussion will help those of you who have subscribed to previous episodes previous seasons of utilizing tech. You know if you were listening to utilizing AI, a lot of these things came up again if you were listening to utilizing CXL, you know a lot of these same discussions there’s so much similarities and yet so much different things different interests different areas. In the end, it really comes down to what you’re trying to do with technology more than the fundamental aspect of the technology. You have to pick the right solution the right tool for the job, and I think that’s what I’m gonna be looking forward to in this season of utilizing edge.
Stephen Foskett: Please do tune in. This is a weekly podcast. We’re gonna be publishing a new episode of utilizing edge every Monday starting now. This is episode zero. Episode one is coming. Oh, by promise, there will be a new one every Monday. We are recording aggressively to try to fill that quota, and also, if your appetite for edges up, please do check out Edge Field Day, the videos for matches tape Edge Field Day, and your favorite search engine, and you’ll find those videos because we talked to a lot of these companies. They did a lot of demos and salon. As well, I know that that that the two of you were doing some really interesting things to where can we follow your thoughts on the edge of Brian?
Brian Chambers: Yeah, for me, two places, it’s @BriChamb be part of my name on Twitter, and then I just started a substack fairly recently where there’s some discussion of edge pretty much every week when I write that. So, it’s called the Chamber of Tech Secrets, thanks to my LinkedIn audience for voting on that name, and went with it and found that it. Brian Chambers dot substack.com.
Alastair Cooke: My online presence is Demitasse, demitasse.co.nz. Or you can just put my name, Alastair Cooke, into Google, and so if you put something tech-related like VMware or AWS, you’ll find me. If you don’t put that in, you’re likely to find some cricketer. I’m looking forward to learning more about what people are doing and building and how you build applications for the entire trip. I do my own little demo application, and I’m hoping to have some Build Day Live events where we take that application is not the plane out to lots of each location. BuildDayLive.com or Build Day Live on YouTube.
Stephen Foskett: And as for me, as I said, you can find me at Gestalt IT where I am doing a weekly podcast on Monday that’s called Utilizing Edge, which you might have heard. Another weekly podcast on Tuesdays, the On-Premise IT Podcast, as well as a weekly tech news show on Wednesdays. So check that out, GestaltIT.com, or YouTube/GestaltITVideo. You can also find me on social media sites like Twitter and Mastodon at as SFoskett. So thanks for listening to Utilizing Edge, part of the Utilizing Tech podcast series. If you enjoyed this discussion, as I said, you can find this in your favorite podcast application, just type “Utilizing Tech.” Please do give us a rating, give us a review. We would love to hear from you as well. Send us some feedback. This podcast is brought to you by GestaltIT.com, your home for IT coverage from across the enterprise. For show notes and more episodes, go to our special website, UtilizingTech.com, or finalizing tech on Twitter or Mastodon as well. Let Utilizing Tech thanks for listening to this first episode of Utilizing Edge, and we will see you next week.