In this episode of Over the Edge, Matt Trifiro sits down with Bill Pfeifer, Edge Portfolio and Thought Leadership Director at Dell Technologies, to go over a special announcement. The two dive into the creation and history of the Over the Edge podcast, as well as the future of the show.
In this episode of Over the Edge, Matt Trifiro sits down with Bill Pfeifer, Edge Portfolio and Thought Leadership Director at Dell Technologies, to go over a special announcement. The two dive into the creation and history of the Over the Edge podcast, as well as the future of the show. Matt shares what he has learned so far on the show and some of his favorite episodes.
---------
“This is a huge technology swing, right? This is like cloud was 15 years ago, but we're still at the very opening acts of edge computing and what it's going to do to the world, the impact it's going to have.” - Bill Pfeifer
“It's fascinating the different points of view that you come across and the different perspectives. I know what my perspectives on edge are and they're changing all the time but it's a single perspective and everybody else has radically different perspectives on this.” - Bill Pfeifer
“One of my favorite definitions of the edge, and I agree there are lots of them, is where the digital world meets the physical world, and I think there's some truth to that.” - Matt Trifiro
“My current fascination is artificial intelligence. And again, I'm coming at it from a perspective of sort of a naive user, right? It's like the tricorder. I see a future where every object in our lives, even the most mundane objects like coffee cups and sports coats might be connected to the internet or a network and might have either AI on board or be able to tap into AI that's delivered through a network.” - Matt Trifiro
---------
(02:00) Announcement from Matt
(02:47) Introduction to Bill Pfeifer
(04:01) The different perspectives on edge
(13:43) Matt’s reasons for starting the show
(15:41) Edge topics that Bill is interested in
(24:15) Matt’s takeaways from the show up to this point
(28:05) Some of Matt’s favorite episodes
(34:43) What is Matt interested in moving forward?
(36:23) Edge questions that Matt still hasn’t gotten answers to
--------
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.
--------
Follow Bill on LinkedIn
Follow Matt on LinkedIn
Referenced Past Episodes:
Three Decades of Vision for Edge with Mahadev Satyanarayanan (Satya) of Carnegie Mellon University
Stateful Computing, Continuous Intelligence, and Edge AI with Simon Crosby, CTO of SWIM.AI
The Genesis of Edge Computing with Victor Bahl, Technical Fellow at Microsoft Research
Solving the Fundamental Problems of the Cloud with Chetan Venkatesh, CEO & Co-founder of Macrometa
The Future of Edge is Messier Than You Think with Dean Bubley, Founder of Disruptive Analysis
Bringing American Manufacturing into the Fifth Industrial Revolution with Walker Reynolds, President and Solutions Architect of 4.0 Solutions
How Standards Drive Adoption and Enable the Intelligent Edge with Alex Reznik, Distinguished Technologist at HPE and Chair of ETSI MEC
Brewing Beer at the Edge with Matthew Steinberg, Co-Founder of Exhibit 'A' Brewing, and Pierluca Chiodelli,Vice President Engineering Technology & Edge Portfolio Product Management and Customer Operations, Dell Technologies
Dell and Exhibit 'A', Continued with Matthew Steinberg, Co-Founder of Exhibit ‘A’ brewing, and Pierluca Chiodelli,Vice President Engineering Technology & Edge Portfolio Product Management and Customer Operations, Dell Technologies
[00:00:00]
Narrator 1: Hello and welcome to Over the Edge. This episode features a conversation between Matt Trufiro and Bill Piper, Edge Portfolio and Thought Leadership Director at Dell Technologies. Matt and Bill go over a special announcement, then dive into the creation and history of the Over the Edge podcast, as well as the future of the show.
Matt shares lessons he's learned while hosting, questions that remain unanswered, and his favorite past episodes. 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 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. com. Technologies. com slash Simplify Your Edge. For more information or click on the link in the show
Matt Trifiro: notes. Two years ago, when I started the Over the Edge podcast, it was all [00:01:00] 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 or OGA. 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 it will improve the performance and economics of new services like private five G 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 Vegas, Nevada.
Narrator 1: And now, please enjoy this interview between Matt Trofiro and Bill Pfeiffer, Edge Portfolio Messaging and Thought Leadership Director at Dell Technologies.
Matt Trifiro: So today's episode of Over the Edge is going to be a little different than the usual. And the reason for that is I'm announcing that I'm stepping up [00:02:00] to executive producer of the show and I'm handing over the host job to Bill Pfeiffer. Bill is here with me today. And how are you doing, Bill? I'm doing well.
How are you, Matt? I'm super. Now, you may recognize Bill's name or his voice because he's been on previous episodes. We're going to spend this episode talking about the history of over the edge, history of edge computing, all the things I've learned and where Bill wants to take the show, where we want to take the show.
So it's great to have you here. It's great to hear about you on the show, Bill. It's super excited for you to take over the show as the host. For those who aren't familiar with you, tell us a little bit about yourself. Well, thanks, Matt. So
Bill Pfeifer: I've got a fair history in the networking, security, computing space.
And I was working on the edge for a while now, AI, before that. So I've been kind of all over the industry, right? I sold to customers. I was a customer. I was a partner. I [00:03:00] trained engineers for a long time, and I'm just kind of enjoying working stories about technology and the evolution. Into what I do and helping people understand where we are, where we're going and how we're getting there.
Matt Trifiro: Yeah, that's, that's great. And what makes you want to take on the labor of love of being the host of a podcast? It's really about the conversations,
Bill Pfeifer: right? So I had listened to many of these episodes and now I'm going back and of them and some of them with a fresh ear and. It's fascinating the different points of view that you come across and the different perspectives.
So I know what my perspectives on edge are, and they're, they're changing all the time, but I mean, it's, it's a relative, it's a single perspective. Mine too. And everybody else has radically different perspectives on this. It's emerging so fast that how far in that maturity continuum you are [00:04:00] and where you are in the, are you a customer?
Are you a service provider? Are you a vendor? Most of the vendors like me, like you, right? We sell certain stuff at the edge, and so we get kind of stuck in that perspective, but other people aren't stuck in that perspective. They're stuck in a very different perspective, or they're evolving their perspective right now while we talk and things like that.
And it, it's just, it's such a fascinating space because. It's moving toward that ubiquitous compute, everything's going out to where the people are, to where the business is, and it's going to change things so radically over the next 5, 10, 15 years. This is a huge technology swing, right? This is like cloud was 15 years ago, but we're still at the very opening acts of edge computing and what it's going to do to the world, the impact it's going to have.
And if you think about 15 years ago, before cloud was popular, when you went someplace, you had to know where you [00:05:00] were and where you were going, and you printed a map and you stopped at a payphone and made phone calls and did all these crazy things that, you know, now you just grab your phone and you're good.
What more do you need? I have my phone. I'm fine. You know, I can get gas. I can get everything I need from there. It can tell me where I am and where I'm going. I don't even know where I am and I don't care. Life has fundamentally changed and I think in five years, at least 10 years, we're going to look back and go, wow, how did we do this before thing that hasn't been invented yet was in
Matt Trifiro: place?
I mean, it's absolutely extraordinary, Bill, just when you're thinking about this, I was thinking about maps and yeah, I learned how to drive and I had a glove box full of paper maps that when you got lost, you'd have to unfurl. And even when. Internet directions, MapQuest, I think was the first, came out, you still would enter the directions and then print it out and take the piece of paper with you because you didn't have a smartphone, because the internet preceded the smartphone.
I think, you know, just the other day, and [00:06:00] we may have talked about this before, but, you know, I grew up watching the original Star Trek episodes. I mean, they, they were mostly filmed before I was, I'm not that old, but. You know, as a youngster, I watched them and was fascinated by them. And I always wanted to own a tricorder because I just thought that was the coolest thing.
You could walk around and, you know, kind of do everything you needed, pointed at something. Like anytime they had a problem in the plot that they needed to pump some mechanical, some duet machina, duets machina to get it out of it, they would pull out the tricorder. And like, we all carry tricorders with us now.
I mean, the smartphone is basically a tricorder. And I think you're right. When we look back at. This time, when my children maybe are my age. They're going to have an entirely different world. I mean, everything's going to be connected. I mean, your clothing might even be connected. Your clothing might have AI.
Your doorknob may have AI. I mean, who knows? And I just think it's such a, you're totally right. It's an incredibly rich field.
Bill Pfeifer: There are these conversations about like by 2050, 40% of the jobs out there haven't even been invented yet. [00:07:00] And so even guessing what people are going to be doing. Let alone what we're going to be doing, you know, what's the industry going to look like?
Well, what, what industries will still be there and be relevant? Who knows? It's all going to shift in such fundamental ways that it just, it's fascinating. And again, we're kind of on the ground floor of that next massive transition.
Matt Trifiro: Yeah, I, you know, at the edge of one, one of, one of the ways that I fond of, of explaining it to people that, because it took me a long time to understand this just wasn't a, a light shift.
For years, people talk about the pendulum of computing, you know, it started as mainframes and then went to terminals and, and I think that analogy actually is a little misleading because it's not quite how it works. In fact, what we're going to is like. Computing isn't in two places, it's on a continuum, on a gradient, from the physical world to potentially a centralized cloud and every possible waypoint in between.
And I think that that's a mental model [00:08:00] that people have to, you know, really have to be introduced to. And the other thing is, and this is so true, the internet that we've used, that we continue to use, that we think of as the internet, as people, was built for people interacting with computers and servers, right?
And so most of the rich content, most of the data, starts in a central location as a Netflix movie and it gets shipped down to you. And so most of the technical evolution of the internet has been to support those processes where you've got, you know, a bunch of centralized resources that are trying to deliver a global service, you know, whether it's Netflix or Amazon.
One of my favorite definitions of the edge, and I agree that lots of them is, is where the digital world meets the physical world. And I think there's some truth to that. And at that point, that is where the data gets created. It's the physical world and it gets handed off to the digital world. And in this world where there are more 4K cameras than there will ever be movies on [00:09:00] Netflix and they're on 24 7, the data generated at the edge that potentially needs to be shipped around, maybe back to the core, maybe just locally, is going to far eclipse the data that we've used historically to ship to humans.
And the other thing is, most of this data will be created by machines and consumed by machines. People will be involved, but people will get the output. People will get an alert that the lathe in their factory, their digital lathe is about to fail, and they should turn it off and get a repair person out.
So they have a little downtime or, you know, whatever those things are. And that's going to have been calculated using a machine learning model that was processed on a thousand GPUs to build that deep learning thing. And then it was shipped down to, you know, a smaller server somewhere that's maybe even on the device itself.
That's monitoring this that identifies an anomaly and then surfaces it to a human that can then, you know, schedule the repair or whatever, or maybe even a robotic repair. I mean, eventually a lot of that stuff's just going to happen as a consequence of the system identifying that there's a [00:10:00] problem it can remediate.
Machines talking to machines happen at a timescale that humans really don't understand unless you're in that world. You know, we operate in ones of seconds. Google has done a great deal of study on website and mobile screen repaints. And how long people are willing to wait. And anything over a hundred milliseconds, people see as a delay and anything less than a hundred milliseconds, they don't notice this kind of is kind of the rule of thumb and machines, a hundred milliseconds is glacial, right?
Machines are operating on nanoseconds and microseconds and ones of milliseconds. And so in order to handle all that data at the edge, to figure out where it needs to go, to store it locally, some of it, all of that, and to do it at those timescales. You're right. Requires a dramatic restructuring of the internet.
And that's quite an evolution from what was originally called edge computing. As far as I can tell, when I did the research, the first use of edge [00:11:00] computing that I can find is a paper that Akamai printed where they were putting caches out at the edge to deliver that Netflix like content to us humans.
And now it's a very different thing. So I'm excited to see the next few seasons of guests that you bring on that help guide us into. It should be a
Bill Pfeifer: lot of fun for sure, and as we evolve into people moving around in the world reacting to you because you're moving around or the world reacting to itself because some situation has changed or whatever, it's just, it's going to be fascinating.
And there are so many different points of view about this and so many different levels of understanding, which isn't to say anybody's wrong, right? They just have a different point of view. And where your point of view sits right now really can impact how you see the edge, how you think about it, what it is to you.
And you know, whether you even talk about the edge as the edge or is it just, this is my [00:12:00] place of business, this is my home, this is the integration that I have, this is the automation that I have. Do you think about that as edge computing? I think most people don't. And so that's probably one of the first hurdles.
We have to overcome is either we have to stop calling it an edge or the rest of the world has to understand that it is the edge. And, you know, as we get closer to that ubiquitous computing, we'll probably stop talking about edge, data center, cloud. It's just going to be available compute and things are going to
Matt Trifiro: happen.
I mean, the early days of cloud, before it was called cloud, the folks at Sun Microsystems and some of these, you know, people working on grid research and things like that, they imagined that compute. And I think in this world, you could even say like AI, but compute and AI would be delivered into our homes and our businesses the same way electricity is.
And if you think about, right, like it's utility, it's just, and it's available, like everywhere there's a socket, you can tap into electricity. It is the sockets, how we invite the power into our house and [00:13:00] our businesses. And it's dramatically changed everything. And I think as compute has been invited to our house through a network connection instead of a electrical socket.
I think it's had the same dramatic transitions and that's a, nobody felt the need to talk about the edge back then. It was really this recent invention with a bunch of us trying to find business there. Well, sure. When
Bill Pfeifer: compute was the size of a room, you didn't really want
Matt Trifiro: that.
Bill Pfeifer: Right. That's that's why it was going to be delivered like electricity.
Right. It was going to be a giant building that was one big
Matt Trifiro: computer. Right. It's the size of the Hoover Dam. So, I mean, one of the reasons I started the show back when I first got involved in edge computing, when I started as the CMO of Vapor. io, there was a lot of talk about edge computing and everybody had a different definition.
In fact, I had a lot of fun going around and asking everybody I could meet. I'd go to shows and things and say, what's your definition of edge? And the joke used to be, ask a hundred people, you get 112 answers. And that was true. And so one of the projects that I co founded was the State of the Edge, which was a [00:14:00] research report that was published for free.
It's now part of the Linux Foundation. And its goal was to try to find the commonalities and talk about the differences, but as differences just so you, you actually can understand this landscape. The show came out of that a little bit because first I'm a marketer. That's my trade. And I'm a wannabe engineer.
And so I love talking to engineers. I love learning about engineering, but. I'm dumber than most engineers, but I'm smart enough to feel like there's a lot of other people like me out there. And there's a lot of people that may think they understand something, but don't because they haven't taken the time to really talk to the practitioners and get into that level of detail.
And so I felt, well, let me try to showcase some of the leaders, some of the champions of this movement, this technology, and let me ask, you know, really basic questions that are explanatory. And so while the show is structured as kind of informal, like we're sitting around, like you and I are today, we're having a [00:15:00] cup of coffee or a beer or whatever, I very consciously spend time trying to explain things and trying to get my guests to explain them.
In, in ways that most of us will understand. Hopefully chat GPT won't, won't replace that function anytime soon. Actually, it doesn't have good answers for, you know, the nuances in Edge that we talk about. I mean, what are some of the topics that are kind of on your, I'd like to learn about radar and Edge that you might find a guest to have on the show to explain?
Well,
Bill Pfeifer: so again, the edge is where so many things are coming together. I like exploring all those different things that are coming together, right? So, we're going to put all these devices out there in the world and collect data and generate insights from them. What's that going to do to the security
Matt Trifiro: space?
Bill Pfeifer: And so, there's a security report out there that's based on actual hacks that have happened. I'm not going to name it right now. Future Hope, right? It's not a vendor specific thing. It's [00:16:00] just how were people actually hacked without any agenda of so buy my product. And it would be fascinating to have folks like that on just to talk about what are you seeing today?
What are you expecting comes next? Because they're probably looking toward next year, several years out. What are you anticipating based on this shift of where the data is? End. AI, of course, right? I mean, the reason the edge is so prevalent, we have all this data that's being created. Why is it being created?
Why are companies funding this? Why are we putting all these cameras and other devices out there? Because you can use AI to generate...
Matt Trifiro: It's the fuel, data's the fuel that drives AI. Right? And then that AI, AI is incredibly
Bill Pfeifer: power intensive. Incredibly. And yet there's this movement towards sustainability so we don't all, you know, burn up the earth and die.
And so how do we get AI fuel efficient enough, power sustainable enough? [00:17:00] Right? That's a business decision. So, you know, we're moving full speed into AI, but then how do we do that in a way that gets it efficient enough? Gen AI is the big thing that everyone's talking about right now. And the servers that run that are like five, six, seven
Matt Trifiro: kilowatts each.
You thought Bitcoin mining, you thought Bitcoin mining was contributing to global warming. I think it's hotter this year, this summer, because of chat
Bill Pfeifer: GP. Like five, ten years ago, a data center rack was five or ten kilowatts. You can hang one of those servers in an old data center rack and nothing else.
Yeah. It's just amazing. And so, data management, all of this data is going to be created. We're going to have AI. Something's got to happen in between there to clean it and get it lined up. AI ops, handling all the models, dev ops, handling all of the deployment. I mean, there's just, there's so much that's all piling on.
We say, as an industry, as a group, you know, you deploy edge computing, and then you can generate new value, [00:18:00] some variant of that. But buried in there is, you generate the data, you clean the data, you collect the data, you run the data through an AI model that's been trained. You do some sort of automation to get that value.
You integrate that into the culture of your people. Change management. That's a huge thing. There's so much, even just looking at the human aspects of it is going to be fascinating. We're going to have this, this sea change of how we all behave. Change management for organizations is really hard, really hard.
10, 15 years. Again, you know, by 2050, most jobs. Or many jobs that people have haven't even existed yet. That
Matt Trifiro: could be a challenge. Yeah. You know, I'm glad you brought up the business and environmental angles for a couple reasons. One is I didn't cover those that much in the first three seasons. And they're very important and very interesting, but also they are hurdles we have to get [00:19:00] over to make all this actually happen.
And so it is really important that we talk about it. You know, you think about power consumption and you and I both immediately go to the servers because those are the things that generate heat. And when you're trying to ship lots of data back and forth over long distances, that's a, it consumes a surprising amount of power.
It's not just the data over the optical fire and wires, it's the data over the airspace. And while your phone may not be consuming kilowatts of power, every little bit of power it uses means diminished battery life. And so you have all of these different objectives that you have to optimize for. You have to optimize for cost, which is a huge thing, the catalyst for.
You know, rolling out things at scale is cost. It's operational cost. It's like, how many people do you need to hire and trucks need to own in order to like service all these things out at the edge, where and who owns the equipment? You know, is it owned by a cloud provider and I can buy it by the bit? [00:20:00] Or is it owned by me?
Is it owned by a leasing company? Is it owned by the hardware manufacturer? Is it a combination of all those things? Who's responsible for it if it breaks? Who's responsible for the fiscal protection, he's like, there's just like a thousand issues. And so reducing the, you know, the complexity, my, my CEO at Vapor says that the killer app for the edge is economics and the easy button.
And that's a, you know, a clever, a clever turn of phrase, but it really turns out to be true. Yeah. I mean, you need a practical outcome, but a lot of the big hurdles that I've seen, you know, I've been at this for almost eight years and I thought it was going to happen. I thought that the sea change that you're talking about was going to happen a lot faster.
There's so much that's
Bill Pfeifer: going to go with that, right? We're just getting real momentum toward cloud. And that was
15
Matt Trifiro: years ago that, that really. Oh, I bought into everything. I thought we were all going to be in autonomous cars, right? I totally bought into that. And like now Thomas cars are geared, but Thomas cars will happen.
When I interviewed you, we had this little debate on, [00:21:00] you know, what is edge? And I think it's still, in some ways it's kind of a banal question now, but it's also got some interesting implications because. You know, you were coming at it from the perspective of Dell, which is currently has a lot of products that go on premises.
And I was coming up with the infrastructure and the network, right? Because I'm saying, well, why have anything on premises if it can be delivered like electricity? And, you know, the answer is both. And I think the simplest way I've found to explain that to people is that the folks in autonomous driving are very, very determined to tell you that all the A's and I's are going to run on the car and the car is going to be separate.
Right. And I think some of those are good reasons, right? Do you actually want the control loop for the brakes to have to, you know, go over the wireless network? Well, no. So there has to be a lot of AI on the car so it can, it can handle situations when it's not connected or, you know, maybe it's a rural area or something, right?
But you also want your car to see around corners. And the [00:22:00] only way your car can see around corners is if it's connected to something else, other cars through the cars or more likely through the infrastructure. Although it could be both. And I think when you start thinking about things that have to happen on a macro scale, when the data in my factory can be combined with the data in a thousand other factories that I may or may not have a relationship with.
In order to build a better machine learning model for all of us, it's just extraordinary. And so it won't just be things running on the edge, on premises, or on the device. That'll be a big part of it, but it'll also be the way that the data is combined in the network, the way that models are computationally intensive models.
Deep learning is very computationally intensive, and that means it's expensive. It means it generates a lot of heat. Probably better to do in the cloud, but then you have the problem of like, well, we gotta ship the data back and forth. And I sat next to the CTO of Seagate once, and we were just sort of playing [00:23:00] around at lunch.
And we were, I can't remember the exact math, but we were, it was something like, okay, let's imagine I have a factory that generates a petabyte of data a month. And I need to send that back. I need to send that from New York to Seattle. How long is that going to take? And basically, his conclusion was, it would be cheaper and faster to put those on spinning, put that data on spinning hard drives and put those hard drives on a plane.
I mean, sorry, on a train and ship them to Seattle. That that'd be fast. And that's just, you know, that's a head scratcher. And part of it is like a petabyte is a million gigabytes or so. It's a lot. It's a petabyte giant, you know, and so you just can't conceive of that. Right. But. Also, we think of bandwidth being infinite, to some extent, and latency being zero, and it's not.
Until data becomes almost infinite, too. Right, right, right. Exactly, right. Yeah. Yeah. Yeah. What happens when you try to sit, fit in an infinitely large cylinder [00:24:00] through an infinitely large hole. Right. So,
Bill Pfeifer: you've been doing this podcast for three years now.
Matt Trifiro: Yeah. What are your takeaways from it? What have you learned?
I mean, wow. I've, I've learned a lot. I think, I think what I've learned most. is to not be dogmatic about how this is going to roll out and not to trust anybody else that's dogmatic. You know, in the early days, I thought it was really important to, for me to form an opinion that was right. And by doing research, I would get to that opinion.
And what I discovered after talking to all these really, really smart people who've been studying this, you know, for their careers sometimes, folks like, you know, Satya at Carnegie Mellon and Victor Ball at Microsoft Research. You know, you talk to these people and you realize that there's a tremendous amount of complexity and there is no one right answer.
And, you know, in fact, interesting enough, I think AI is going to help us come to the optimal [00:25:00] answers. And it's going to happen faster than humans can think, because it's going to constantly be changing. I believe workloads are going to, there's no one right place to run a workload. You're going to have a set of resources and a set of objectives, latency, cost, reliability, all those things.
And, you know, and a bunch of AI models are going to say, okay, well, this is what I need to do to make that happen. And that's going to change constantly. I mean, you look at the. You look at like a 5G RAN, that's a radio access network, and they've virtualized it now, and so. It runs on the edge. That's what a radio access network is.
It is by definition, something that's running on the edge. And it's changing the RF signal thousands of times a second. And you need compute to do that. And it needs, and, and, and there's no human that's, you know, back in the days when we were, you know, developers that had two choices of where to put a workload in the cloud, US West and US East.
So when we had like two regions, okay, that was easy. Okay. I fell over one to the other. I can run duplicates, right? [00:26:00] Which by the way, always
Bill Pfeifer: made me laugh, right? Cause it's cloud computing. It's not a location where you can run it here
Matt Trifiro: or you can run it here. Well, that's that. And that is, that is, uh, I mean.
As a developer, as someone who wants to deploy and build applications, I want it to be like the cloud. I want infinite resources at increasingly lower cost that just kind of manage themselves. Yep. And you don't know or care where it's sitting. I don't care where it runs as long as it gets the job done.
But the reality is where we are today, a lot of that still has to be, we don't know how it's going to work. But we know that AI is going to play a really important role. Because maybe with 20 regions or data centers, a human, maybe a hundred, like you have a spreadsheet. I mean, that's what they used to do in the old days, right?
These guys would have a, this is a calc and they would say, okay, this workload runs on this server and this workload runs on this server. And now you kind of, it's in the cloud and I don't know what server it's on because Amazon's going to figure that out. But eventually it's going to be, I don't even know where it ran.
You know, maybe I can go look at a log [00:27:00] somewhere, but it's happening so quickly. And as the whole
Bill Pfeifer: FinOps process spins up, right, FinOps is all about dynamically pushing workloads wherever it makes the most financial sense. So now we just add some latency and some performance characteristics to that. And FinOps becomes just what enables.
I don't know or care where that thing ran. It just
Matt Trifiro: happened. Yeah. So I think that's what, that's probably the most important thing I learned. And then, you know, each guest has something to teach me, which is part of the fun of it. I mean, and my, the dirty little secret is I ask these straightforward questions that are designed to get at some simple answer to a complex topic.
Cause I want to know, like, this is just, this is just my dirty little secret. Like it's. I can actually talk to the guy that invented the cloudlet, which is, you know, a small version of the cloud that runs at the edge. And that's, you know, Satya at Carnegie Mellon, and he's been doing this stuff for 20 years.
And, you know, I can ask him, you know, how this is supposed to [00:28:00] work and what he's seen. It's hard to do that unless you're a, unless you're a podcast host. I don't even think I'd get them on the, on the phone if I wasn't a podcast host. So in addition to like learning everything from guests, I mean, I've had to, I said, some of my favorites.
So, so Satya Karnimel was definitely one of my favorite guests. Victor Ball. I'm fond for the research, for researchers, as you know, but also like Simon ai. I mean, partly because he talks like a sailor and partly because he just, like, he's got a very, very different view of how AI is going to run at the edge.
And it's fascinating. You know, it's going to be thousands of little processes that are all just bubbling up a little piece of data from whatever they're supposed to do. And then, then aggregate that system is going to make intelligent things. And, you know, when I spoke to him, he's talking about these very large deployments that were making meaningful improvements in sophisticated systems like a radio access network.
Using that, I liked talking to Alex Resnick, partly because he's just a great guy that I've run into at conferences, but he was at HPE when I, when I first interviewed him, [00:29:00] what he mostly did was. He ran Etsy MEC, which was the, well, it started as Mobile Edge Compute. That's the idea of the telcos putting servers in their network at the edge.
And now it's multi access edge compute because the telcos are becoming cable companies and the cable companies are becoming telcos and, you know, it's, it's, it's sort of a political decision, but they did it. And, you know, this actually comes to one of the sort of this nobody's right answer. I mean, I remember having, not personally having these debates, but seeing debates where people were talking about like the telco edge, right?
And, you know, somebody pointed out to me once like, well, why would you define an edge by who owns it? Like, is it really, is that how we're defining edge? And so I think you've got a really comprehensive body of work that Etsy has done in thinking about embedding, compute. into the radio access network. And I think one of the things that, well, I won't say they didn't get right, cause that's not, I don't really, not my place to say that or not, but.
Their big [00:30:00] vision was, we're going to figure out how to run all workloads at the edge and telcos are not going to be dump pipes anymore. And we're going to, you know, become cloud providers and all that. And, and I think they've, they've realized that they're not ever going to be really good cloud providers that developers are going to want to develop for the existing clouds and they're never going to catch up.
But what they realized is that if you've got compute running in the radio access network, you have unique data you can get access to. You can network congestion. You can, you know, request network slices for particular workloads, which would give you sort of a guaranteed throughput quality of service and an application to request it.
So. You know, what they really did get right was, okay, we need APIs to all these different systems. We need a ton of data that comes off of all these systems because we need these workloads in the AI that's figuring out where they go to be able to reason about them. And as we said earlier, data drives AI.
So it was fun to talk to Alex about the evolution of, you know, a standards body built around edge. And then watching how Mobile Edge Compute [00:31:00] has become something different than when it started. You know, it's not a bad transition, but like it found its place in the ecosystem. And the ecosystem is much larger than telcos.
And I think the telcos have recognized that, right? You see all the partnership with the cloud providers. I really had one of my favorite conversations with Chetan Venkatesh. I think it was episode 9 in season 1. And Chetan, his company, Macrometo, is one of the original sponsors to State of the Edge. And so that's how I know him.
But his company basically built a globally distributed database because that's kind of an obvious thing you want, you know, you want to be able to write and read data. out at the edge that you can be confident is going to become eventually consistent and that it's going to be optimized for its use case.
And that, like we talked about earlier, like somebody else is going to figure out where all those bits need to go, where they need to be, how many replicas there need to be in order to be performant and all those things. And he's built an entire company around this. Which I think is just [00:32:00] fascinating. And then I think my absolute all time favorite episode, and this one's going to sound a little strange because we've been talking about edge computing, but this was Matthew Steinberg, who's the CEO and founder of Exhibit A Brewing.
Yup. I learned how to make beer. That's what I learned from that show. But as you know, he's part partner with Dell and they put all these, you know, sensors and compute around his bottling line and his brewing system so that he can make more consistent, high quality beer, which is really his vision. And he doesn't want to become an edge compute expert.
He wants to deliver best beer possible that tastes exactly the way it's supposed to taste when it reaches the consumer at the lowest possible cost and the highest quality and reliability. And he's recognized that, wow, sensors and compute can really take me a long way in diagnosing problems and maintaining consistency and setting off alarms when things don't look right.
But that was, that was my, that was probably my favorite episode. My second favorite was probably Walker [00:33:00] Reynolds, who, we just recently published Walker's interview. Walker, he's got an amazing YouTube channel where he talks about Industry 4. 0. What I really like about him is he's a, he's a problem solver, he's an engineer at heart, and he's very, he's very opinionated, but he's thought about his opinions.
And what I really like is his goal is to help bring back the small American manufacturer. Because he thinks that's the way to do it. Like you say, Oh, offshore stuff is so cheap. Well, look, if we automate. How many humans do you need to make a piece of furniture if your furniture factory is highly automated?
And I just said, like, that's a really interesting twist on how age computing is going to impact the world. Well, you're right. I mean, if we have a lot of small factories and maybe governmental incentives or other, you know, mechanisms for bringing automation at scale into our factories, in our small factories, I mean, it could be an incredible entrepreneurial boom opportunity.[00:34:00]
Especially like 3D printing and all these new technologies, like, I mean, maybe everybody could be a furniture manufacturer and, you know, you buy furniture from the guy or the girl down the street and not from, you know, Thailand or wherever your furniture is made today. Yeah. That'll be
Bill Pfeifer: interesting to watch.
So that's looking back, but now looking forward, you've decided to step out of the host role. Which probably comes with some regrets, I would imagine. Maybe some small ones, maybe
Matt Trifiro: some big ones, I don't know. I don't know. I'm just hoping you do, you do it better than I do, Bill.
Bill Pfeifer: It's not a low bar, my friend.
It's a pretty high bar,
Matt Trifiro: so. I mean, I mean, look, I, like I said, I've been in edge computing for almost eight years. And I did the State of the Edge, and I did the Over the Edge podcast. And so part of it is just like, you know, I kind of, I like, I want something new. But it's still an important topic, and so I don't want the show to go away.
And so collaborating with you and Caspian to keep the podcast evolving and, and growing the audience, I think is, is just an amazing way to [00:35:00] transition out. But my current fascination is artificial intelligence. And again, I'm coming at it from a perspective of sort of a naive user, right? It's like the tricorder.
I see a future where every object in our lives, even the most mundane objects like coffee cups and sports coats. Might be connected to the internet or a network and might have either AI on board or be able to tap into AI that's delivered through an app. No, my coffee's too
Bill Pfeifer: hot. Cause someone hacked my cup.
Matt Trifiro: Yeah, right. Well, yeah, I mean, as you pointed out security, I'm glad you're going to cover security cause that's going to be a, I didn't really cover a lot of security in the first three seasons, but that's exactly right. Like, I mean, what could be made easier, more interesting, more efficient if my coffee cup had sensors in it?
You know, I don't know. I'd have to think through some of those use cases. I would like to know whether my coffee's hot or not. I'd love to know the bacterial count of whatever liquid I'm drinking. [00:36:00] Maybe you wouldn't. Depends. You're right! The little censor on the nut bowl at the bar. You know, they say like the dirtiest thing in the world is then the communal nut bowl at the bar, salty snack bowl at the bar.
Bill Pfeifer: What questions do you still have pending that you haven't gotten answers
Matt Trifiro: to? About itch? About the itch, yeah.
Bill Pfeifer: I mean, or about life, but that's going pretty broad and we're going to go down
Matt Trifiro: a red hole on that. Yeah, no, no, I know. I'm, I'm thinking, I'm thinking about that. I mean, cause I have lots of unanswered questions.
I'm trying to think of the ones that are kind of like front and center. Big questions like, well, when are we actually going to get self driving cars at scale? Right. Like I, the big questions. You know, when are these things at the edge going to really start impacting my life in a way that's transformative and not just kind of interesting?
I think that the question I have comes back to what my CEO, Cole Crawford says, the, you know, the easy button in economics. I think my question is like, who and how is going to crack the code and when of the ease in economics [00:37:00] that just creates the hockey stick? Because we're all working on it and you know, it probably won't be just one, but there's going to be a moment, like the thousandth monkey moment for edge where we'll just say, okay, that's how it's going to get done.
I have some hypotheses, like I think shared infrastructure is a big part of it because I've seen the economics of shared infrastructure. And if you need to deploy expensive equipment like GPs and you can get somebody else to fund them and pay for them as you go and then share them with other people who are also willing to pay for them and do that with just about everything you need for your service or application, that just has dramatic effects on both.
absolute cost, how much it costs you to do it, but also just like your ongoing operational costs.
Bill Pfeifer: Sure. And that takes us to an as a service model for customers, which is what everybody
Matt Trifiro: wants anyway, so. That's my other, my other big hypotheses. And I've seen this, actually. I believe the market is driving this.
Increasingly, corporations of [00:38:00] all kinds, whether it's, you know, Kroger grocery store or General Motors, want to purchase things as a service. And part of that is the CapEx, OpEx, you know, financial jiggering. And part of it is, well, can I narrow the scope of my business so that it's, I'm focused on the things that are most differentiating, where I can build core expertise and focus on those and then outsource lots of other things, but it's also this idea.
That you want the vendor to take more risk. So, you know, I saw an interview recently with a CTO of a grocery chain, and he's being asked about the technology being deployed in the stores. And he said, well, we've put self checkout in every one of our stores and most of our customers use self checkout, but very few of them like it, right?
So it's saving us a lot of money. And customers have, have adopted it, but it's not as cool as if load the stuff into your cart and walk out [00:39:00] of the store and it charged your credit card. Like that's what we want, but that's expensive, especially when you're putting everything, all the compute, all the AI, all the cameras on the store, and you're paying for it for every store by itself, whereas.
If maybe I hang the cameras and I use a 5G network and maybe a lot of the compute is actually running nearby and it's serving, you know, the 20 stores I have in the region and not just my one store and a vendor is providing it as a service, I can now buy automated checkout per camera per month or per customer per month or 5% of, you know, whatever it is.
And it's attractive because it de risks it, and it accelerates the deployment. I think that's a really interesting, it's a really interesting trend that has emerged in all, you know, I mean, back when, when I first entered the workforce, the only things you bought as a subscription were magazines. Now we [00:40:00] buy everything as a subscription, cable TV and magazines, I think were the things you bought as a subscription.
Nothing, everything else you just had to buy and own. You lease a car, I guess that kind of counts, sort of. But it wasn't like today where you can get stakes as a service. And car as
Bill Pfeifer: a service and keep changing the car because it is actually as a service. That's right. Like a lease
Matt Trifiro: was... Well, and a car with a driver as a service.
Oh yeah? Which I guess was a taxi, but it wasn't a subscription. You know, now I can pay a hundred, a hundred and however many dollars a year to Amazon to get free shipping as a service. It's a fascinating shift in the entire economics of everything, which I think is actually good. I actually think, you know, if you think about the origins of money and why that happened, it's like, cause the person who knows how to make baskets.
May not be as good at catching elk. And so if you had seashells that represented value, you could figure out how many seashells is an elk and how many seashells is a basket. And so now you can have this liquidity that you wouldn't have had if you were just trading elk and baskets. [00:41:00] And I know I literally stretched the analogy, but, and now, but I am going to go somewhere with this.
And now with companies offering things as a service and me being able to buy things as a service, it creates more liquidity because I can spin something up and spin something down if I make the wrong technology decision. I can fire that vendor and hire the next one. Yeah, they get to pick up their equipment and go home because like I'm not using their service anymore.
And which makes them more competitive, right? I mean, when you have to compete for your customer to renew every month because they're paying on a subscription, you know, your relationship with your customer and your business model is very different. You focus on retention as much as you do on growth, which, you know, I think a lot of businesses were made like, all I got to sell is one.
And it can be a mediocre product because I only have to sell one because it's unique enough. And I don't think that's the case anymore. Anyway, those are the questions I have is like, well, when is that going to reach its tipping point? And what's that going to look like? And what new, you know, exciting [00:42:00] things are going to emerge from these new business structures and these new ways we figure out how to push things down the cost curve and deliver them ubiquitously.
I just don't know the answer to that. And I have lots of questions there. That's an
Bill Pfeifer: interesting question that I want to explore some now. I mean, When you talk about shifting over to that as a service, you're right. You de risk it for the customer, but you're pushing that risk onto the vendor. The risk still exists.
Matt Trifiro: Yeah, but the vendor should be better at taking it, right? Well, the
Bill Pfeifer: vendor has to have a profit motive. So that when they guess wrong, they still have, you know, some pad and they don't cease to exist. Otherwise, why would you want to be a vendor in that space if all the risk is on you and that's that? But then there has to be some way to make sufficient profit taking on that risk is worth it.
And so where is that risk reward benefit that shifts the whole cycle of risk reward and who owns it and where the profit sits, right? The customer will make less profit [00:43:00] because they're taking on less risk. Because they have to pay someone to take on the risk. That's, that shifts to a really economic
Matt Trifiro: conversation.
Yeah, so let's push on this a little bit. So it may not be that they have to give up profit. It may be that the company that's focused on delivering automated checkout as a service is so good at doing checkout as a service and cost reducing it and spreading the risk over all their clients because that's their business.
That it's not costing you any more than it would, but they've managed to create margin out of deficiency. Yeah. Shifting the value creation. Okay. Yeah. I'll keep my subscription to Over the Edge if you have those guests on. Interesting.
Bill Pfeifer: Value creation versus risk versus profit.
Matt Trifiro: Yeah. That's super exciting, right?
And I'm not an economics expert by any stretch of the imagination. Maybe we'll find one. Hmm. I wonder, I bet there are. I bet there are like... Economic [00:44:00] PhDs that study as a service. I'm sure. I'm sure.
Bill Pfeifer: Interesting. That'll be a fun thing to look for. Challenge accepted. So Matt, we're coming up on time here and I'm sure we could talk about this for days and days.
It would be fun to continue the conversation over time, whether we do it offline or. You
Matt Trifiro: periodically come back. I would love to come back from time to time. That's fun. Yeah. You can
Bill Pfeifer: report in on how I'm doing and how the world outside of the podcast is behaving. And, uh, we'll see, we'll see where that takes us with the conversation, but I wish you the best of luck.
I hope you have a great vacation from the podcast. And find the next big thing that you're working on. I can't wait to see what you're doing next and I thank you for the opportunity to step in as host. This should be a lot of fun. It's been a lot of fun to listen to so far and this conversation was a lot of fun, so I'm looking forward to continuing that and we look forward to hearing more from you as we move forward.
Matt Trifiro: You bet, Bill. I'm super excited like you are.
Narrator 2: That does it for this episode of Over the Edge. If you're enjoying the [00:45:00] show, please leave a rating and a review and tell a friend. Over the Edge is made possible through the generous sponsorship of our partners at Dell Technologies. Simplify your edge so you can generate more value.
Learn more by visiting Dell. com.