430: Interview with John Overton of Kove, Software Defined Memory

In this episode of Destination Linux, we interview the founder of Kove, Dr. John Overton, about the journey from co-inventing distributed hash tables that powered the early cloud to his latest breakthrough Kove:SDM, a Software Defined Memory system that literally lets servers “download more RAM”. Overton dives into the open source ethos that shaped his career. If you’re passionate about Linux, composable infrastructure, or tech that bends the laws of physics, this conversation is a must watch.

Forum Discussion Thread

Download as MP3

Support the show by becoming a patron at tuxdigital.com/membership or get some swag at tuxdigital.com/store

Hosted by:

Ryan (DasGeek) = dasgeek.net
Jill Bryant = jilllinuxgirl.com
Michael Tunnell = michaeltunnell.com

Links:

Chapters:

00:00:00 Intro
00:01:21 Why we turned a 3-minute booth chat into a full interview
00:02:40 John Overton of Kove
00:03:48 Early career & inventing distributed hash tables
00:16:10 Foundational tech that made today’s cloud possible
00:24:56 Sandfly Security, agentless Linux security [ad]
00:26:48 John’s take on AI
00:39:52 The birth of Kove SDM – why it started
01:03:16 Making “download more RAM” real – memory-pool magic
01:17:40 Kove SDM vs. Compute Express Link (CXL)
01:24:54 What are there new challenges in computing you’re excited to tackle?
01:35:39 Lightning round – guilty pleasures, movies & more
01:40:03 Outro

Transcript

View full transcript

Ryan:
[0:00] Welcome to Destination Linux, the Nestromo of open source exploration, where every commit is a pulse rifle blast against the xenomorphs of proprietary software. I’m your Hicks this week, Ryan, locking and loading Linux with my rocking crew, ready to vent closed source into the void. Jill is our Ripley, fierce code-slaying survivor, who can debug a crashing app faster and she can blast an alien queen on an airlock. Her creed, if it’s not open source, it’s just a facehugger waiting to latch on. I’ve heard her say that multiple times.

Michael:
[0:37] Many times. Many times.

Ryan:
[0:39] Michael is our Carter Burke, slippery hardware hustling company man.

Michael:
[0:45] What?

Ryan:
[0:45] Okay, I may have personally picked this character.

Michael:
[0:49] I feel like you had to have.

Ryan:
[0:51] Rewiring servers with a smarmy grin and always pitching proprietary deals. He swears are for the team. His mantra, trust me, code free unless there’s a contract. Michael says that all the time.

Michael:
[1:06] I have never said that one time. Or proprietary deals. I’m also against that part too.

Ryan:
[1:12] Pick the perfect character for you, Michael. I feel like you’re insulting it.

Michael:
[1:17] Yes, with good reason for sure.

Ryan:
[1:19] Well, listen, this week’s mission you’re going to be excited about. We’re interviewing the founder and CEO of Kove. Oh, we have been waiting a long time to bring you all this interview ever since the Red Hat Summit. When we ran in to this gentleman, we wanted to have them on and we got the interview. It’s absolutely amazing. You’re going to absolutely love it.

Michael:
[1:44] We had like a snippet of like three or four minutes of an interview that we had at the booth. And from that point, we realized, no, this needs to be much, much longer. So if you looked at the timestamp of this episode, it’s much longer.

Ryan:
[1:58] We’re talking about the person who figured out how to download more RAM. Like… Literally.

Michael:
[2:05] Exactly. To make that actually a thing, we’re not kidding, really does exist.

Ryan:
[2:10] And one of the people involved in the innovation of the cloud, like the cloud, the thing that the cloud, like they’re a part of it. This is that.

Michael:
[2:17] Cloud thing that.

Ryan:
[2:18] That that so grab your flamethrowers, sync your repos. And remember, in open source galaxy, no one can hear proprietary scream. Let’s lock and load. This is Destination Linux.

Michael:
[2:33] Oh, yeah.

Ryan:
[2:40] So we are here to welcome a guest that we’ve been so excited to have on this show first of all i have to tell a little origin story before i even introduce the guest and that is when we’re at the red hat summit me and michael were walking around we were doing recordings for you know a couple of days that we were there but the entire time we kept having people come up to us saying you’ve got to go to this booth. You’ve got to go to this booth of Kove. And we’re like, okay, we’ll go to this booth. And so finally we did. And we met. I want to welcome to the show because we want to bring you the same conversation. We had Dr. John Overton to the show. We met John at the Red Hat Summit. Like I mentioned, it was an amazing discussion and we just knew we had to bring this to all of you. John, thank you so much for coming on the show and welcome to Destination Linux.

John:
[3:27] Wow. I mean, there’s nothing better after that introduction. Thanks. Great to hear.

Ryan:
[3:33] Well, we love having you here. And there’s so many things we want to dig in because the stuff that you’re doing there at Kove is truly mind boggling. But before we get into all of that exciting stuff, I really want to start with

Ryan:
[3:47] your origin story and understand how your career in tech, because it’s been a fascinating career, lots of innovations. Can you take us back to what first drew you to computing to begin with?

John:
[4:01] Yeah. Well, I think as a kid, I always was attracted to big ideas, big think. And in the beginning, I grew up in the South. At the beginning, that was kind of the humanities and so forth. And then that transitioned in my early 20s into tech. And I loved the idea that you could do things in a sweeping fashion to influence or impact the world in ways that were just huge. And how I got into tech was I finished a graduate degree in theology at Harvard. And I rode a bike across the country. And I was going to write a novel.

John:
[4:44] And when I came back, I had to figure out how to make some money. So I became a temp secretary. And I was a pretty good temp secretary. And so they started sending me around. And I ended up landing at this company called Open Software Foundation. And OSF was the first, I think, the first real open source project that started all of this, right? It had about just shy of 70% of all the computer manufacturers in the world. And I got hired as a temp secretary.

John:
[5:18] And I was working on my manuscript. And I’m not really very good at the humanities. I was always kind of good at math and science and stuff. But the humanities was harder, so I did that. And this work order passed my desk, and it was this outrageously expensive, from my view, this was back in 1986, 87. It was a database project, and it was like $80,000 for eight months or something like that. And I went into the boss, and I said, this is stupid. I could do this. And he kind of chuckled because I was a temp secretary, and this was a big consulting thing with Informix at the time for the company stuff. He said, ah, just go ahead and give it a whirl, come back in a couple of weeks and tell me how. And it was kind of dismissive or whatever, but you know. I was just thrilled to be there, but I, I worked on it really hard and I’d got it done. And, uh, and he’s, you know, you got it done. I said, I think so. And he checked it and yeah, sure enough, I took this job that was going to be an eight month job, knocked it off in a couple of weeks or a month or something like that.

Ryan:
[6:17] Oh my goodness.

John:
[6:18] And, uh, and so then he said, well, you know, we’d like to hire you. What do you want to do? And I had no idea. So I, uh, I got to walk around and talk to people and I ended up having the best bosses in like ever I could have imagined. I think the first guy that I worked for coined the term widget. He ran the HP operating system for HP. Then I worked for another guy who was a professor at MIT. And there was this amazing woman that I got to work with named Elizabeth. And there was this big meeting with about a hundred of the smartest technical people on earth. and she went in and was trying to organize this and inspire everybody to collaborate. You know, at that point, nobody was thinking about open systems, right? IBM had its version, DEC had its version, HP had its version of an operating system. And I saw her inspire this crowd of people. And it turns out she was running development for Apollo, which was at that time one of the biggest workstation. And then another boss of mine was the VP of engineering for the digital VACS operating system.

John:
[7:29] One of my bosses was one of the original authors of the System 5 kernel, and the list goes on. I had people from CMU, MIT, and I just got pushed around, and so I ended up working with, I don’t know, half a dozen, maybe more people, and, And then they all had this big problem. What do we do to organize all of this work? We’ve got hundreds of companies, and we’re building a huge offering. It was called Motif. It was the first commercial implementation of X that had been done, the first windowing system at the time. And they were building what we called OSF1, but it was an open systems, a Unix operating system before Linux. And they couldn’t afford to have writers in each company do this. So they wanted to have a project where every person at every company could establish how many writers they wanted. There might be two or three here, five here. There were a couple of hundred, I think, around the world. They were all working and they needed a way that they could work locally, but they could hit a button and it would grab the documentation from everywhere in the world. And then it would generate typeset books, really like 40 or 50 manuals. And, you know, at that time, you know, XML didn’t exist.

John:
[8:48] You know, SGML was the first, you know, generalized markup language that had been out, which was some of the prototype work. But this is the first time it was done. And I worked on that for about nine months or so. And then at the end of that, you could hit button. And in 20 minutes, it would contact hundreds of places around the world. And it would generate an IBM manual set, a digital manual set, a Hewlett Packard manual set, a Toshiba, you know, you could go over. And for, I don’t know, a number of years, pretty much anywhere you went in the world, I would walk in and it was, you know, to a computer store or whatever, anywhere. And it’d be like, yeah, there’s that thing that I wrote. And I had no background in it. And at the end of it, I really started thinking, And what a gift this was. I got to work with some of the smartest people in the world, and they basically trained me. And that really launched my career. And since then, I kind of never looked back, I guess. I just kept going.

Ryan:
[9:46] I want to take that apart a little bit because there’s some really interesting nuggets in there. When you talk about the fact that you got to work with all of these different people, how much of that was just luck? And how much of that was actually because you were fearless in looking for new projects to take on?

John:
[10:07] All of the above. I mean, I was fearless. I’ve always been fearless in some sense because, you know, I put myself through college. I mean, I just, you know, look, everybody does this, right? You get up at some point, you make some decisions in life, and you go for it. But to say that what it was is just not loud enough. I was at the right place at the right time. And I met these amazing – I mean, to say amazing people isn’t loud enough. I mean, it’s just – I met, you know, people who changed the world. One of the guys at the company wrote the first – you know, he was at CMU, and he wrote the first microkernel for operating systems. It’s the basis for all operating design since then. I mean, I was touching the thought leaders of the whole world at that time. And I worked, you know, I work, I do, I still do. I work really hard, but, you know, without the luck. And I think that’s kind of the way it works. You know, maybe if you work, you know, I think I learned and heard once upon a time ago, you force your own luck if you’re in sports and you train enough. I think it’s the same thing. You work all the time. You eventually find a couple.

Ryan:
[11:16] You eventually get lucky. Yeah, because when I hear you were a temp and then you saw this email come in and it was an $80,000 project and you’re like, hey, I think I can do this. What I see is somebody who was kind of fearless and saying, I’m going to do this thing. I’m going to put my reputation at risk. I’m just a temp. You could have failed. You could have lost everything.

John:
[11:38] It was actually kind of charming because at some point it got around that I was, you know, I was doing some stuff because I did a lot of stuff. That was just one of the projects. And people would come up and they would offer me new opportunities around the company. So I had people approaching me. Hey, you want to work for me? You want to work for me? And so it was just a blast. But really, you know, I think we’ve lost a little bit in the modern culture. Maybe it’s social media. I don’t know what the cause of it is, but maybe it isn’t. But, you know, this is one of the great times to be alive. I mean, look at the opportunities that we all have, right? You just got to, you know, work for them. And that was the first one of my many, many since then. I mean, the stuff we’re doing now with Kove is like, it’s over the top. And it’s the same thing where, you know, we’ve just buckled down, done our business and so on. So I think that the opportunity for global impact has never been larger than it is.

Ryan:
[12:35] There’s a sticky note I keep on my laptop that says, seek the uncomfortable, right? It’s just a constant reminder to go after the things that other people are too uncomfortable to take on. When that person announces a project and you’re like, hey, I don’t want to do that. I think I look at that sticky note like, oh, I’m going to take it. I don’t necessarily know all the ins and outs of it, but I’m going to take it. And I think, you know, when I look at your story, when I listen to it, it’s a perfect example of why that works. Is you weren’t afraid to seek the uncomfortable, the things you didn’t know to work with the people who were really brilliant and things. And I think that’s a really awesome story to let people know about.

John:
[13:08] The first time I did a code review, so, you know, it was this guy, Fred, and he had worked on the System 5 kernel. At that time, that was what made Unix, right? I mean, and I sat down. I remember when I was finished, you know, we looked at the code on the screen together and stuff. And I remember when I walked down the room, I thought, you know, if somebody had said to me nine months ago that I would have had a hundredth of this life, I wouldn’t have believed it possible. I mean, I just sat down and had somebody look at code and work with me who’s like among the best people on earth. And, you know, and he’s delighted to help me because he wants to he wants to allow me to grow into the person I want to be. Like, how awesome is that?

Ryan:
[13:54] Yeah.

Michael:
[13:55] Yeah. And I feel like that has been kind of going on through the whole idea of the open source philosophy, like the whole community is that way.

Ryan:
[14:03] And talk about that all the time, that this is the only place people looking to get a job right now. I know it’s really tough, but we just mentioned this last week, the week before, that you have the opportunity in open source to get a portfolio built working with some of the greatest software writers in the world.

John:
[14:19] In the world.

Ryan:
[14:19] You don’t even have to wait to get hired. You just go start contributing.

John:
[14:24] I couldn’t agree with you more.

Ryan:
[14:25] It’s crazy.

Michael:
[14:27] It’s awesome.

Jill:
[14:28] And John, I love hearing about your history with the VAX machines. I actually have a VAX machine in my collection.

John:
[14:37] Oh, you do? Yeah. That’s crazy jill.

Jill:
[14:41] I cannot believe that i mean.

John:
[14:43] It’s jill so of course you do yeah that’s awesome yeah that’s awesome hey that was one of the best built systems i mean i think you know digital may have made some mistakes that were business mistakes but i think that the art of building systems they were the best in the world at that right you bought things i don’t know if it’s true i heard lore once that um at some point there was a subway in new york and somebody broke down a wall and they found a vax still operating the subway that they didn’t know because when the power went off, it would reboot. And this is like, you know, 30 years ago. I mean, I have no idea. But the psychology of the culture was that, right? Like it should be built to last, right? It’s fantastic.

Ryan:
[15:23] We’ve lost that for sure.

Michael:
[15:24] Jill, does it boot?

Jill:
[15:26] Yes, it does.

Ryan:
[15:28] That’s amazing.

Jill:
[15:30] It comes from one of the universities I went to. I’m a college professor.

John:
[15:37] Cool people. I met at USC with Steph Bailey and he and I shared an office space in, the computer science department. And a few years after that, when I started my first company, he was one of the first people I hired. He was the chief architect and I was the CTO. We worked really, really well together and co-invented the distributed hash tables for location service. And we’ve been friends ever since.

Michael:
[16:07] That’s awesome. So also, I mean, And speaking of all this crazy, awesome stuff you’ve worked in in the past, you’ve also worked on some foundational technologies. And we want to get deeper into those, like technologies that can be attributed to the cloud itself. Tell us about your work there.

John:
[16:24] I finished up at OSF and I went back and I did a PhD. And I went – I’ve been flipping between humanities and sciences for a while. And I went back and I did a PhD at the University of Chicago in theology, semiotics, linguistics, and anthropology. And that was my official job but while I was there I missed computing so I ended up working as a kind of the maybe one of the senior administrators for the computer science department and we ran the computing systems for computer science math and physics so I was in this quad and my friends that I sat down with almost every day for you know eight years were physicists, mathematicians computer scientists economists linguists and theologians okay and we would sit down and I mean at the time when I was at UFC I mean it was amazing like the linguistics department is number one in the world divinity school number one world anthropology number one physics number one or number two math number one in the world computer science wasn’t as strong at that time and it was economics number one.

John:
[17:36] This was the place where the uber geeks of the uber geeks, you know, were like, and people would get together. And I remember that I would get together with the people who are doing robotics and I would share some of the linguistic stuff that I was doing in Dychesis, which I was coming up with a linguistic model to analyze natural language. This is like 30 years ago, right? So, you know, now we talk about that, but I was doing that then. And I was showing them the stuff that I was doing having to do with how robots could move and understand three-dimensional space. Like, what a crazy time. So, so I go through my, my PhD and, and at the end I started thinking, you know, what do I want to do? Do I want to be a faculty? I had a great, I mean, I, I just, I had the best experience at the university of Chicago. I can’t believe, I kind of can’t believe it happened to me still. It was just so nice. But, but I looked around and I thought I couldn’t imagine being a faculty. And I thought, well, you know, what would I like to do? I thought, well, I’d like to swing back to, you know, computing. I had been doing stuff all over the world through the computer science, math and physics departments. And I had been doing consulting for MIT. So I was connected still in the community.

John:
[18:45] And I started thinking, well, maybe I want to go back to that. So I kind of wrapped everything up, but didn’t complete my PhD and then left the university so that I could kind of be unencumbered from any kind of intellectual property. And I went back to the same kind of thing I did at OSF, the same thing I did in my PhD. You know, what’s the hardest problem that I could imagine that I’d like to work on for years? Like, it’s that uncomfortable comfort, right?

Ryan:
[19:09] There you go.

John:
[19:10] What’s that thing?

Michael:
[19:11] I think it’s a little more than that. It’s like, what is the most uncomfortable thing that anyone could imagine? I’m going to do that.

Ryan:
[19:17] I like he takes my saying to the next level.

John:
[19:20] It was cool, right? But I thought, what if you could create, invent a technology that could track everything all the time everywhere? Okay. Now, that sounds like an unimaginably big challenge, right? At that time, the idea of big was a relational database. You could get Informix, Oracle, SAP. I don’t know if SAP was around at that time or not. But those were the big think, kind of big monstrosity of machines, and you’d buy big hardware. And I was thinking very much differently than that. I had done a lot of photography, and I was imagining, think about the life cycle of an image.

John:
[20:00] Right now, this is before he had cell phones, right? So I had cameras and stuff. So I take a camera, it takes digital, not digital, but analog film. I take it somewhere, I develop it, they do something to it. They give it back to me. I have a picture. If I like it, I then go somewhere else. I go back to another place to enlarge the picture. There’s this life cycle of motion that’s going on. And I thought… You know, you have the internet, you know, the beginning of the internet, the university systems are all wired. What happens when that becomes commodified? And it’s no longer that big universities have networks and they can produce objects, but you have devices.

John:
[20:35] You know, phones, cameras, digital cameras, and everything goes, what happens? And you start doing that and you start producing and all of the end devices, what we would call the edge now, all of that became active contributors of data. And this was back in 1996, 1997, you know, before anybody was really thinking about that, what would happen? And if you were to really dig into that, what you would come to conclude is you have to have something that’s linearly scalable. You can’t have something that a search infrastructure that is a tree because the tree goes out of state. And if the tree goes out of state, then you no longer know where something is. You have to be able to compute it. So I came up with distributed hash tables for location management. And, you know, that’s attributed for all sharding of databases in the modern world and all cloud storage. You know, we have had unfortunate legal cases that we’ve fortunately won in and out of court, you know, with Google and Amazon and so forth.

Ryan:
[21:35] So let me get this straight. You kind of have a foundational technology that you worked on that powers the cloud, right?

John:
[21:44] Yes.

Ryan:
[21:45] I think we deserve to be on this interview with him.

Jill:
[21:48] Michael.

John:
[21:48] I don’t know i kind of feel the same way it’s like this is great right but uh jill does but not us no it’s cool because you know and i think it’s you know that that’s the theme i think that you know it’ll lead to Kove when we get to Kove like you know we need to go back to where we build the best right and we don’t do it because there’s ego you do it because it’s the it’s the old like why do you climb everest you know because it’s there or why do you do it because it’s possible right let’s imagine what the possible is and let’s go do it i like.

Michael:
[22:21] How you’re imagining what the like this this thing that people never even was thinking about being possible you’re you’re trying to come up with this and while most people on the planet at the time were just worried about their aol minutes.

John:
[22:31] Yeah yeah that is what was going on yeah i mean that’s historically it was aol, yeah that was what people were and you know and the view that i had i mean i i remember i had a And a little bit later, I was a year, a couple of years after that, the math department at Yale asked me to come by and give a talk. And and I was in front of, you know, all the math department at Yale. And there’s this, you know, amphitheater of people. And I’m showing what we’re doing. And this guy gets up and starts screaming at me and saying, this is impossible. OK, you can’t do this. I’ve worked on this. For two decades, it’s impossible. And I looked at him and I said, I don’t, you know, I don’t know what to say. It works. Like this is not a debatable fact. It works. Right. And, you know, and I think a lot of stuff that I’ve been interested in doing, it’s finding the people. And I think that’s why the open source, there’s a lot of alignment there. Right. Because I think that the people at open source, you know, they have that appetite for adventure and for sacrifice and doing the impossible.

Ryan:
[23:48] Right?

John:
[23:49] And doing the impossible. Right?

Ryan:
[23:51] Because who would have thought this little project, Linux, could have taken over?

John:
[23:56] Yeah, that little Linux thing. Like, who’s this guy, you know, Torvald? Not Minutes. Exactly. Richard Stallman who? Okay, you know, like.

Ryan:
[24:09] You know, all those companies were probably laughing, you know, like, this guy will go nowhere. This project will go nowhere. You know, we’ve got all of this proprietary. We’ve got all this technology. We’ve got all this money. How could this person possibly compete with us? And then they did.

John:
[24:26] It’s because that person didn’t. It was a movement. I mean, I think that the most exciting thing about tech, going back to kind of the origins of this, like, you know, what makes tech really exciting is, you know, the best of the best don’t care where you come from, who you are, whether you’re purple or red. From mars or not from mars or what your preferences of this they just want to know are you really kind of a geek at heart want to do stuff yeah and let’s

John:
[24:51] get to work and if you are let’s get to work yeah it’s awesome love.

Michael:
[24:56] It as linux users we know what’s up security is non-negotiable but with threats getting smarter your security tools need to keep up without dragging your system down of course and traditional agents i mean they slow you down they leave blind spots, it’s time for a smarter approach. And that is why Destination Linux is proud to be sponsored by Sandfly Security, the revolutionary agentless platform designed for Linux. Go to destinationlinux.net/sandfly to see how Sandfly can transform your security strategy. Now, Sandfly doesn’t just detect and respond. It revolutionizes security with SSH key tracking, password auditing, and drift detection, Kovering threats from every angle. Whether your systems or in the cloud, on-premises or in embedded devices, Sandfly ensures they’re all secure without the headaches of agent-based solutions.

Ryan:
[25:43] You know what, Michael? You can also get a free trial of the software because you’re a listener of this show. And by the way, I was lurking on our Discord. And this is why I love our community. I lurk a lot on our Discord. I don’t say a lot, but I lurk a lot. And I noticed someone was talking about how cool Sandfly is. And this is how innovative our community is. they had created like an emoticon of sandfly’s logo to put in their text in there and i was like how cool is that that sandfly is a big deal but as a listener to our show you can get a discount by using the discount code destination and that’s on the home edition and it’s 50 off that’s 50 off that’s right edition of this and.

Michael:
[26:30] Remember that coupon code is destination and you get 50% off the Homelab edition. So dive into the future of Linux security at destinationlinux.net/sandfly. That’s destinationlinux.net/sandfly and see how Sandfly can transform your security strategy.

Jill:
[26:48] So you have been involved with so many incredible projects. We would love to get your take on artificial intelligence. Is it the revolution everyone thinks or is it 90% marketing and 10% reality, as Linus Torvalds has recently said.

John:
[27:08] Innovation is innovation and computing is computing and they don’t change a whole lot. And they change a whole lot. OK, that’s what I think about AI.

Michael:
[27:16] I like that.

Jill:
[27:17] Nice.

John:
[27:18] Innovation is innovation. Computing is computing. Things change a lot and things don’t change a lot. OK, it sounds like a lot of words, but in a lot of mumbo jumbo. But here’s where I think we are. So every technology that gets some splash has some utility. OK, it just does. Or it wouldn’t get some it wouldn’t get some splash. Right. So you can’t say there is no utility here. On the other hand, I mean, you know, we’re talking about big graph theory. Big graph theory has been around for a long time. Right. That’s not new. OK. We’re talking about, you know, conceptualizing a universe of probabilities. We’ve been working on this kind of stuff for a long time.

John:
[27:56] So at some level, I’d say, yeah, it’s not really a big to-do about anything because it’s just more innovation and more computing. Now, on the other hand, socially, it has lit a fire, and now people are thinking socially in a way that they wouldn’t think is possible, right? Could we have self-driving cars that we can believe in? Can we have, you know, drug therapies that can be created by labs that we build that build labs to think through some stuff? A friend of mine at Argonne who got at Gordon Bell in supercomputing, you know, when COVID came out, he came up with a model using AI and it predicted every derivation of the COVID virus before they all happened. That’s cool. OK, you know, like there are things that are really real that it can do. The problem is people are people and sometimes people are people and they, you know, get wrapped around things and they lose their frankness. They lose their ability to be straightforward and honest. They like to follow, you know, pack where, you know, pack animals and herd animals at some level. And you get a big, you know, you know, it’s like you all are the anti that. Right but you’re infected with the open source craziness of wanting to do real things right that’s that’s our lingo right i mean we’re all the same.

Michael:
[29:23] I like the phrase of saying we’re infected with the end with the open source it’s true.

Ryan:
[29:29] Though it’s a.

Michael:
[29:30] Philosophy right yes it’s.

John:
[29:32] A perspective of how you want to.

Michael:
[29:33] Live your life right and it’s what got me into like the being so ingrained into the ecosystem and the in the platform and everything is that it wasn’t just because i wanted to use something that, you know, some that I just found an alternative to windows or whatever. It was because of the message behind it in the mission behind it. Yeah.

John:
[29:51] And that’s where I would say, um, that’s where I’d say that, you know, um, there’s value of saying 10, 90, 90, 10, right? Because the social messaging does give life to things. And you can’t say it doesn’t. And I think the open source is a perfect illustration. When I was at OSF, I mean, we were doing open source. And the reason that people were doing this is because they couldn’t beat the AT&T System 5 and Sun Microsystems, their alliance, because they were controlling the Linux through the licensing. And so they wanted to come up with an alternative. It was built off of capitalist principles to open it up. It wasn’t what happened later, which is, no, no, this should become a public good, right? So there’s one that didn’t have the messaging behind it. But when Linux came, the messaging came with it, right? And now look where it is. And I think AI is a little bit the same way, right? To the degree that it inspires our imagination to what is possible by joining together to try to contribute to build things better, I think there’s lots of opportunity in AI. To the degree that it becomes the new search engine that you’ve got to pay $10 a month for.

Ryan:
[31:02] I think it’s – They figured out how to monetize the search engine.

John:
[31:06] I mean, just saying, right?

Ryan:
[31:10] You’re not wrong. I remember saying that in an early episode that this will replace Google and when AI was really first starting to take steam and watching that actually take place has been fascinating. Because now you get an AI result even in your search browser that honestly most of the time is now answering my question better than my search could. So it’s changing things, you know?

John:
[31:35] Right.

Ryan:
[31:36] Yeah.

John:
[31:36] But there you go on the 90-10-10-90, right? I mean, so I guess it depends on which vector you’re looking at the object for because I think from the point of view of people saying, oh my God, there’s all of this amazing new innovation that has never been done in the history of computing. For all of those of us who do real computing, this is an evolution. It’s not a revolution. It’s not this, you know, you can look at it and say, yeah. You know, this was coming, we knew it was coming, right? I mean, there’s not a lot of surprise there. On the other hand, it does seem to be percolating imagination in the public. And that’s good, especially for the people who are highly introverted geeks that, you know, don’t want to do those kinds of marketing things. Right. So, um, so I don’t know. Is that answering your question? Do you think or not? I’m not trying to talk around it, but I think there’s more, there’s more to the question, right?

Jill:
[32:27] Yeah.

Ryan:
[32:27] No, I think it’s perfect. And it kind of leads to my next question on ai is you mentioned people being people and so when you think about people being people and you work with some of these massive companies doing amazing things and you think about all the innovations that you’ve seen do you think ultimately ai is changing the world to be a force for good or do you think we’re going to go through this period where it’s going to be bad and then maybe good so.

Michael:
[32:55] Or just straight up uh robot overlords.

Ryan:
[32:58] Yeah i i’m.

John:
[33:00] Uh i don’t know if i’m considered radical or not radical on this so i’m not sure where i land in the universe but um i you know when i was at the university of chicago i play squash on the squash courts and below those like 20 or 30 or 40 feet of concrete was where the atom was smashed or.

Ryan:
[33:17] Broken split for the.

John:
[33:19] First time like i play above where that happened okay Right. I mean, it’s super cool. Right. And, you know, and there you go. So is that good or bad? Right. So I work with, you know, one of my friends that I work with on a project is a physicist at Stanford and is working on fusion, you know, fusion reactors. Okay. The fission reactors, we can have debates about fission reactors. Right. But because of the splitting of the atom that let the fission work, now we’re doing fusion. If we ever get fusion working, it’s a big deal, right? That came from what? Okay, oops, I forgot to mention, there also was Hiroshima.

Michael:
[34:00] Yeah, true.

John:
[34:01] You know, I think that the technology, they’re always going to be good people and evil people, and they’re always going to be kind of halves of a whole. Unfortunately, I think that AI has the opportunity to create so much good done well. But it’s inevitable that they’re going to be people that figure out somehow to take advantage of it and misuse it. You know, one of the reasons I like the open source community so much is it’s a lot harder to be just an outright jerk in the open source community because you get ejected, right? If you don’t do real work that’s legitimate work and people recognize that, you know, nobody wants to work with you because everybody’s doing something for public good. And when you get that vibe going, you go all kinds of cool places.

Michael:
[34:52] Yeah i love that yeah yeah absolutely and also it’s kind of funny because there’s a lot of times where you know you’ll see some drama happening with various open source and then the way it kind of ends is that whoever is starting that drama is just no longer part of that project and then it just kind of fizzles out cleans itself out yeah well.

John:
[35:14] That’s what’s cool right i mean And again, that goes back to the original, you know, the 90, 10, 10, 90, you know, I think, what I think is wrong about what people are concerned about with AI, you know, the AI overload, I mean, maybe that will occur, right? It’s too early. But what we have currently is AI isn’t going to do that. I mean, I think we’re a good hop and a skip away from that yet. But what we aren’t is the idea that we can start using technology aggressively for social good, social advancement. I mean, I think that’s already the idea of social communities that are electronic. Think about that. I work with on a regular basis people in probably five or 10 different countries. Tell me a time in history where you could vigorly do that yeah and.

Michael:
[36:02] Then that kind of started with the whole process of the open source community because like.

John:
[36:07] The chat.

Michael:
[36:08] Rooms and everything that.

John:
[36:09] Just all started.

Michael:
[36:10] That way yeah.

John:
[36:11] Yeah the work you did with the cloud i mean now we’re.

Ryan:
[36:15] Listening to in 190 countries.

John:
[36:17] Yeah because.

Ryan:
[36:18] Of the cloud.

John:
[36:18] 190 countries i.

Ryan:
[36:20] Can name about 25.

John:
[36:21] If you hit a timer Like.

Ryan:
[36:24] But we go through the list and we’re like, how? And it blows our minds. That’s pretty much most countries.

Michael:
[36:29] If you think about it.

Ryan:
[36:30] It’s crazy.

John:
[36:32] Most countries in the world.

Jill:
[36:34] Yeah.

Michael:
[36:34] It’s amazing. It’s crazy.

John:
[36:38] I think both of those are true, right? I think it – I do think that in general, you know, here’s where, you know, probably my, you know, kind of theological, you know, leaning is, is that I think in general, things move toward better in general. But it doesn’t go like this in the best case, it goes like this.

Ryan:
[36:57] Yeah. I like that. That’s a good way of looking at it. Yeah. You’re going to have your ups and downs with it.

John:
[37:01] You always will.

Michael:
[37:02] Right. But ultimately the main line is going up. Yeah.

John:
[37:05] Yeah. But I do. I mean, here we are talking, uh, you know, think of the serendipity that happened. You were at a conference. I was at a conference. We had a chat. Okay. You know, the chat went in a certain way where we all thought this would be really fun to do it again. Right. Um, look at all the things that had to happen. For that to be natural that happens on a regular basis right and i’m not attributing that to divine forces and stuff like i don’t mean it that way at all i mean it that that’s the way society works that in general you get these currents of thinking or like the open the open source is a great illustrations the currents of things somebody says i have a project anybody want to work on it right and then people get together work on it and then you can have like oh i don’t know a free operating system that’s putting almost every device on earth because of an idea right yeah Or I could come up with an idea and say, you know, before anybody was seeing it, you know, what if we could find things with, you know, a one try, you know, you make a query, you fail, and you get it right the second time. Okay, that’s basically what a distributed hash table is. So you can disperse everything. I can find anything anywhere, anywhere in the world on the internet. And I can do it within, you know, less than 100 milliseconds. Right? That’s possible because of the society of being at a school where I could be with all these other kinds of people or being on a podcast with you guys and lady where you have a vax machine. Like, it’s awesome.

Ryan:
[38:30] Oh, awesome.

Jill:
[38:31] You know, I’ve always kind of looked at it in terms of evolution that we’re on, yeah, those… This super information super highway right and but linux has perfected it That’s an interesting way to do it.

John:
[38:50] That’s awesome.

Michael:
[38:52] And also the technology.

Jill:
[38:53] Yeah. Linux has perfected it and we’ve gotten to be a part of that history.

Michael:
[38:59] Yeah and also the technology is growing at a pace that is kind of like just staggering and that’s what my favorite part about is that when um when people talk about like a year a year in in reality is like is is a year but a year in technology is like a decade of reality so it’s the advancement is so much so it keeps going faster and faster right and i.

John:
[39:23] Think it is related to the kinds of tech these kinds of global technologies that we’re talking about right 100.

Michael:
[39:28] Yeah absolutely and also like the we were talking about being able to meet at a conference then have this interview this is also in a way that we’re the technology is going to the point where we can do a live communication over video and have it all recorded and have it all like structured.

Ryan:
[39:45] From all different states.

John:
[39:46] Yeah, all over the- From different places around the world and it all just works.

Ryan:
[39:50] Yeah, it’s crazy.

Michael:
[39:51] It’s crazy. But I want to get back to your story because there’s so much to talk about. And I want to start with what led you to start the Kove?

John:
[40:03] So I think the consistent theme from the beginning to the end is big think. I mean, it is exciting. It’s not hard to work really hard on things that you think are really exciting. And I was fortunate to do that at OSF. And then when I kind of cooked up the concept of finding anything anywhere, you know, so I started back in about 2003 and I thought, you know, what would be, um a nearly unimaginably hard problem to solve that’s the stuff that i think is just really once.

Michael:
[40:35] You solve one that’s incredibly impossible let’s.

John:
[40:38] Go to.

Michael:
[40:39] The next one.

John:
[40:39] So what would be something that was so scary that i i probably couldn’t do it and therefore i would want to try and try and try and over and over and over so um you know we’ve known for a long time that memory operates at the speed of light right so you can move an electron i don’t know what the actual number is It’s like 9.98 inches. It’s a metric system or something on a piece of copper. And it’s 13 inches-ish in a vacuum. There’s some number, right? So you put memory, and memory feeds the CPU. So you put memory 100 feet away, and you’ve got 100 nanoseconds, right? So you’re doomed, right? This is the so-called speed of light, distance of cable problem.

John:
[41:21] And pretty much every company has taken a swing at this. And I think we’re the first and the only, I think, to have ever figured out how to hide that latency. And so we spent probably from the first five years just looking at the physics of the media that were available. This is before Flash came out, if Flash could be sufficient.

John:
[41:45] I came up with a board design thinking we’d had to encode everything into silicon to be able to do it. We were getting ready to put down a million bucks on building a board. Then we decided, well, if we do that, it’s not a real long-lived company. It’s a point solution that will get flipped. And I feel thrilled to be alive. I wasn’t looking to build something and sell it. I want to build something to build it. That’s part of why we’re in Chicago. right i i want to you know i want to feel they have everybody come into work and feel proud that they’re building something that’s awesome so we um so sorry we started looking at that and then we started looking into smp symmetric multi-processing systems thinking well cray and sgi were doing you know really good work on those kinds of systems and maybe we could do it only in software or we could do it in a software hardware hybrid and then we decided well that can’t scale linearly and therefore it won’t ever hit the scale that we want to hit. Then we found out ScaleMP about four years after we had been pursuing that came out and then they produced the first virtualized stuff. So we were doing all this work thinking, what’s the big problem?

John:
[43:00] Where we landed on this is, if you look in the history of computing, here are a couple of axioms. I do not know if these are true. These are the ones that we believe to be true. So note, I am ducking a little bit on this, right?

Michael:
[43:14] Fair.

John:
[43:15] But if you look at it, first of all, you often go to some kind of super well-engineered system or engineered hardware, some kind of hardware gadget, a mainframe, right, or whatever. And then through time, it starts swinging to the other direction toward a commodity, right? So you have a mainframe, then you have a workstation, then you have a PC, then you have an open systems architecture chip.

Michael:
[43:44] And then eventually you have a Raspberry Pi.

John:
[43:47] Right. And then you have a Raspberry Pi, right? Okay. And then when you get to the very end of that hardware innovation, someone says, wouldn’t it be cool if I can virtualize all of those things? Instead of having a disk drive that has to sit inside of a box, what if I could take a whole bunch of disk drives and make them look like a big disk drive? Or in the case of some multi-metric process, some SMP, in case of that, instead of looking at this from the point of view of, well, I’m going to have a bunch of cores on there. What if I could take cores on a bunch of different servers and make them all act like one big server? You could go and you could say, instead of having a bunch of PCs, why don’t I have a Beowulf cluster? And now I can have an inexpensive supercomputer. So if you look at this, and this was what hit us in 2005, every technology has been virtualized in the computing stack. CPUs are virtualized. VMware did that. Storage. Networking. Okay. CPUs. The only thing that was never virtualized was memory. And so the first thing we did is we started thinking, can we do it and can we physically do it in memory? So what do I mean by virtualization?

John:
[45:02] Let’s imagine I have a computer and I’m going to kind of keep it in a kind of simple, easy to frame perspective. So I’ve got a computer and I’m going to put a lot of memory on there because AI needs a lot of memory. So memory costs 65% to 85% of the cost of that server. So if I buy a $20,000 server, $15,000 of that can be memory, maybe $14,000 of that. That’s stuck in the box. Now, what’s the probability I can use the memory all the time? You can’t.

John:
[45:35] Basically, the most expensive stuff I’m buying, I can’t use all the time. It’s not a very cost-efficient, so I’m going to buy my engine, okay and i’m going to use it at 25 all the time or i’m going to buy my uh my the rubber on my wheel and i’m going to get a brand new wheel when it gets to be 25 tread used okay yeah the whole approach doesn’t make any sense well if you go our approach you have some servers and you can aggregate all of that memory together so i can pool it this is the way we would the lingo we We put the pool over in the corner, and then you can have a server, and it has a much smaller amount of memory inside of it. In fact, it could be a 64-gig server, and it can raise its hand and say, you know what, I need a little bit more memory, or I need a lot more memory. And then the process makes the request, and then it can get the memory from across yonder. Now, the problem with that is that everything goes to hell because of the speed of light distance of cable. So that’s 100 feet away. That’s 100 meters or 100 meters away. That’s 300 feet away. That’s 300 nanoseconds. A processor operates in the core at one and a quarter nanosecond. Okay, can’t work. Everybody agreed that it can’t work. And I mean, the biggest of every single big, and I know most of the CTOs and most of the big companies, they’ve all tried it and they’ve all agreed. And the big names that we have lawsuits against, unfortunately, also tried it.

John:
[47:04] And we figured it out. And so what we can do is, with software only, you can have any x86 server that’s supported by Linux of any variety, no matter how big or small. It can be 10 years old. It can be 12 years old. We have systems in production that are 12 years old and supercomputing facilities that are servicing the latest and greatest CPU with no latency penalty. None. Okay?

Michael:
[47:31] It’s crazy.

John:
[47:32] Right now. Right. I mean, the world’s monetary system, SWIFT, which does, I think it’s it’s I think it might be five trillion dollars of money passing every day. OK, is built on this approach. OK, that’s like 93 percent of all money transferred globally every day. I mean, there are things that you can do. Once you can control that last area, that thing that hasn’t been virtualized, you can do things that you would never imagined to want to try. Build bigger servers or use smaller servers. Change the design of a data center. Change the design of an edge instance. Change what inferencing means. Change how you’re going to build computation systems. Change, I mean, you can just go on and on. Your creativity can go crazy. And for the people who are the geeks here or in big companies, you want to change your company, somebody reach out to somebody at Kove, okay, and you can get a download, and you can install it on any Linux device that you’ve got in your data center, and you can see that this stuff works. In the case, I’ll give you some stats. So in the case with Swift, they were doing some modeling on a very popular, but I can’t name who, virtual machine manufacturer that people aren’t pleased with right now. But I can’t say names.

Michael:
[48:55] There’s actually a couple of those now.

John:
[48:57] It’s whoever it is. And I don’t want to cast aspersions because everybody makes decisions. But they were running a job and it took a certain amount of time. So they installed OpenShift plus Kove and then they ran the same job. And it was 60 multiples faster. Okay? Not 60%. that’s a 60 times computer difference oh wow okay but software okay so no no no it’s.

Michael:
[49:26] Not it’s not using software it’s.

John:
[49:27] Using magic let me give you another one let me give you another one that we like to talk about so redis open source redis it’s the most widely used database in the world right everybody uses it so it’s a single threaded application you have uh in a dma is two inches on the motherboard and then it does a hash lookup. Like it’s hard and it’s single threaded and you can put it on one socket and there’s not even Pneuma overhead. So like you can’t reduce that a whole lot more. It’s like really simple. So we have, we can run that benchmark with our software and we can put a memory target 24 meters away plus two switch ups. And all of the list ranges, we are between 30 and 70% faster than the local memory in the server, even though of the memory is 24 meters away. We can hide, we can hide latency 150 meters away. That’s a football field and a half. So you can put memory on the other side of the data center and it will be the same speed or faster than local memory. Now it sounds almost unimaginable, except that we, it’s being used for critical mission systems in the world all over the place right now. So you can’t debate it. Right. But, but it’s super cool that. So going back to the original, like, you So what gets us going?

John:
[50:44] I want to live in a country where you dream to do the unimaginable. And then you can do it, right? Let’s put a man on the moon, okay? Let’s build an open source civilization that can have innovation, innovate at a rate that nobody else can do. Let’s put memory in places that no one ever thought. Let’s change the cost basis of computing.

Michael:
[51:06] Let’s make the meme download more RAM real.

John:
[51:09] Yeah.

Jill:
[51:10] Yes.

Ryan:
[51:10] I really feel like in my mind.

John:
[51:12] That’s what John was doing. If you took this and put it in the data center, you can get, so let’s say you’re in a compact space and you can’t put any more power servers. You can install our software and double the computational efficiency. Double it with software. Like it’s super cool.

Ryan:
[51:29] So John, we don’t usually do gotchas on the show, but this is one chance we’re going to do it. Do you feel guilty for stealing this from Martians? That’s crazy why everyone was telling us by the way for those listening of why we had to go speak to Kove because everyone would go there and then and then you’ve got the receipts to back it up so this is just john talking he’s got independent receipts of companies that do these benchmarks and tests backing this up and we’re all like this is impossible how is this possible everyone was blown away at the conference. And there’s a good reason why, because it’s just, it’s crazy.

John:
[52:09] I have to say one of the thrills of my job, you know, the majority of what I kind of think I do these days is I try to help educate people to think the unimagined is possible, right? You can, you know, we have a logo, you know, like achieve more, right? Like, what if you could go back to the time where you could say, I really do think that my expectation of my job is to come up with things that are going to make it better. Like not, not the little five or 10%. So I get my bonus, but that open source psychology, I want to make things better. What if what if we could give you software to do that? Like, yeah, that’s as cool as it gets. Right. And oh, yeah. And the job that I have is to educate people so that they can start to open their their minds. A part of this goes back to this kind of, you know, very unconventional background of sitting out in the University of Chicago with physicists, economists, mathematicians, computer scientists, theologians, linguists, anthropologists. you know, those are my friends. And, you know, and when you’re in that kind of community, nobody’s sitting there and no one cares. It’s like, that’s what used to be good about the academy. I’m not sure if the academy is as good at that now. I think that what used to be in the academy is kind of in the open source, which is why I love the open source community. Like nobody cares what school you went to. What do you do? Are you useful? Do you do stuff that matters?

Michael:
[53:34] Do you write good code? That’s all I care.

John:
[53:36] Who cares? It’s awesome. And I think if you get to those people with the message, what if you could think of the unimaginable and make it real? That’s what I think Kove SDM is from the perspective of Kove, right?

Jill:
[53:54] Yeah. John, I am actually so excited to talk to you because I’ve known about this technology for many years. And one of the reasons is not only did I teach computer animation and motion graphics for 30 years. But I’ve worked in Hollywood. I’ve done commercials. I’ve done movie trailers. I’ve worked with Alias Wavefront on SGI machines.

John:
[54:20] Well, that’s just awesome. Yeah. So this is how you have your collection?

Jill:
[54:27] Yes.

Michael:
[54:28] That’s what we were saying earlier, that she deserves to be here, not us.

John:
[54:32] Yes.

Jill:
[54:33] Oh, well, one of the things I used to, when I’d set up render farms, I would send up RAM disk render farms because it would always render a lot quicker to a RAM disk.

John:
[54:46] Yep.

Jill:
[54:47] Well, I mean, I needed this back in the, in the eighties, early nineties and early two K’s when I was working on a time machine next Jill.

Ryan:
[54:57] So we’re going to get on.

John:
[54:59] So the thing that, you know, the stuff that you’re talking about, which is super cool, like we had a project where I don’t know what the actual math, but, you know, they use these rotational diamonds to create opacity and color and all this kind of stuff. Right. And when you start doing that at scale. It turns out that there’s some components of that that are reused. Okay. In traditional rendering farms, you pay that cost over and over and over. Right.

Jill:
[55:23] Yeah.

John:
[55:24] If you could keep that, all of that in DRAM.

Jill:
[55:27] Yeah.

John:
[55:27] You could have a central. I mean, it’s exactly that story. Right. It’s awesome to meet you. Oh yeah.

Jill:
[55:32] Same here. It would just be, be nice to have in memory. You’re, you’re rendering, you know, six sides of a cube with ray tracing, you know, keep that in memory for me. So that the next project I have with 1,000 objects.

John:
[55:47] It’s easy peasy to do, in fact.

Jill:
[55:50] Just saying.

Ryan:
[55:51] Easy for you, John.

Jill:
[55:56] So, John, for our listeners who might not know Kove, can you tell us about your company? And we already know the innovation that it has brought to the world. This is the next information superhighway, people. This is.

John:
[56:12] It’s super cool. So thank you for asking. I mean, I, you know, especially in this kind of a, you know, this kind of a chat, it’s easier to kind of say what you really think a little bit, like not just the, I work with some of the coolest people I’ve ever met in my whole life. I mean, on a daily basis, we have people, I think it’s 13 states now, we select, I don’t want to say misfits, because I don’t think there’s anything about misfits in anybody who’s in our company. I think they’re just as awesome as they get. But it is that psychology of don’t come to work here unless what you want to do is change the world. And I don’t mean this as a little… Jargon thing that you get in Silicon Valley that people put on the, you know, front step of the, or the front door, you know, we’re here to change the world. I mean, like, don’t do it unless you, unless you really want to do something, don’t come. Right. And, and it has inspired, you know, you know, the ability to spin innovation at a rate I never saw coming. And, you know, the people that we have, which is good. It’s also inspired, you know, kind of access in a way that I also, So, you know, I get to work with some of the most mission critical infrastructure in the world, in the globe.

John:
[57:37] And it’s on a merit by merit basis. And it’s off of, you know, telling the truth about what you do and showing it. Right. And what I so I think, you know, going back to Kove, you know, what, you know, what I’d like to say kind of about anybody, If you’re the kind of person that you really do want to dream big, for real, and you work in a company that that could matter, you should reach out to us because we’ll work with you. And we won’t try to sell you a product. We are not that kind of company. We are trying to build the best stuff in the world. And if you’re inclined like that, you know, we become fast friends because it’s fun. It’s really fun. And we’re a little bit different than open source.

John:
[58:20] Is, you know, this is the, like, I can’t, I’m the vice chair of something called the Enterprise Neurosystem, which is a, it’s a, I don’t know, it’s a 50C31, 503C, whatever the charitable thing is. But we’ve got about 180 members from most every major company I work with, one of the CTOs of IBM on there, Stanford professor, faculty, all kinds of people. And we do work with Uganda, Tanzania, the United Nations. It’s all free, open source, everything that we’re doing for, or, you know, a bunch of different kinds of research. And that’s as open source as it is open source. And people sometimes say, well, you’re proprietary. How does this fit? Okay. And the answer, and this goes back to the Kove question, the answer is everything we do uses an open interface. So if you want to use Kove, you use Malik and Free. Okay. That’s on every operating system in the world. You don’t like us? You can unplug us in a microsecond, okay? If you like us, you can use us. Your choice, okay? And don’t use us unless our value is so valuable that you want to pay us, right? That’s a very different attitude. I’m not figuring out how to turn the crank. It’s that we have invented something that lets you cut your computational overhead 50%, your power 50%.

John:
[59:44] To do things like you were talking about before, Jill, that you never imagined, right? What if you could do that on a regular basis and do it at scale? So once we’re installed, I mean, there is no scale limit. You know, you can get in a 40-U rack that’s eight years old, you can get a terabyte per second of I.O. At 1.7 billion IOPS. That’s eight years ago, okay? That takes 625 racks of EMC VMAX to be able to do that today.

John:
[1:00:17] From eight years ago. So if you want to go back and you’re the company, you want to build something awesome, whether you think that we provide value or we don’t, come talk to us because we’ll tell you the truth. And if there’s a way to do it, we can go build really awesome stuff. And that’s where the open source community really bleeds through in our personality type, because we really do look for people that want to build the best stuff in the world.

Ryan:
[1:00:46] So Kove is working on not only this, but I assume when you have the best people in the world, you have all of these geeks, that there are other things that you guys have that you’re thinking about or doing. Is there some of those things you can share with us yet, or are there still?

John:
[1:01:03] Oh, we have so much cool stuff going. I hope it’s a perpetual motion machine, and we keep attracting the people that we’ve been attracting. But, you know, let me give you some things that one could dream about.

John:
[1:01:21] So what if I didn’t define a server the way I define a server now? What if a new version of a server was one CPU, okay, one DIMM and one fan, that big, okay? And that could be arbitrarily scaled in real time to anything that you needed, whatever it is that you needed, okay? Now imagine that there’s no limit to the scale of that. okay that changes data center design it changes rendering farms it changes uh genomics it changes um um you know at some point machine learning so right now think about machine learning one way to do it is the way that you would get with nvidia big huge hunk and iron okay i would argue that in the history of computing things go to big iron and then they go to commodification and virtualization. What would that virtualization look like if you weren’t thinking that your H100 or the next version of that was $20,000 or $30,000 times eight in one box times this times that, and all of a sudden I spend a billion dollars? What if that whole approach could be commoditized so that you were thinking about it the same way we did one day previously with Intel? Wouldn’t it be great instead of having to go to IBM to buy mainframes like if I have PCs? I think that’s going to happen with AI. And I think the way that happens with AI is through memory.

John:
[1:02:46] And, you know, I’d say we have a, you know, we have a pony in that race and that’s something that we’re really, you know, interested in. That’s maybe a little opaque, but that’s all I can say because of patents and stuff. But I think there’s a lot of space where you could start imagining that things that take big iron to achieve, you could do not on big iron and that you would look at us as a way that you can commodify very high-end infrastructure into less expensive infrastructure.

Ryan:
[1:03:16] So you mentioned that Kove, it’s amazing, and you mentioned that Kove can take a server, like a 64 gigabyte server, and make it basically process with 100 terabytes of memory if you wanted, right?

John:
[1:03:30] It’s essentially 64 petabytes of memory per process.

Ryan:
[1:03:37] That’s mind-blowing.

John:
[1:03:38] That’s the actual limit, is 64 petabytes of addressable memory per process, not per server, just to be a geek.

Michael:
[1:03:45] Per process.

Ryan:
[1:03:47] No, I love it.

Michael:
[1:03:48] It’s awesome.

Ryan:
[1:03:49] I know our audience loves it too. Can you give me a real-world example? You’ve given a couple, but maybe another real-world example where you’ve really changed how, you talked about how you could change the way a data center operates, how it looks. A real-world example where that’s happened, maybe within the Linux community or ecosystem if you have something.

John:
[1:04:07] There are many. There are many. Okay, but let me give you one that I think is kind of especially juicy. The easy piece.

Michael:
[1:04:14] Always love juicy.

John:
[1:04:15] Okay. Easy peasy ones, you can say, well, I’m going to make a big server, okay? And, you know, I would argue that 64 petabytes of memory per process is a big server. You can do that, okay, if you want to do that, right?

Michael:
[1:04:28] I mean, I guess you could say it’s a big server, yeah.

John:
[1:04:29] Yeah, yeah. I mean, you could, you know, but who’s counting among friends, right? So, you know, there you go. I mean, so you could. And there are some instances where people really do need big servers, that is. the one where people, what people really need, which is much more exciting. And I’m going to use the AI because we’ve already talked about AI. Okay. And therefore this fits the thread of the conversation that we’ve had. So let’s imagine that I’m a programmer and I work at a big oil company. Just going to make this, I’m going to deliberately obfuscate some things so that it makes it so I can talk about it. Um, um, you know, and, and so it’s not an oil company by, Anything that I say, it’s not that kind of company because I’d get myself in trouble. But let’s just say I’m a big oil company, not an oil company, and I’ve got 50 data scientists. They’re the most expensive resource that I have, right? They are the premier, smartest people I can get for that specific industry, and I want them to be super productive. Go figure, right? I’m paying them big money. I want them to be productive. So I go to him, I say, look, we’re going to do AI because, you know, this is the future and all of this. And we can go through the 90-10, 10-90 and go back to that part of the conversation.

John:
[1:05:42] But we say, but we’re doing this, so I’m going to buy a bunch of hardware for you. What do you want? And you can go through and you can go max out your GPUs or you can go to using like an AMD approach, which is going to be less expensive.

John:
[1:05:57] There are different ways that you could go, but let’s just say now I have it.

John:
[1:06:00] So I’m going to have, let’s say, 10 million, you know, threads or CEs or whatever you want to call it in your own lingo. I’m going to have 10 million of those going. And they’re going to all be using memory because everybody knows AI is all about memory. So I’m a data scientist and I’ve got to figure out how do I segment my model so that it fits into 10 million of these suckers, whatever my segmentation is, so that they go off and I submit the job. And then about four hours later, it all completes. Why four hours well i’m making it up but it’s because i want to spend the system the big system that my big oil company has bought me i’m going to spend that as many times during the day so if i do four hours i get six of them okay and so i you know i rotate them and so i wait i’m 50 people i wait for two weeks i get my moment and then i run everything so i spend all this time segmenting my data so it fits into the hardware okay all good so my time comes up and hallelujah you know It’s all queued up. It’s all auto-magical. It comes up, and I check in, and my job’s off cranking along. And one of those 10 million threads happens to use, I don’t know, five times more memory, six times more memory than it’s supposed to use. Well, why? Because, well, I’m doing math, and I have 10 million of these going to 10,000 or whatever it is. And one of them went a little bit differently than I thought. That’s why. Is that possible? Yeah. So what happens? it goes above the local memory in the server.

John:
[1:07:30] And, of course, the server says, no problem. I’m going to kick in, and I’m going to swap to disk. Now, I’m a big oil company. I can afford anything. So I put the best hardware that money can buy, and I put the fastest NVMe drive in there. So when it swaps, it gets the biggest best of the best of the best, and I’m awesome. Okay. Well, in AI land, that swap going to NVMe compared to DRAM is 125 times slower. And you say, well, okay. 125 times, that sounds pretty slow. So what’s that mean? That four-hour job now takes 19 days to complete. Because one of my 10 million guys went above the local memory. So that’s no problem. I’m a big oil company. I’ll just make those 50 other people that I’m paying $50 million a year with, wait, 19 days. Well, my job finished.

John:
[1:08:24] No problem. Okay. Well, obviously that’s not going to happen. So the first thing I do is if ever I go outside of the local memory of the server, the first thing that I do is I crash my system because I want to get, crash it. I lose my role. Somebody else pops into my slot. I go back and I figure out what I got to do. So my moment is there, I’m not going to run out of local memory. So you know what everybody does? People, engineers are really stupid, right? And so they obviously, they start saying, well, what’s the statistically meaningful number that I can run hot on my memory in the system? Here’s the answer. It’s 25 to 30%. So everyone targets 25 to 30% of local memory, right? So that if one of those guys goes up 3x, no problem. I’ve got enough. And now my job completes in four hours. Everybody’s happy. Okay. But riddle me this.

John:
[1:09:23] I’m now targeting to use 25% to 30% of my cost basis of efficiency because it’s the only way I can do it. Everyone knows this. Okay? So let’s interKove. Okay? Let’s now imagine, because we can put memory 500 feet away or 150 meters away at the same performance or faster than local memory, you can now, instead of using 25% of memory, target 75%. So right out of the gate, my model is 3x more efficient. It runs 3x more data. Or if it took three years to train that, it now takes one year. Because I’m tripling the memory on this. Now, let me give you something else. That’s just in local memory. So now when it goes above, it goes to 3x the memory and it goes above, it dynamically gets the memory from across the data center. So I’m now running a terabyte server can be an effective 2 terabyte server across the data center, whenever it needs. So my efficiency on that doesn’t just go up 2x. It can go up 5x. So now if I’m going to take three years to train my model, it can take three months to train my model. Okay. Or what does the big oil company think about that? Right?

Ryan:
[1:10:49] It’s money.

John:
[1:10:50] It’s money. Right? And it’s better product. It’s better quality. So the point about this, to go back to like the real, you know, think about this from the real world perspective, the way that you think about software defined memory isn’t necessarily I want bigger and better. What you want to do is change the way you think about solving problems, where you think, I no longer worry about stranded memory. I no longer worry about starved process. I give the memory when it needs it. It’s all automated. I let the software build custom hardware in response to the model. It’s inverting the paradigm of programming. Right now, those data scientists are saying, what’s my hardware constraint? How many memory channels do I get? How do I optimize so that my software will fit into the hardware? What if I didn’t do that? And I said, I want to work on the data science. The hardware should shape itself however the bloody hell it needs to shape itself. Right? That is what Kove brings to the table. Okay? And it is game changing. Up and down and all around. Is that making sense? Is that pragmatic?

Ryan:
[1:12:00] No, it makes complete sense.

Jill:
[1:12:02] Memory is scalable.

Ryan:
[1:12:04] It’s like you said at the beginning. Everything else has been able to be scaled and virtualized, but memory well. And you guys have figured out how to do that. And that’s the simple way of looking at it.

John:
[1:12:17] Yep.

Michael:
[1:12:18] I mean, it’s also just like you’re talking about examples of like a terabyte or two terabytes. And you can have effectively, based on what you’re saying, like the amount of like the limit is basically the size of the data center. Like you just is whatever you can put in there.

John:
[1:12:33] We have a couple of cloud folks that we’re talking with. I mean, the challenge that we’ve had is that sometimes we talk to people, we end up in uncomfortable situations five years later with lawsuits, right? But we do have a couple of cloud folks that we’re talking to. I mean, this has the potential to change the whole dynamics of cloud, right? I mean, if I could improve the efficiency of cloud computing 5x, who wins the consumer wins everybody everybody wins right it’s actually kind.

Michael:
[1:13:06] Of funny because of the i just kind of chuckled because i thought of something so it’s first you started you you basically started the distributed hash tables to make the cloud possible and now you want to make it five times more efficient you know.

John:
[1:13:19] Yeah let me give you let me give you a really good another really good illustration of this this is something that we’re just doing right now so um we just ran some benchmarks because somebody asked us to do it for elastic search so are you all familiar with Elasticsearch.

John:
[1:13:33] So here’s the beauty of Elasticsearch. It’s DHT based, right? So it scales wide, right? So every year, let’s say I’m a big organization and let’s say I’m making stuff up. I’m deliberately because I don’t want to step in anybody’s cabbage patch of writing. So this is fictional. Okay. I mean, it may be true, but it’s fictional. So I have a thousand servers and the value of Elasticsearch is, you know, at the end of the year, let’s say I have a budget and I’m going to refresh 30% every year because my refresh cycle is once every three years. So year one, 30%. So what I do is I go in and I add 30% more servers, 30% more licenses. It’s good. And I’m not having to re, you know, forklift, refactor everything. I just add more servers. It scales across the servers and I get better benefit every year. And I can do that over and over and over. Okay, that’s one beautiful, wonderful way to do it.

John:
[1:14:28] Now let’s imagine a scenario where each one of those servers can become five times larger using pooled memory. Okay? So how much money do I save if every year I increase my server and license count 30% every year? Now let’s imagine the scenario where I, in one year, improve my densification of each server 5x. That’s 500%. So in one year, I can go five. So how many years does it take 30% growth rate to meet the 500%? You can throw software defined memory and each server becomes five X larger at the same performance level.

Michael:
[1:15:05] That’s crazy.

Jill:
[1:15:07] And you don’t have to get rid of your older servers because they can be in the pool of memory.

John:
[1:15:14] Exactly. I mean, it’s exactly that. And so where you really let this start to kind of bang around in your head is you can take any kind of commodity infrastructure and you can make it really badass just by adding software.

Ryan:
[1:15:34] Love it.

John:
[1:15:35] And, I mean, it’s not terrible that I can just apply software, third-party software, and if you don’t like it, you can uninstall it. If you like it, you can install it. I mean, it’s not intrusive. It works all with open interfaces. We don’t have any vendor lock-ins. sticky stuff, right? It’s really, I mean, it takes 30, less than, I think it’s 14 minutes to install our software. 14 minutes and you can prove it. But we have customers that have skipped two generations of hardware refresh. Think about the money.

Ryan:
[1:16:07] That’s a huge cost savings.

Michael:
[1:16:09] Massive savings.

Jill:
[1:16:10] Yeah.

Michael:
[1:16:10] I mean, just hardware, they want to keep the hardware as long as possible just because of the amount of money it costs to do that.

John:
[1:16:17] We have a customer that has memory targets in active production that are 14 years old, okay? The organization that makes the hardware quits supporting them eight years ago. They can’t get hardware support. So they’re in racks. And what happens is if something burns out, it fails. Our software identifies it. It X’s it out, kind of like what you do in a cloud thing. It just takes it out of the pool, and then it gets a new allocation. That full reKovery cycle takes less than 200 milliseconds. That’s how long it takes you literally blink your eye. That’s how long it takes you blink your eye. That’s five, I think it’s five generations old of memory technology in active production for like a… A badass organization.

Michael:
[1:17:07] That’s actually also really cool because of the fact that if if something does fail like when you’re in the old style that you you have to deal with a lot of headaches right but in this kind of structure even if it does fail you know with the literal blink of the eye it fixes it kind of makes a new pool but also it means that you can yes your pool is slightly smaller but because of the technology you’re still at like you can add another old.

John:
[1:17:32] Server it’s like what jill was saying you can just add another old server and make the.

Michael:
[1:17:36] Pool bigger yeah it’s awesome right yes so exactly exactly i’m curious about something um so i wanted to talk to you about the compute express link how does Kove sdm compare to emerging technologies like compute express link so.

John:
[1:17:52] Um i have lots to say about this um so so first of all i think there’s plenty of room for lots of people to do lots of different things in the world, right? And, you know, ExpressLink is, you know, it’s a consortium of 180 companies. Intel is behind it. They spend a lot of time. There are a lot of smart people behind it. So, you know, say lots of good things about that, that way. However, now I’m going to say the way I would, I would look at it, what I would consider to be an analytic set of comments, right? And I’m not casting aspersions. I don’t, there’s lots of ways to think about this in lots of different ways.

John:
[1:18:33] ExpressLink is hardware. So if you want to use it, you want to use CXL, all you got to do is buy new servers, new chips. You may have to have new programming models. You may have to have new networking. And it’s awesome. And when you do that, they have grand claims of the future. I don’t know what you do about the old hardware that doesn’t work with the new hardware, which I’ve invested however much money I’ve invested. And now I’m in a new hardware game. Now, the argument might go something like this. Well, yeah, but, you know, you’re already buying new hardware. So you’re just doing a migration path. And this can be a cost efficient migration path because it’s going to be awesome. And that’s good. So, our view is we could work with any CXL target that they want to make and it could just plug into our software and we’d be happy campers, okay, if they want to do that. We work with 14-year-old hardware, 13-year-old hardware, 12-year-old hardware, 11-year-old hardware.

John:
[1:19:36] Picture. And we’ll work with two-year hardware in the future. And we’ll work with 10-year hardware in the future that works with the same stuff because we’re a software layer. And the approach that we took because of the kind of background that we have, everything is written to the open interfaces in the Linux kernel at the top level kernel. We regression test on 57 kernels every night, pretty much every major distribution so that our stuff is solid. Everything is anchored to the migration path of how Linux goes forward. We are a vigorous supporter of that kind of approach because I think that’s where innovation goes. So to the degree that Linux supports anything, we support anything. To the degree that Linux doesn’t support something, that’s our limit, okay? So now here’s where we’re different with something like CXL. So from our perspective, we hide 150 meters of cable latency and we are the same speed as local memory. We’re faster than local memory. And that’s not us saying it. It’s customers saying it’s benchmarks, third-party this, third-party that. In the best case, the CXL specification is designed for 100 nanoseconds up to a microsecond. Okay?

John:
[1:20:52] The memory inside of a box is 60 nanoseconds or less. So it’s slower than local memory. You can do some things to try to make it fake it down a little bit. But we’re designed to be the same speed as local memory or faster, except we’re software. So my theory to the worldwide community on this would be we have a commitment to work on any piece of hardware that is ever supported by Linux. We don’t have any obligation to any vendor lock-in. We don’t require anybody to use any special hardware. Everything is open interfaced. Everything is designed that you shouldn’t know that we even exist. You should never have a forklift upgrade. You should never have to change your hardware. You should never have to change your software. You should never have to change anything. It should work transparently and just like memory. Oh, and by the way, it ought to be the same speed as local memory. So I’m just saying a little bit different philosophy.

Ryan:
[1:21:45] So CXL, you have to have the compatible hardware, and it’s limited by, of course, the speeds of your PCIe lanes, right, because of that hardware. And everybody knows the PCIe lane is what you want to be on. That’s your fastest bus lane to be on for your system. Couldn’t you use both though? Couldn’t I use, if I have some of my systems that I put CXL on and I’m upgrading a couple of them, I could still use Kove across.

John:
[1:22:11] Like I said, we don’t care. Go for it. You like CXL, do it. You can plug into us. That’s awesome. We don’t really care. I, like I say, I’m not trying to cast dispersions. Like I, look, I come from, you know, I think that money matters. I think that, you know, efficiency matters, cost matters, performance matters. Those things really, really matter. Right. So you want the best way to do that? Use software. Don’t buy new hardware. Right. What Jill was saying, use that hardware. That’s 15 years old.

Jill:
[1:22:40] Yeah.

John:
[1:22:40] Yeah. And make it, make it valuable. Yeah. Just saying, right.

Ryan:
[1:22:45] All of a sudden, Jill, Well, your whole computer museum isn’t a museum. We could use it to process the AI.

John:
[1:22:50] Large language.

Jill:
[1:22:52] Well, you know, in the 80s, when I was setting up render farms, I was using 486s in the farm just to increase even a little bit that processing power.

John:
[1:23:04] Yeah, yeah, no, exactly. Exactly. To go back to that, I would even say, you know, there’s, you know, I know, you know, what I anticipate is there are going to be a lot of startups and, you know, we’re an unusual, you know, animal because of the way we’ve done it. We’ve been around for 23 years. Right. So we haven’t done marketing at all. Right. If you all, you know, hadn’t really walked up, I wouldn’t have been on this. Right. And if you had been jerks, I wouldn’t be on it either. Right. So, you know, we’re.

Michael:
[1:23:31] That’s why you talk to me and not Ryan.

John:
[1:23:35] I just got to make sure that I say out loud, like these are like the really nicest guys. Making sure something by sarcasm could be taken the wrong way. I do not mean it that way in this case.

Michael:
[1:23:46] And also Ryan’s okay, I guess.

John:
[1:23:48] Yeah.

Ryan:
[1:23:49] He’s okay, I guess. Love it.

John:
[1:23:53] But I’d say that, you know, the future is memory. I mean, if you want AI to be real, memory is the thing that holds it up. So there’s going to be a lot of people doing a lot of things. And everybody, it just, you know, I kind of chuckle because, you know, we were looking at hardware accelerated approaches like CXL in 2003, 4, 5, 6, and 7. That’s just a few years ago, right? That we were looking at the same stuff and we decided I was getting ready to build cards to go into the PCIe bus for the same instinct of why people are thinking about CXL. The problem is when I look at the history of computing, it always goes to commodification and it always goes to virtualization of commodification. That’s what you get right now with Kove, right now. So if you want to use your old stuff and you want to use CXL and you want to get all upgrade everything, go for it and we’ll be your friends.

Ryan:
[1:24:45] Yeah, that’s awesome.

Michael:
[1:24:46] Nice.

John:
[1:24:47] Is that helpful?

Ryan:
[1:24:49] No, it’s super helpful in explaining the difference. Yeah.

Jill:
[1:24:53] Yeah. Well, John, there are actually new challenges in computing. What are you excited to tackle to meet the demands of future technologies like quantum computing and AI?

John:
[1:25:08] So I have a schizophrenic, not actually schizophrenic, but metaphorically schizophrenic approach. So the number one thing that I think of in my day job is to help find and educate people so they realize that the art of the unimaginable is here, right? Like you can do that. And the vector on that initially is what if there really is no more of a memory wall, period, okay? That’s part one. Because if you can get people thinking like that, you cannot have them think like – So I knew the person who was running.

John:
[1:25:47] Motorola and when Motorola failed. And I had a chat and I said, you know, how this happened? And I’m not going to use names, even though I know people can look up who it was. But, and he said, well, every quarter somebody came to me and said, you know, we own 98% of the analog market. Okay. 98% or 96%. And I can guarantee you this amount of revenue every quarter. And somebody else on the team would say, you know, digital stuff is coming. Smartphones, They’re coming. It’s going to be big. And the person will say, yeah, but I can guarantee you this money. What’s the amount of money that digital thing is going to make? Right. And that’s OK. That’s a certain kind of thinking of that. Right.

John:
[1:26:25] That’s when you lose imagination. It’s when instead of thinking I should be building for the future now, you think I need to be protecting my future and my past. Right. So going back to the initial question, I think the first thing is you want to inspire people to go back to thinking, I want to build the unimaginable and make it real and valuable, like really valuable, not as a science experiment off in a lab. But I want to build something like a render farm that goes five times faster using old hardware, okay? Or I want to build commodified AI so that people in the open source community can do kinds of training on local systems without having to be in government labs or in $30 billion a year facilities, right?

John:
[1:27:18] And the first part of this is, I would say, that really is memory. And it’s not just because we built. I mean, we’ve gotten here after 23 years of building to get here, right? So we’ve put on our dues here to get here. But I do really think that the first part of what the next future is, it’s getting people to understand you don’t have to be memory constrained at all. It’s now, not tomorrow, not two years from now, not when CXL might be CXL.3.0, which might go across a rack as opposed to inside a server and buying new hardware. But you can do that right now. So the first educational part of this is software-defined memory will change anybody’s anything. It is the equivalent, in my opinion, of when I cooked up DHTs for location service and I said, what if I wanted to search anything, anywhere, all the time? Right? That was the theory of DHTs. The theory of software defined memory is I never want to talk about memory. I just want it to work. I just want my job to be built and to let the hardware sort itself to give me what I need.

John:
[1:28:28] Okay. Now that’s first part of my schizophrenic stuff out of my side of my mouth. Second part of the schizophrenic stuff out of the other side of my mouth all of the futures when you think of quantum or other kinds of models or edge or ai or commoditized ai and we can go through a list just take a machine gun or no take a fire hose and just go right and you can come up with all kinds of ideas that stuff happens because i have virtualization of compute network storage and memory.

John:
[1:29:02] OK, that’s where that happens from. And when that and right now we still haven’t licked the it’s not in common. And I hope the people who are in your, you know, in your community listen. And after they hear this, they go back and they say, what what could I do if I really wanted to change my organization for real? Like not not as a thought experiment, going to my boss and saying, I get this cool science experiment that I want to do, you know. But like what if we could be more competitive because we could reduce our cost infrastructure by half or we could do things that we could compute models that would have taken three years in three months? What would that do to allow us to change? And, Jill, if we get to that, things like these other things aren’t specialized. Yeah, aren’t going to matter.

Jill:
[1:29:52] Yeah.

John:
[1:29:52] Of the open source community. And if anything that I think open source has shown is however awesome quantum is, and we can go through all the big ideas, the ones that win are the things that make everybody be able to do everything everywhere all the time, right? That’s what wins. And that isn’t those big, it’s the Linux operating system. It’s the most successful piece of technology ever invented, ever. I mean, maybe there’s something, maybe the car is better, right? Maybe because the wheel.

Ryan:
[1:30:22] I’d rather have Linux.

Michael:
[1:30:28] But here’s the thing. It’s not even just that. Exactly. Linux is actually powering most of these cars now. I know.

Jill:
[1:30:35] Yeah.

John:
[1:30:36] But, and so was that done in a, you know, in a, in a think tank with, you know, the other one is like the Wright brothers, right? So I’m about 200 miles from where the white Wright brothers did it. So, you know, when you go in and out of DC, you go to that, that hub where that was the person that got all the money from the Europeans to come up with a flight machine, how the Wright brothers do it. They’re bicycle mechanics and they, they could work all day to get a chance to fly, you know, a hundred feet and then have it crash and they’d have to resolder everything.

John:
[1:31:04] Over and over and over. Innovation comes from the knuckleheads that really do think like the open source community. And so I do think there are things like quantum, for example, that’s very intriguing. I do. And that’s, you know, that’s from a proper kind of research approach. But when I look at, you know, things like the telephone also invented about less than a mile from where I’m sitting right now, the cell phone was invented. And, you know, Chicago has a lot of these old, you know, technologies that were invented because you had this Midwestern approach of like, I got to make stuff work, right? I’m not looking for fame. I’m making it work. I want the plane to fly. I want the phone to work, you know, Bell Labs, right? In the Midwest, right? That was the Linux operating system, right? I mean, that style of put your feet on the ground, roll up your sleeves and go make things work. I think that’s where the big innovation comes. Linux is maybe be the best. It’s certainly one of the best. I mean, I think all of the open source stuff that Richard Stallman did is also awesome. The copyleft and so forth. There’s a lot of people that have done a lot of things. So when I start thinking big innovation, I think there are so many things coming. But right now, my view on this is the quicker that we can get the world to start thinking, I now have the virtualization of all infrastructure, is the quicker that those big ideas hit more people.

Ryan:
[1:32:28] And when.

John:
[1:32:29] They hit more people get more innovation.

Ryan:
[1:32:31] That’s where open source really shines because when we talked about this um you know the digital divide is a real problem and it hasn’t gotten smaller in a lot of ways with ai the digital divide is now widening again where you’ve got this pay gap to utilize this tool you have to have this this uh constant internet connection to utilize a lot of these tools. And so now the next John Overton, the next Jill Bryant, the next genius innovator out there may not have the resources. They may not have been born into the family that allows them to be able to pay that subscription fee, to even have internet. There’s a lot of people in the United States who have no access to fast, high-speed internet. So when you look at Linux, what was it? It was Unix was this amazing operating system, but only a certain small group of people could afford it and could use it. And then Linux came in here and said, what if I gave that same technology to everyone and everyone had access to it? And that’s where I look at this stuff and I think about the real innovations for people to do what you’re saying, to think about the unimaginable, to go where it’s uncomfortable, is also looking at how can you get it into the hands of everyone.

John:
[1:33:47] I 100% agree with that. I really would like to go, and I don’t want to get too sappy on this, but I’d like to go, but I do think it’s true.

John:
[1:34:02] The U S is a country of innovators, right? That’s what we started, right? We’re supposed to be the people that think out of the box, you know, like, you know, that’s one of the jokes that we have a Kove thinking out of the box. Like we actually do, right? Like, you know, it’s not the memory isn’t the box. It’s out of the box, right? Think out of the box, right? So, but, um, we got to get back to where we’re encouraging communities to do that. Uh, you know, where people are thinking, okay, so, you know, there, you know, there hasn’t been a man on the moon or there isn’t a car, there isn’t a, you know, an AI LLM, or there isn’t just, we need to go back to that. And the way that happens is through very vibrant intellectual communities, where it is driven by necessity and lack, not surplus. The surplus isn’t what gets you there, right? It’s that you feel like, I, you know, I got to make everything that I do count. You know, I, you know, I can’t afford to build an airplane over and over because the Europeans are giving me money. I’m a bicycle mechanic. I better make it work. I mean, that sense of innovation is very American. And we definitely try inside of Kove to inspire that kind of commitment that your job here isn’t to come and collect a paycheck. It’s to do something that’s original. And I think that really is America. And if we can do that effectively, that’s where the innovations keep coming.

John:
[1:35:32] And I think sometimes we lose sight of that in the modern era. Oh, yeah.

Ryan:
[1:35:37] Absolutely. I think for sure. Now, John, this interview, as we thought, was amazing. It’s incredible. And you have passed the first interview, but this was the easy part.

John:
[1:35:48] Oh, okay.

Ryan:
[1:35:49] This was like the warm-up test, okay? Because what we have next is where the real interview – this is what makes or breaks our guess.

John:
[1:35:58] Oh, I see. Okay.

Ryan:
[1:35:59] This decides whether people like you or whether – This.

John:
[1:36:03] Is the gauntlet right here.

Ryan:
[1:36:04] Yeah, get some water because we call this the lightning round. And what we’re going to do is we’re going to ask you a question. We want you to answer the first thing that comes to your mind. And you will be judged on these answers.

John:
[1:36:17] Just know that.

Michael:
[1:36:18] And these are very intense questions.

John:
[1:36:20] I get the kind of two guys and gal that you are. I get it. I’m ready for it.

Ryan:
[1:36:25] All right. Here we go.

John:
[1:36:28] Do you have a guideline of how long my answer can be? Can you notice it?

Ryan:
[1:36:32] If I can get set off. We really don’t. Just as quick as you want, you can answer it.

John:
[1:36:36] See, I can already game this out. If I talk for 10 hours.

Michael:
[1:36:41] He never stops talking.

John:
[1:36:42] I never get the second one, right?

Ryan:
[1:36:45] I love it.

John:
[1:36:45] I love it. Okay, so I’m in the vibe.

Ryan:
[1:36:49] All right, here’s your first impossible math equation. No, I’m kidding. Here it is. Favorite band or musician or genre of music?

John:
[1:36:57] Oh. Country.

Ryan:
[1:37:03] Country okay nice.

Michael:
[1:37:05] Your your favorite guilty pleasure like candy or soda or something like that.

John:
[1:37:09] Um um kombucha kombucha.

Jill:
[1:37:19] Nice what’s your favorite thing to do that’s not technology related.

John:
[1:37:27] Um playing squash or going bicycling or or better than that hanging out with my wife oh.

Ryan:
[1:37:34] Love it good answer too by the way.

John:
[1:37:36] Got you out of the doghouse there uh.

Ryan:
[1:37:38] Chicago style pizza or new york style pizza.

John:
[1:37:42] Yeah flat i i have had i’ve been i’ve been i’ve had too much of chicago style i used to go here and i overdosed on it and now i’ve.

Ryan:
[1:37:53] Actually done the same with chicago style pizza uh overdosed on it.

John:
[1:37:57] It’s very good though like it’s so good that eventually it’s like i can’t do it anymore it’s.

Michael:
[1:38:02] So rich and thick and yeah it’s delicious what’s your favorite movie.

John:
[1:38:06] Oh my dinner with andre oh.

Michael:
[1:38:14] That’s a good one that’s a good one.

John:
[1:38:17] Okay you prefer okay.

Jill:
[1:38:20] Cupcakes or muffins.

John:
[1:38:22] Oh muffins for sure.

Michael:
[1:38:24] I got him right man.

Ryan:
[1:38:26] You know you just you can’t nobody’s perfect michael that’s what it proves.

Michael:
[1:38:31] Nobody’s perfect you know well okay so for for reference john if you don’t know so we have this weird debate that he likes cupcakes i like muffins and we ask everybody who comes on the show just just to do it i mean john let.

Ryan:
[1:38:45] Me give you a perspective a muffin is just an ugly cupcake.

Michael:
[1:38:55] But a cupcake is actually a cake that just didn’t have a lot of ambition oh.

Ryan:
[1:39:00] My god oh my.

Michael:
[1:39:01] God it.

John:
[1:39:02] Goes deep i am humbled at the uh at the the like the.

Ryan:
[1:39:07] Thought that goes into.

John:
[1:39:08] This depth of thought on this.

Michael:
[1:39:10] I’m humbled but also ryan i can’t help but notice earlier you were talking about geniuses and stuff, and you mentioned John, and you mentioned Jill, and you didn’t mention me.

Jill:
[1:39:21] There’s a reason for that.

Ryan:
[1:39:23] I’ll let you reflect on that after the show.

Michael:
[1:39:25] I see. All right. Well, the last lightning round question we have for you is what is your next unimaginable thing that you’re taking on? Is it time travel or teleportation?

John:
[1:39:39] No. I’m saying teleportation for sure. if that’s the choice that’d be the one.

Ryan:
[1:39:46] We need it i mean star trek has inspired so much of our technology but we didn’t get teleportation that’s the only thing.

John:
[1:39:53] You’re missing really i’m bummed about that right like.

Ryan:
[1:39:56] Yeah i.

John:
[1:39:56] Want to be able to be up somewhere and beam me down or beam me up right.

Ryan:
[1:39:59] Oh yeah exactly so great well john you made it through the interview you passed the gauntlet in the lightning round thanks you did it i.

John:
[1:40:09] Really appreciate it.

Ryan:
[1:40:11] Yeah and all we have left to do is thank you for really thank you for taking the time to talk to us.

John:
[1:40:18] What it’s like, Jill, it is so delightful to meet you. Okay. You know, again, I mean, I did get to meet you, but like for the first time and to you guys, I like, I knew this was going to be a hoot. I mean, this is the first time.

Ryan:
[1:40:32] We’re so happy.

John:
[1:40:35] Creativity is great. Thanks.

Ryan:
[1:40:36] We’re so happy to introduce you to our audience too. Yeah.

Jill:
[1:40:39] We are honored to have you on our show.

Ryan:
[1:40:42] Thank you. And likewise, we are going to have you back on the show because I know you’re going to have more exciting adventures. And we honestly, we had a list of questions and we didn’t even get through it. So we have to have you back on because we’ve got more questions. This is the problem with talking with John. Like we ran into it, the summit, John was supposed to leave, by the way, he was out the door and we called him back. And I think he was like, okay, you know, we’ll talk for 10 minutes, like an hour and a half later, John’s still there sitting there talking with us. So this is one of the issues that we have is I think all of us have the gift the gap, but we’re going to have to have you back on. We’re going to be really excited to follow Kove and see what you guys are doing. So again, thank you. And we’ll have you on real soon.

John:
[1:41:23] Thanks so much.

Ryan:
[1:41:23] I told you that interview was going to be incredible. Did not disappoint. Huge thank you to each and every one of you for supporting us by watching or listening to Destination Linux. However you do it, we love your faces. It’s because of you that we can bring incredible guests like this onto the show without you listening, without you subscribing, without you following, without you joining our communities, We wouldn’t be able to pull this incredible talent to come talk about their origin story and the innovations and all this incredible stuff. Go to tuxdigital.com/discord, where you can discuss the show. You can discuss your love for Linux. You automatically make friends. All you have to do is walk in there and go, hey, I like Linux. So you type in there, hey, I like Linux. And boom, you have thousands of people who are now your friends who you have something in common with. I mean, that is amazing right there.

Michael:
[1:42:13] Exactly. It’s amazing. And also, so without you helping us, without you being followers and subscribers and that, we wouldn’t be able to get these kind of interviews like Ryan said. But also, this show wouldn’t even be possible without all of you, and especially those that we have as patrons. Because tuxdigital.com/membership is where you can become a patron. And just to support the show, it makes it possible for us to create this show. And we appreciate you so much. And because of that, we also want to give you a bunch of cool perks. And that’s why you can watch the show live or watch unintended episodes of the show or check out the merch discounts. So if you want to get some of the cool merch we have like the Linux is Everywhere t-shirt that I’m wearing, the Destination Linux shirt that Jill’s wearing and absolutely nothing that Ryan holds up or wears ever, you can go to tuxdigital.com/store and get all that cool stuff with discounts if you become a patron. So there’s so much there, hats, mugs, hoodies and more at tuxdigital.com/store.

Jill:
[1:43:11] And make sure to check out all the amazing shows here on TuxDigital. That’s right. We have an entire network of shows to fill your whole week with geeky goodness, like Michael Tunnell’s This Week in Linux.

Michael:
[1:43:23] Oh, yeah.

Jill:
[1:43:24] Head to TuxDigital.com to keep those Linux penguins marching.

Ryan:
[1:43:29] Everybody, have a great week. And remember that the journey itself is just as important as the destination. Thanks, everyone.

Michael:
[1:43:38] Thanks, everybody. We’ll see you next week.

Ryan:
[1:43:40] You wouldn’t download a car, would you? No, but I’d download memory. I’d download memory.

Michael:
[1:43:45] I mean, yes, I would download a car. If I could download a car, I would download a car.

Ryan:
[1:43:48] I’d download memory and a car.

Michael:
[1:43:50] Both of them. Both of them, yes.

Ryan:
[1:43:52] Now I could do the memory part. The car. Well, I guess Kove. Kove’s probably going to work on it. They’ll probably figure out a way to download a car.

Michael:
[1:43:58] At this point, with the 3D printing and the speed at which you could do, I mean, maybe. Yeah.

Ryan:
[1:44:03] I’m going to get Kove on that. They’ve got this.

Jill:
[1:44:05] Yeah.

Ryan:
[1:44:05] Now, every impossible task we come across on this show, we’re going to be like.

Michael:
[1:44:09] Hey, Kove will do it.

Ryan:
[1:44:10] Kove’s going to get this done. We’re going to send this over to Kove. We’ll have it in a few weeks. We’ll be ready to rock and roll. That’s the way to go. Love it.

Leave a Comment

Start the discussion at forum.tuxdigital.com