This episode of Over the Edge features an interview between Matt Trifiro and Paul Howarth, Senior Director of Business Development and Product Commercialization at Rogers Communications. In this episode, Paul describes why powerful networks are so important in heavy industry, why how autonomous tech can increase efficiency and safety, and looks forward to certain accelerations in his tech field.
This episode of Over the Edge features an interview between Matt Trifiro and Paul Howarth, Senior Director of Business Development and Product Commercialization at Rogers Communications. At Rogers, Paul leads a team that designs, builds, launches and sells world leading IoT products that leverage the power of 4G and 5G networks.
In this episode, Paul describes why powerful networks are so important in heavy industries. He elaborates on how autonomous tech can not only increase efficiency, but also safety. Paul and Matt discuss the difference between 4G and 5G, what works best, where it works best, and look forward to what might be coming next.
----
“Heavy industries have challenges getting data offsite. You'll see some pretty amazing things. I've seen fiber get built down in some very remote locations. I've seen microwave links being built over hundreds of kilometers because it's worth it. That data is so important to get back.”
“Think of sitting in your gamer chair with a joystick and being able to operate a drill. It's already hitting the mine face, drilling into the rock, and then maybe you switch screens and use the same joystick. But now you're operating a truck in another area. So now you've got someone who's highly skilled, operating multiple pieces of equipment, and doing it away from the danger.”
“With a highly available cellular network, you can actually pull people away from a dangerous work area, one that's not a great place to work. And they can do remote drilling, controlling vehicles, really helping also drive a diverse workforce, right? It's very difficult to get employees to go work in some of these places. And if you can tell a remote operator, you're able to pick up a lot of employees that might not want to live in a camp.”
“I'm really excited about what can happen and how we can bring software to the masses. Like it's incredible. How a lay person can now just say, I have this idea, I'd like it to do this. And these tools can now produce a lot of code.”
“When I think about the future, I've always had a principle that when you see new emerging technologies, brute strength and bandwidth is gonna win. So that's what makes me excited about 5G. It's just a lot of bandwidth they're able to drop on top of it.”
---------
(02:15) How Paul got into tech
(03:44) All about Rogers
(04:35) What is heavy industry?
(06:07) Heavy industries and wireless
(08:30) How data drives the industry
(09:00) What is the edge?
(10:59) What mining customers need
(14:58) Autonomy in heavy industry
(18:52) Autonomy and safety
(19:51) Deciding between 4G and 5G
(23:08) How Paul uses Edge
(28:04) The challenge of consolidating servers
(29:39) How fast should we make performance?
(31:51) How to build a network
(36:13) Rogers’ use cases
(39:33) What the future might hold
(41:21) What Paul’s looking forward to
(42:17) What Paul wants to accelerate
--------
Over the Edge is brought to you by Dell Technologies to unlock the potential of your infrastructure with edge solutions. From hardware and software to data and operations, across your entire multi-cloud environment, we’re here to help you simplify your edge so you can generate more value. Learn more by visiting DellTechnologies.com/SimplifyYourEdge for more information or click on the link in the show notes.
--------
Narrator 1: [00:00:00] Hello and welcome to Over the Edge. This episode features an interview between Matt Ferro and Paul Hower, senior Director of Business Development in product commercialization at Rogers Communications at Rogers. Paul leads a team that designs builds. Launches and sells world-leading iot products that leverage the power of four and 5G networks.
In this episode, Paul describes why powerful networks are so important in heavy industries. He elaborates on how autonomous tech. Can not only increase efficiency, but also safety. Paul and Matt discuss the difference between 4G and 5g, what works best, where it works best, and look forward to what might be coming next.
But before we get into it, here's a brief word from our sponsors.
Narrator 2: Over the Edge is brought to you by Dell Technologies to unlock the potential of your infrastructure with edge solutions, from hardware and software to data and [00:01:00] operations across your entire multi-cloud environment. We're here to help you simplify your edge so that you can generate more value.
Learn more by visiting Dell technologies.com/simplify your edge for more information, or click on the link in the show notes.
Matt Trifiro: Two years ago when I started the Over the Edge podcast, it was all about edge computing. That's all anybody could talk about. But since then I've realized the edge is part of a much larger revolution.
That's why I'm pretty proud to be one of the founding leaders of a nonprofit organization called the Open Grid Alliance for O G A. The OGA is all about incorporating the best of edge technologies across the entire spectrum of connectivity. From the centralized data center to the end user devices, the open grid will span the globe and will improve the performance and economics of new services like private, 5G, and smart retail.
If you want to be part of the open grid movement, I suggest you start@opengridalliance.org where you can download the original open grid manifesto and learn about the organization's recent projects and activities, including the launch of its first innovation zone in Las [00:02:00] Vegas, Nevada.
Narrator 1: And now please enjoy this interview between Matt Shapiro and Paul Hower of Roger's communications.
Matt Trifiro: Hey Paul. Good to see you again. Thanks for coming on the show. How are you today?
Paul Howarth: I'm doing really well. Great to see you again and chat with you again.
Matt Trifiro: Yeah, likewise. So one of the things I love to start with is just get an idea of how my guest, uh, first got into technology. Can you give me a sense of your, your arc into this technology world?
Paul Howarth: Yeah, it's, it's less than an arc, more of multiple arcs where you, you kind of weave back and forth those forks in the road. But believe it or not, I actually started working in the telecom industry. I actually climbed some telephone poles when I was really young as a technician working for one of the big operators, and then I.
From there, transitioned into operations, customer care. Ultimately sales. Did a bit of a stint with Cisco Systems for 10 or so years, and then I've been working at Rogers ever since. You can see I've got a few gray hairs on my head. So been doing this for, for quite some time. And within Rogers, I've worked on our wireline portfolio.
Rogers is a big cable [00:03:00] operator here in Canada, so I worked on that portfolio for businesses along with fiber. And then, For the last three or four years, been really working on four and 5G and driving our strategy around building private networks, mostly for those heavy industrial customers.
Matt Trifiro: Yeah, that's amazing. Do you have an engineering background?
Paul Howarth: I do not. Believe it or not, I was educated on the job. I have no, can teach you a lot? No, no formal university education. Although I'm proud to say that I was just looking at a a paper, a master's student we were working with on around edge compute. Published his paper and I see at the end he put me in as a nice thank you to me.
So I, I feel proud that, uh, I was able to help somebody who did make it through.
Matt Trifiro: Yeah, well done. Rogers is like a major carrier like we have in the US and it's, it is probably the, one of the, if not the largest, one of the largest wireless carriers in Canada. Is that correct?
Paul Howarth: Yeah, definitely the largest wireless operator.
And you know, we have a pretty big wire line footprint now. We just actually completed our transaction with Shock, another family owned company, so [00:04:00] we're really proud to bring the two families together very recently.
Matt Trifiro: How do you classify heavy industry as opposed to, you know, medium and light industry?
Paul Howarth: I suppose for us, you know, heavy industry and you, you'll find definitions different from everybody.
But for us, you know, think of mining. Canada is well known for its minerals and has been for years a lot more intensity in that space. Now that there's a need for more nickel copper in particular for the electrification, not just of automobiles, but within heavy industry as well. On top of that, we would kind of lump in.
Industrial plants. So people that sort of take some of those raw materials and turn them into different chemicals or substances needed for products down the road. So those are in heavy industry. And then we've been dabbling a little bit in logistics, but more the shipping container side of that business, looking at logistics and, and shipping ports more the.
The heavy equipment side of that particular business we'd love to be in lighter. Lighter to us would be manufacturing, warehousing, those types of industries. The [00:05:00] market today is really dominated by wifi, but we see the. That transitioning is, there's more automation in those spaces as well, and, and autonomy, I guess not just automation, but autonomy in those light sort of industry environments is really gonna drive a need for more secure, better coverage.
Something that a cell cell network's better suited for.
Matt Trifiro: Yeah. So let's, let's take that apart a little bit. The. Types of services and capabilities that you wanna deliver to heavy industry are both on the sort of private wireless, I imagine on the public wireless too, where there's coverage, but the private wireless and then the mobile edge compute.
So let's start with the wireless piece. Why is it that heavy industries seem to be the first to adopt these private wireless capabilities?
Paul Howarth: It's a couple of reasons. Largely, it's a coverage issue, I'll call it. They, they tend to have very, very large facilities, so it's not unusual for an open pit mine to be.
You know, 50 to a hundred square kilometers in total area. There's been lots of different wifi technologies that have been used over the years to try to deal with that, but ultimately, [00:06:00] cellular is just much more suited because they can, you know, the radios are more powerful. We can get them up onto structures a lot higher to give them broader coverage.
The second aspect is, Reliability, those heavy industries generally employ, say, one to 2000 employees. And when communications goes down, it's basically tools down. It's a safety issue on a lot of those heavy industrial sites. So you can imagine if you're the CEO of one of those companies having, you know, a thousand to 2000 employees sitting idle, your opex is just, you know, going through the roof.
So the business case works, you know, much easier. Right. You can, you can say, well, I got 1500 people tools down. I'm losing a hundred thousand dollars an hour or more. It's worth it for them to pay what is a premium for a cellular connection over wifi. Let's be clear about that. But the premium is worth it to them to get the, to get the uptime right.
So really coverage and uptime are the two drivers that we see in heavy industry.
Matt Trifiro: Yeah, that makes a lot of sense. And then when you think about what it takes to create a [00:07:00] private network, like what, what do I need to do? I've got a pit mine that's out. I don't know. Ken's a big place, so I imagine some of them are pretty remote.
Uh, pit mine that's remote. And I imagine the communication may just be internal. There may not be a, a, you know, a fast connection external. It may just be all like literally private within the a hundred, 200 kilometer facility, right.
Paul Howarth: It's twofold. So yes. You know, there's a lot of automation going on in in those industries, in mining in particular.
So part of it is the private infrastructure for something they call autonomous haulage. There's also remote drills. There's a lot of tools now that need that high available connectivity to one, increase productivity, but two, just keep employees safe and away from some of the more dangerous aspects of it.
What's interesting though is there's not a. A heavy industry, no matter how remote they are. And believe me, we've worked on some very remote places. One of the systems we put in, we actually had to load equipment on a ship and you could only get in, [00:08:00] you know, when the ice broke kind of deal. Every one of those customers needs to get that information back into a cloud.
They're, they're very data driven, these businesses now. And so it's pretty exciting, right? Cuz it's, it's definitely about the local connectivity, but we have to work with 'em and help them also get that data back because, That's the real value you're finding. The minerals are obviously important. That's kind of a known value.
The data is actually, in some respects just as important cuz it's telling you what to do next, where to go next, where to, where to point those 1500 employees to maximize your productivity. So they do have challenges getting that data offsite, you'll see some pretty amazing things, right? Like I, I've seen fiber get built down some very remote locations.
I've seen microwave links being built over hundreds of kilometers. Because it's worth it. That data is so important to get back. You know, we do, obviously there's still satellite, right? Uh, where needed and. You know, you probably know, Matt, there's there's more going on in satellite right now than a lot of [00:09:00] businesses, so pretty exciting there as well for, from a connectivity perspective.
But you'd be surprised how sort of cloud native these heavy industries really are, and I would say they're early adopters of some of the cloud technologies as well.
Matt Trifiro: Interesting. No, I w I wouldn't have, I wouldn't have, uh, naturally thought of that. That is sort of interesting. W when you think about, you know, it's like the first year of this show, uh, everybody was like, well, what is the edge?
What is the definition of the edge? And like, you've got a new definition now, right? The edge, the edge is not the edge of the. Enterprise premises. It's like the edge of the earth.
Paul Howarth: It's like, it's all, uh, it's a great question. I I, I've always said, you know, edge is depending who shoes you're in, right? If you're, if you're an automotive manufacturer, edge is the car.
If you're a mine, edge is at the mine site, regardless of where that is, you know, we would hope as a telecom operator that we play some role in the edge. Is somewhere, in some respects inside our network as well. But mobile phone manufacturers would say the edge is the device that you have in your hand.
So it's, it's a good point. And when you're having those discussions with customers, it's, it's always good to find out what their view of where the [00:10:00] edge is, because it's not, it's probably not your view.
Matt Trifiro: Right, right. Well, and I've discovered customers wanna buy the use case or the outcome of the use case.
No. They don't care where the edge is as long as it works.
Paul Howarth: Fair, fair point. Most don't know edge as a term. Right? Exactly.
Matt Trifiro: So, so let's talk about, um, one of your mining customers. You don't necessarily have to name it if you, if you don't want to, but if you, if you can, that should be, that'd be great. Walk through the process of, you know, how they approach you, what they were using before, what they needed to do, what they're doing now that they could never do before.
Is there somebody we can talk about that?
Paul Howarth: A mine or heavy industry customer would approach us from various different positions, but generally mines that mining specifically are self-contained units. E even the big global multinationals, the mines that are being developed are, you know, they're in different countries, there's different regulations, there's different technology available.
So generally it's the local mine site itself who we're dealing with, not sort of a, a corporate head office. Mm-hmm. And really the first step in building private. [00:11:00] Cellular or you know, edge Communications is really getting to the site. If it's an underground mine, we've done some work underground as well.
You know, the materials are different. Sometimes it's hard rock, sometimes it's soft materials. You know, radio frequencies penetrate differently, so, so we really need to get on the ground. That's actually the first step that customers need to take is make an investment in doing a radio frequency study. We can do cellular and wifi and others at the same time while we're there.
Once we have that, then we can plan it out pretty accurately. At that point, we can say, this is where we can provide coverage. How much coverage do you want? What performance do you want? And then it's simply working through the design aspects of it. The next big piece that usually comes up, and this is partly remoteness, but partly the important of production is, is availability.
How are you going to fix this thing? If you're eight hours away or 10 hours away, if you've gotta take a boat, Paul, to get up to where this equipment is. And the answer on that is, you know, we're gonna use the things we've been [00:12:00] using for 20, 30 years in telecom. We're gonna build designs that are highly available.
And so if something fails, there's always something else there to, to step in. Cuz you need to be clear. You're not getting to a remote mindset in four hours. Um, That's just not sensible. You really need to plan and build the infrastructure. So that's actually usually quite a long discussion, right? People want to make sure that you've planned that out correctly cuz it's, it's a technical discussion, it's a process discussion and it's a people discussion to make sure all those things work correctly, to give you that really high availability infrastructure contract and then, and then build.
And the builds are quite lengthy. And then with supply chain issues, some components can take three, four months to get on site. And we've been held up by. Things you wouldn't expect. I've been held up on a job by an air conditioning unit that was needed for an equipment shelter that was backward, oddly enough, so all the high tech stuff would arrived and was ready to go.
But, but the, a darn air conditioner was sort of the [00:13:00] holdup, uh, on that case. And then, and then we turned it over to the customer. So, you know, it's not like, You're gonna turn it on in a couple weeks. These things generally take six months to a year from that initial get on site, see what it looks like to design and then support.
We have one public mine site you were asking about, so that would be the Detour Lake mine. It's now an Agna Co eagle mine. Really well run operation. You know, they've got great coverage along the road for safety reasons. At the request of the mine installed. Solar and uh, wind. On a number of the cell sites that support that location.
Mm-hmm. Because they really wanted to make sure it was a very e s g friendly installation on top of that. So, um, great group of people there.
Matt Trifiro: That might also help with resilience in a remote location where electric power may not be as reliable.
Paul Howarth: Yeah, it does. I mean, we couldn't get it a hundred percent.
We're still seeing the results of, of the installation, but probably there's a few sites where there'll be no wind and not much sun. In Northern Ontario, and so we do have [00:14:00] propane there. If we did look at hydrogen, we just couldn't get enough of a supply. There's, there's not a diverse supply chain for that right now, but in the, in the future, I think you'll, you know, that will be another option to help some of these remote minds.
Get off of the grid or having to build grid where there isn't
Matt Trifiro: any. Yeah. You know, you mentioned autonomous. I don't remember what the, what the word was, but autonomous something, which, you know, I, I think of like, oh, that's the future of cars that everybody's been talking about that we may never see, but it sounds like it's really happening.
Paul Howarth: It's called autonomous haulage. Ah, in the mining industry. It's not com. Completely autonomous. There are, there are still people in the cabs. These are very, very large vehicles. You know, like two story high vehicles. The autonomy aspect of it makes sure that they're dispatched in the right order, make sure that they can travel down what could be a very narrow roadway in an open pit mine so it's not being driven by an operator all the time.
You know, the autonomy definitely kicks in and it's really important from a safety perspective. You can take fog and [00:15:00] things like that as you're driving down. The, the edge of a mine face, but it's also a productivity gain, right? If you can get the truck in the exact spot when the load is dumped into it, you'll get more load into the truck.
Secondly, you'll see, and this is where a, a cellular network really shines, is you're starting to see a lot of just remote operated equipment. Think of sitting in your gamer chair with a, with a joystick and being able to operate a drill, get the drill set up. It's already hitting the, the mine face, drilling into the rock, and then maybe switching screens and using the same joystick.
But now you're operating a truck in an ar, another area doing mucking, you know, picking up the, the blasted or, and moving that over the truck while the drill is already operating. So now you've got someone who's highly skilled by operating multiple pieces of equipment and doing it. Away from the danger, right?
Particularly underground. The mine face would be one of the most dangerous areas in a mine and underground. Just to give you an idea, Matt, like it's [00:16:00] a hundred percent humidity. Imagine when you're working underground in some of these deep mines. Imagine the rock is, you know, 60 degrees Celsius and you'll have to translate that for our Fahrenheit friends.
But, Trust me, that's hot. They've gotta blow cool air down the mine just to cool it off enough, down to 40 Celsius, which is still very hot. So employees can actually go work in that environment. So with a highly available cellular network, you can actually pull those people away from that area. That's not a great place to work.
It's not very comfortable. And they can be doing remote drilling, controlling vehicles, things like that, uh, underground and. Really helping also drive a diverse workforce, right? You, as you can imagine, it's very difficult to get employees to go work in some of these places. And if you can tell a operator, tell a remote operator, you're able to pick up, uh, a lot of employees that might not want to go live in a camp for two weeks.
Yeah.
Matt Trifiro: Wow, that's amazing. I mean, I still have this image of mining of, you know, like Snow White in the Seven Dwarves with the little helmet with the, the headline on it. You [00:17:00] know, go going into the little cart down, down two miles underground or whatever it is. But it makes perfect sense. I mean, it, it's a very labor intensive industry, and if you can make machines do most of that labor, and if you can automate the machines, I mean, I hadn't even thought of like the cost reduction in not having changed the tires as often.
It's like thinking of the entire. 200 kilometers as a, as a factory, as a machine with all these different parts that contribute. Yeah, it's really amazing. So I can see, you know, all the investments that in a, in a very significant wireless infrastructure, may be relatively small compared to all the other investments that they need to keep track of
Paul Howarth: on the site.
Yeah. They, they, they are. Yeah. In some respects, I should be in the truck selling business. I know as they're. Dispatching these vehicles. They need to know in the autonomous aspects, where are the vehicles relative to each other at all times. And the tolerance is zero packet loss. They're not dramatically requirements on speeds.
Generally. 30 megabits on download and 10 or 20 on the uplink is plenty. The [00:18:00] requirements on things like jitter. Packet loss are very, very tight because they need to know where the vehicles are. If they ever get outta sync, they have to stop all vehicles so they know where they are. Exactly. Again, and you know, just like working in a factory automated environment.
Yeah. You know, a robot would be in a cage to keep it away from people. Mm-hmm. When you start working on these massive properties with machines that have autonomy elements in them, you need to know where they are relative to the people. And so, so how do you know where the people are? Well, it's another good question, right?
Partly you might solve that with cellular or gps when you're on a surface. Mine much more complex. When you get underground. That would have to be something like Bluetooth or RFIDs often used actually to track, and again, it's. You know, a cell network, we can track things and triangulate relatively well, say to, you know, a number of feet that is getting better with 5G as we get things more dense and do software.
But they need to know where things are in a mining environment, generally [00:19:00] down to a couple of inches. Think about it all from that perspective, and it's really for safety reasons, right? As you can imagine. These vehicles are so large, having humans moving around them is very dangerous. Right. So they've gotta know where everything is.
And again, a cell network is critical for that coverage, but also the reliability.
Matt Trifiro: Yeah. So, so when your customers are deciding between a private 4G network and a private 5G network, how, how, what are kind of some of the trade-offs they look at? It's a great
Paul Howarth: question that the, the truth is that they're all going in with, 4G that like a hundred percent of the, not just in Canada but globally, would go in with some aspects of 4g.
And then there are some customers who are layering in 5g. 5G today is really just about if you need a tremendous amount of capacity, you know, 5G can really handle that a lot, a lot better than a 4G radio network. Can you mean
Matt Trifiro: bandwidth capacity or like, uh, unit capacity with
Paul Howarth: it's, it's both, but it's really bandwidth capacity.
Right? So you think about some of the [00:20:00] spectrum assets that have been allocated to 5g, the 5G radios themselves. Of course, the future is multiple gigabits, depending on the spectrum that's available, but generally you're getting into. You know, 500 megabits very reliably to gigabit speeds. We have a client we're working with right now that we haven't finalized, but they need to replace some fiber.
The particular automation they're using, the fiber optics break every five years or so. It costs about a hundred thousand to replace the fiber. They've got 10 or 12 machines. It's so, it's millions of dollars. The only way to do that is 5g. But candidly, the real reason, you know, we're still in a 4G world is devices.
Right. Think about all the things you need to plug in to all the machines there are. Thousands upon thousands of vendors that sell equipment all based on the 4G standard. So it's just, it's readily available. 5G devices will come, they are coming, like we do a lot of work with Credit Point that's now an Erickson company where they've got great 5G [00:21:00] platforms that we can put into the industrial space.
Again, for very specific applications though, right, because there is a premium for 5G if you really need it and. So it'll be a 4G world for, for quite some time in, in the industrial space, I believe. But certainly we have done 5G installations where needed, and we're happy to do more, but limited by devices
Matt Trifiro: really.
Now, I Is the, the, the rans that you install, are they sort of appliance-based or are you starting to move towards more virtualized rans running on white box or traditional kind of off the shelf?
Paul Howarth: Perverse very, we're very, You know, we we're deploying traditional architectures. Erickson's, the vendor, uh, has been our vendor since we started in the cellular business.
So much simpler for us to manage all that infrastructure. You know, we've got thousands of sites across Canada. We can manage the radios all as one infrastructure. We do have. Big innovation investments with a couple of universities in Canada, university of British Columbia, university of [00:22:00] Waterloo, amongst others.
But those two in particular, we are starting to look at some of the virtualized technologies there from a, an innovation perspective, but nothing in production. The
Matt Trifiro: other use of servers you've got, you talked about the cloud a little bit and I would like maybe to come back to the, to the cloud, cause I still think that's an interesting far edge relation related to what's happening in a centralized cloud.
But you know, mobile edge compute or mech or multi-axis edge computing, I guess as they call it more of these days, is really the compute locally attached to the ran. Essentially. What are you seeing happening there, especially in these heavy industries, A typical
Paul Howarth: application. That would run at the edge again, seems kind of boring, but it's, uh, would be pushed to talk.
So all those employees remember you, you'd asked how did we locate them? Mm-hmm. You know, a push to talk device generally will have the gps, it'll have the loan worker software in it for safety reasons that needs to be kept. Local running on an edge platform, for example, because it can't rely on a cloud connection in a very remote location.
You know, you can imagine, again, if you [00:23:00] can only get satellite back haul that we're gonna have to install some edge type services in those where we really see edge compute taking off those in that lighter industry. Things like think, think of retail supply chains, there's a lot more density. They have the same sort of real-time coverage requirement we're talking about with heavy industry, but they've got a lot of locations.
You could pull that back into an edge instead of having to put equipment in things at every single one of those locations. We really see there's a great cost optimization to be added by running.
Matt Trifiro: I don't know, a lot of my, my listeners will know the details of a retail supply chain, so like break it down.
What are all these pieces that are out and all, like how does that actually work? Is this about how. Goods get to my grocery shelves sort of questions.
Paul Howarth: Yeah. It, it can be that it's the, it's the trucking aspect, but it's, it's warehouses. So inside those warehouses, they are highly automating. They've always had some level of automation along with employees and those particular facilities, but now they're needing some of that autonomy, right?
They're needing to [00:24:00] change, you know, where things are. They're needing. Automation and robotics to be able to move more seamlessly through there. They're much closer to urban markets, and so those use cases start to make sense for you to embed some of your applications in a more. Centralized edge, I'll call it.
So you're, you're not gonna put them all in one region of Canada, just by our geography is so big. But we might be able to have something in the province of bc, the province of Alberta, Ontario, Quebec, where you might be able to centralize the application for three or four locations. Just, just make things a little bit simpler to maintain and automate and keep the latency at a level that the automation is still useful.
Versus having to send it all the way back to the cloud. Like in, in, in Canada, most of the cloud providers have their infrastructure in the east, we'll call it. Mm-hmm. Anybody writing software and they're testing it out, is working great when they're, if they develop it in the east and then you go [00:25:00] drop that application all the way out in Vancouver and suddenly there's 82nd milliseconds of latency and it's not working so good.
That's a great use case for edge compute. Retail also are heavy users of video for various reasons. Video is a great use case for layering on top. And then Matt, we actually see just in the retail store itself, it's a very wired world today where we put fiber optics in wifi wire to the different checkout lanes.
We do see a future with 5G in particular, where a lot of that can be virtualized, right? You could put a a SIM or an EIM at every checkout lane, bring all of that traffic back. Using something like 5G or network slicing, run applications at the edge within a region. Free up a whole lot of capital and opex that you've got and equipment in all those particular facilities.
Well, yes,
Matt Trifiro: and hassle and threat vectors and, you know, service nightmares and all of that. Yes. I mean, the, the last thing that, that a store actually wants to have on its [00:26:00] premises is a server room. Right? You got, it is an equipment closet, but the new applications don't, you know, a rack of NVIDIA servers does not fit in an equipment closet.
Paul Howarth: It's a, it's a challenge. How do you, how do you fix all that when there's maybe a thousand or 2000 retail outlets. How do you find technicians that know the store layout and, and how to fix that equipment is a real challenge. You, you're bang on. Yeah, I do. You know
Matt Trifiro: when, when you think of the edges like this, well, think of the internet as, as a continuum from the large centralized servers that are out.
Who knows where they are? They might be 8,000 milliseconds away. And maybe you don't care. And then you've got the server that's like on your premises or on your device, right? That's about as close as you can get. And then there are steps in between there. And you, you mentioned region. And so, you know, region's gonna be sort of halfway between, you know, maybe it's not 80 milli milliseconds, maybe 40 milliseconds.
There's lots of applications. You can support it 40 milliseconds, but there's a bunch you probably can't. Like maybe you're trying to do AI inferencing for automated checkout. Well then you need. You know, you may need like, like three milliseconds or something like that. But what's really interesting is the [00:27:00] cost piece.
Because if you say, okay, I can get sub three milliseconds. I just put a, a rack of servers in every store, but as we just discussed, that's, that may not, that may be cost prohibitive. And what may make a lot more sense is to pull those servers back and consolidate them into, you know, one set of servers that now serves.
However many stores you can in a region. And maybe like you said, the service thing is a really interesting thing and maybe it's in a data center environment that's the same everywhere.
Paul Howarth: You raised a good point. I mean, you now, now you've got some density, I'll call it, meaning you're probably can afford to have technicians in a.
Large facility, 7 24, who can fix power supplies or cables or things like that. As you can imagine, you know, we talked about how long I've been working on these things. Like I've just, I've seen networks and compute move back and forth. Mm-hmm. You know, over the years, you know, with a 5G connection. And, and even a modern 4G connection with the speeds we're getting now, you can actually put a wireless network in place and [00:28:00] you, you could pull that traffic back and get the latency capacity you're looking for something we couldn't have done, you know, maybe five years ago, certainly not 10 years ago.
So this concept of, of edge, you know, it's probably ready, the time is now. Like they're, they're facing these refreshes of equipment in the store. But they now have sort of wireless speeds that they didn't have access to. So I think it's a, it's a perfect storm. It hasn't happened yet. Like I said, heavy industry's moving first, but we definitely think it's gonna take off there in the next wave.
Yeah,
Matt Trifiro: I think it's interesting, you know, there is that, that sort of technology trope, which is like, well, where's the swinging pendulum gonna go? Right. You know, mainframes were centralized and then we had PCs and now we have the internet and all that. And I, I sort of think that maybe that paradigm's gonna break that it, it isn't swinging back and forth, it's gonna be wherever.
Wherever it's best placed. And it's gonna be a combination of cost, performance, performance, reliability, and, and those are gonna be the factors cuz there is, you are gonna have the ability to put it just about everywhere in this new distributed world. Yeah. And that's,
Paul Howarth: yeah. And [00:29:00] we, we haven't had that as network designers like I.
I still remember my first network class where I was taught. I said, well, how, how fast should we make the performance? When I asked the instructor, he said, well, Paul, it's always gonna be the best the customers ever had. No one will ever accept their laptop booting a second later, right? Like, once you've opened your new laptop, you'll never use the old one, right?
Because the new one boots up a a second faster. And so it's gonna be very, very similar. So in a, in a, you know, that retail example, When you tap your card for payment, you're expecting that to be, you know, less than a second or at most one or two seconds for that transaction. They're never gonna accept that it takes 10 seconds for, oh, you're
Matt Trifiro: absolutely right.
When I check out, like the fact that it takes five seconds is infuriating. I. Real, I mean, really. I just realized how ridiculous that sounds.
Paul Howarth: I know, but it's true.
Matt Trifiro: It's true. It's like, why am I waiting to take my card out?
Paul Howarth: Right. So you can't centralize everything cuz you're gonna add a [00:30:00] whole pile of things.
But like you said, the pendulum is going to be infinity. Now we'll put the edge where it needs to be to meet the, the needs of the customers and, and the types of transactions they're doing right. Yeah. So
Matt Trifiro: let, let's talk more about the, I don't know, the, the interface between a mobile network, a wireless network, and the sort of compute infrastructure we talked about.
So Rogers as a company that's primarily in the networking business, um, how do you relate to the application providers and the. Cloud companies and like, how do you see your role in that world? Are you gonna build your own applications? Are you gonna look to create an environment where others can, can come and sort of have an open interface?
Like how do you, how do you, how does Rogers look at
Paul Howarth: that? It, it's a good question. Uh, I mean, the answer could be any of the above. I would say it's unlikely that we'll build. Many of our own applications. And it's not to say that we're not capable as a group operators build a lot of software, [00:31:00] right? Like we are, yes, we're networking companies, but we're also great billers.
Like we all have. That's right. We all have the ability to scale billing and so, oh, it'll all pass through
Matt Trifiro: our network as long as we can bill for it.
Paul Howarth: So there's a lot of software involved in building a network and we, we, we invest in a lot of software to stay at the leading edge from that perspective.
You know, my comment is we're unlikely to go build the end user applications. Mm-hmm. Simply cuz it's not our skill or space. And there's really no need. There's, there's a good ecosystem of people out there that are building the software from that aspect already. But your point, we will end up having APIs or interfaces available as needed to get access to some of the networking features.
You know, one of the things we're seeing though is, you know, as it relates to. Using Edge Compute is just a general comment. There's, there's not a lot of developers or, or any that are graduating from universities that have a concept of [00:32:00] building distributed software. Right? They're building software that's generally designed to go to a single database, quite monolithic, I guess you would call it.
You know, we're going to need to find a way to move. The needle on that a little bit where people are starting to design applications where, you know, the classic example that's always given is, well, you'll, you'll collect the data from video at the edge, maybe in a retail application. Mm-hmm. You'll send it back to cloud to do the training, and then you'll push the model back for inference at the edge.
Which is actually all true. I've got a lot of background in working with mm-hmm. Video machine learning. There's not a piece of software out there that actually does that. Right. That, that we've seen where there's a platform that, I'm a retailer, I know nothing about video machine learning. I can just say train models and it'll start collecting data, move it back to cloud and say model's ready and I can just click deploy and it boom, it goes out to 1600 stores and starts doing its job at the edge.
That just doesn't [00:33:00] exist, right? Everything can be done. You could go custom build that, but there's not a piece of shrink wrap software, and I think that's a gap, right? If we really want this to take off, we need to find ways for our academic partners to start producing people that are thinking that way that they can actually build.
That's
Matt Trifiro: really fascinating. I mean, you know, you, you think of the companies that are, or the organizations, or not just companies that are doing that kind of work. Yes, it's the research institution and it's the major cloud providers. I mean, major cloud fires had to figure out how to deploy. Search and advertising across 60,000 servers.
Right, right. And have it all run. I mean, that's called distributed computing. And whether those 60,000 are in one room or spread throughout a city or a region, those are more similar to each other than, like you said, the sort of monolithic applications using a single database. So tell me, like you said you had some partnerships with the universities.
Do you, do you have any insight into, into how they think we should help? They, they could help us solve this problem.
Paul Howarth: The one I was personally involved in, university of Waterloo, a bit of a shadow to them. They, they. Are going to release, uh, [00:34:00] it's called Mech Bench. So we, we talked about this problem for a number of months, and ultimately what we landed on was if we're gonna work with Edge, and we're going to explain to customers the value of Edge.
We need a tool that'll actually say, here's how it works. If you have the database or the, the inference centralized. Mm-hmm. And here's how your model will work. If we had it out at the edge. And we also needed to be able to say, what does the network look like? So, so mech benches, I'll call it a 1.0 tool that was built that you can put in the characteristics of the network.
So it looks like 5g, it could be 4g, 3g, whatever characteristics you want to put in. Then you can run some models that have already been publicly made available, uh, against the tool and say, if it had Edge compute, here's how the model would perform. And if it was centralized here, it'll, here's how to perform so that we could go to a client and say, Hey, hypothetically, we think Edge Compute will help.
Here's a tool, let's, let's [00:35:00] model your application. Similar to your use case, here's how it would work if we had it at the edge and how accurate it would be. Here's how it would work if it was centralized. Now we've got some meat, right? To say, you know, edge is something that's valuable. So ultimately the tool should be open source.
Soon. I'll say if it not already, you know what it really needs now is. It needs a lot of people to go validate it. So we've, we've built it, but now we need the community to go validate, like, does the model work, did, did our assumptions around network? Are they, are they right, right. That kind of thing. So that, that's kind of where we're at from an academic perspective.
We've definitely tested the use cases we talked about. We've definitely tested, you know, doing inference at the edge. We've done some, um, great work around. Collision avoidance. We use lidar for example, at mm-hmm. Intersections. So we can actually see compliance, you know, someone's gonna call 9 1 1, but imagine you're in a very rural community at a uncontrolled intersection and there's an accident.
So that's a [00:36:00] great edge use case where if we had edge compute out there, a system like we developed at University of British Columbia, You would know that that accident happened. And the great thing about LIDAR is it doesn't do facial recognition. It just does object detection. So you know, it meets all sort of the privacy requirements.
Yeah. But you would know.
Matt Trifiro: So we're doing some work with the city of Las Vegas and one of the things they talk about is, you know, if there is something, you can't be everywhere in the city with law enforcement. And so if there's an active shooter, which they had on the strip, you know, not that long ago, you can not only like identify, do an, identify what's happening on the ground and private and situational awareness, but you can forward pulling drone drones that may be nearby and can get on site with, with video much faster.
Well, video, this is probably transmitting over 5G video much faster than, you know, human in a car or whatever. Transportation could ever get there. So I, I can see how these will greatly enhance public safety, uh, in large cities, but also in the rural example that you, you talked about. So
Paul Howarth: definitely a, definitely another good emerging market for Edge Compute for sure.
[00:37:00] And we've done those demonstrations. We're just getting to the point where we've proved it out. We've now built a tool that we can sort of business case how it works. We've now got 5G deployed at scale, at least in Canada anyway, so, so it's good timing. I think, you know, we will start to see these things really come together and some, some of that technology get adopted over the next couple
Matt Trifiro: years.
Yeah. You know, the, the, the mech bench work is really fascinating and it's, it's really cool that you're open sourcing it. If we can make, collect enough tele realtime telemetry in, in the network, let's call it, right? Um, so that we can feed a model like this with realtime data, we can make realtime.
Autonomous decisions around placement, because as a developer, that's kind of what you want. You wanna like make your workload and your little, you know, GitHub and then give it a profile like, okay, I need this level of resilience, I need this level of this, it needs to be below this cost. And then you hand it to an orchestrator and it says, well, I, I can't run it or, Yeah, I was able to run it and I maintained your SLAs and I used, [00:38:00] you know, low carbon because that's what you preferred and all those things.
So it's, it's, it's neat how these two, these worlds are, are converging. We get to see how, you know, it's like autonomous cars. We're only gonna have truly autonomous workloads
Paul Howarth: and, and, and I think all the hyperscalers would be in that case, right? But again, it's not, um, it's not something that's just part of the DNA of those graduates.
But hey, gi, give a shout out to the, the work you're doing for the mech bench team and. They'd, they'd be happy to engage in the wild and see how their tool works. Yeah, that'd
Matt Trifiro: be a lot of fun. So when you look, look to the future, like what are you most excited about with, with all these technology trends?
What, what gets you excited? I. Besides
Paul Howarth: retirement. I'm kidding, I'm kidding. Besides doing my podcast. Yeah. Oh, oh my gosh. So many things like it's, it's, it's so hard to even imagine, but you know, what's going on with some of the chat bots right now and the skill of that, you know, that could be the answer to the software dilemma now.
Right. I, my son is in computer science. Him [00:39:00] and I have been playing around with some of the, AI is for years, but it's almost at the point where I could say, build me a software platform that does this and deploys these models and brings it back and it, and it'll actually dump all the code on top of it.
And I, I've run through some of my, my poorly written Python code and you know, it cleans it up in 15 seconds. So I'm really excited about what can happen there and how we can bring. Software to the masses, like it's incredible. You, you know, how a lay person can now just say, I have this idea. I'd like it to do this.
These tools can now produce a lot of coats. I think that's a, that's an interesting aspect. I can, I can obviously see the negative sides of some of that as well. Bad actors can use the same tools to write software to do bad things, so we need to keep a close eye on how those cutter tools get deployed.
You know, from a pure network perspective, I can say that brute strength, speed is always one. In my career I've seen. Token ring back in the day, had a token with quality of service, but then [00:40:00] 10 megabit ethernet came along and it was half the price there. You know, I saw ATM disappear for the same reasons.
Yeah, with gigabit ethernet. So when I think out to the future, I've always had a principle that I. When you see new emerging technologies, brute strength, bandwidth is gonna win. So that's what makes me excited about 5g. Right. It's just a lot of bandwidth they're able to drop
Matt Trifiro: on top of it. Yeah. I mean, you and I, you and I both, uh, probably observed the, the circuit switch versus the T switch war, and we know who won.
Yeah. Big enough pipe. It doesn't matter.
Paul Howarth: Yeah, so, so I find that that part of technology really exciting. But, you know, generally, generally, you know, I, I just, I love and have always done, you know, I'm, I'm very lucky to be in an industry that I love working in every day. I love seeing what's coming new, and I love seeing the new types of jobs, the new roles and new capabilities that come out.
And, and honestly, I'm mostly excited now. I spent a lot of time just tr trying to help the next generation, right? I've had, I feel like I've had, uh, so many great opportunities working for great [00:41:00] companies and I see the technology landscape. I want to help others get involved and, and do more.
Matt Trifiro: That's awesome.
All right, I have my last question. So there's a lot of forces pushing us forward, right? You know, innovations in large language models and innovations in in use of spectrum and compute and orchestration, all these things. But they're probably things that are also holding us back. Uh, that maybe shouldn't be there.
If, if you could name one or two candidates of things, like, you know, if you think about how to get these dominoes going, if you just go like, you know, give it a little nudge, give it a little push, push one of those dominoes over, what would it be? What would you, what would you try to accelerate?
Paul Howarth: I, I can give you the first one right off the bat.
Yeah. Well, well, I think about the second one. The first one, it's been the same my entire career is anytime there's a new technology, it's very difficult to. Create the business case for it, because there's business cases to do it the old way and it very well documented that people will accept. It's always, you know, the most innovative companies have [00:42:00] always found a way to break through that and say, you know what?
Don't really understand the business case, but we think this is gonna give us competitive advantage and we're gonna make a bet. And fail fast if we have to. So I would say, you know, each new technology, it's always been the business case that's held back. Its adoption and, and for good reasons. There's some technologies that just made no sense, right?
Yeah. So they shouldn't have a business case, but, but it also holds back the ones that make a lot of sense. I would say, you know, we need to become. Better at accepting new technologies, taking a few more risks in that space and looking for the business cases a little bit more accurately, broadly. From an IT perspective, it's very similar.
It's just not getting stuck and doing this w the thing, you know, doing it the way we've always done it is always the biggest pushback. You know, we had a great chat about retail and perhaps getting the infrastructure and virtualizing it, bringing it back to the edge. That would. Go against the DNA of a lot of network architects, right.
That would, that would be [00:43:00] not what they've been taught. They've been taught to put in WAN and then maybe do this SD WAN thing and then wifi, and you're walking in and telling, we're just gonna blow that up and virtualize that. You know? So it's not a business case issue, it's just that one will just be inertia.
An openness. Yeah, inertia. An openness to, to starting and trying something new. That's fascinating.
Matt Trifiro: Yeah. The, the business case piece of it, I think is, is, uh, I mean they're actually, they're related in some ways. A lot of times the reason there isn't a breakthrough on the business case is because there's a comfort with doing things the existing way, and I get it.
Paul Howarth: Everybody's worked on a, you, you can work numbers any anyway you want. You know that I've been involved in lots of big business cases in my career. There's always some variability in the numbers. It's never black and white. All, all I'm saying is, you know, to get more inertia, we're probably, you know, the, the most innovative companies find a way to lean on the side of the new innovative solution a little bit more.
How, how does Rogers do that? Well, partly we do it through those [00:44:00] academic research facilities. So we, we, we work very closely. We listen intently though to our customers, like our customers are saying, this is what we need. Sometimes you just can't do it the old way. Like, you know, they, we knew they were all using, say, wifi, for example, in those heavy industrial sites.
We didn't just go build a private network product. We went out and interviewed 20 to 25 of the top heavy industrial customers in Canada, including some utilities. We just listened, what are your problems? And it happened to land on that cellular was the way to solve it. It may not have, it could have been something else.
So I would say, You know, we're willing to listen to our customers. We're willing to work on innovative things within our, our academic institutions and our partnerships that we've got there. We just have an innovative culture because we've always been a family company and, you know, proud to say the name on the building still belongs to a family, the Rogers family, and that helps a lot, right?
They're very supportive of making sure that we stay a [00:45:00] leader in our industry.
Matt Trifiro: That's really cool. Hey Paul, thanks so much for joining us. I learned a lot about the heavy industry and some of the interesting work that Rogers is doing. I really appreciate you spending some
Paul Howarth: time with us. Hey, it's been great.
Look forward to doing the next one. Cheers. That does it
Narrator 2: for this episode of Over the Edge. If you're enjoying the show, please leave a rating and a review and tell a friend. Over the edge is made possible through the Jenna sponsorship of our partners at Dell Technologies. Simplify your edge so you can generate more value.
Learn more by visiting dell.com.