The Hacking Open Source Business Podcast

HOSB - Building a Bootstrapped Open Source Company: Insights from Jetstack Founder & President, Matt Barker

March 07, 2023 Matt Yonkovit & Avi Press Season 1 Episode 18
The Hacking Open Source Business Podcast
HOSB - Building a Bootstrapped Open Source Company: Insights from Jetstack Founder & President, Matt Barker
Show Notes Transcript Chapter Markers

On this episode of the Hacking #opensource Business podcast, we talk with Matt Barker, the founder, and president of Jetstack. The discussion is about his experience in building a bootstrapped open source company based on his previous work at Canonical and MongoDB. In the interview, Matt talks about sales, business, and open-source technology (Linux, Kubernetes, Mesos, and more!). Matt covers the challenges of selling free open source software, support as a product, retention challenges, and measuring growth and adoption in open source software. Matt also shares his experience in comparing MongoDB and Canonical and starting Jetstack by choosing Kubernetes as the next big thing. The conversation delves into bootstrapping Jetstack with a services model, moving from services to a software product, comparing Kubernetes to Apache Mesos, and betting on truly open software. The interview provides valuable insights into sales, business, and open-source technology, making it an informative resource for anyone looking to learn about bootstrapping an open-source company.

Two Matts and an Avi walk into a conference in London and end up in the Matt Cave... and this is the result.  

00:00:00  Welcome to the Matt Cave
00:01:48  Getting to know Matt Barker and Jetstack
00:02:30  The Early Days of Canonical
00:07:17  Support as a product, retention challenges
00:13:21  Trying to go to the proprietary world from open source
00:14:28  Comparing MongoDB & Canonical
00:17:27  Bootstrapping Jetstack with Services
00:19:50  Moving from services to product
00:22:53  Convinced Kubernetes was the right space
00:24:54  Betting on truly open software instead of open source controlled by 1 company
00:29:43  Measuring growth and adoption
00:33:04  Rapid Fire

Checkout our other interviews, clips, and videos:

Don’t forget to visit the open-source business community at:

Visit our primary sponsor, Scarf, for tools to help analyze your #opensource growth and adoption:

Subscribe to the podcast on your favorite app:

 Episode 18a - Matt Barker

[00:00:00] Welcome to the Matt Cave: at the State of Open Con

Matt Yonkovit: Hello everyone. Welcome to another episode of the Hacking Open Source Business podcast. Um, I'm your host, Matt Yonkovit. Uh, once again joined by AVI Press. And today we're joined by Matt Massey. Matts Matt 

Matt Barker: Barker. Matt, 

Matt Yonkovit: that's it. The, the two Matts. And you know, we're in a Matt Cave. . We're in the Matt Cave right now.

Gosh, I don't know if you knew that. Yes, we are live from the Matt Cave at State of Open Con. . 

Avi Press: What? What the heck? Like 

Matt Barker: never actually heard that one before. Before Isn't great. Matt Cave at home. Yes. 

Avi Press: Yeah. Yes. 

Matt Yonkovit: Yes. See? Yeah. The mats have Matt Cas. 

Yeah. Are you jealous, 


Avi Press: No. I learned something today. Mm-hmm.

Matt Barker: you know, do you ever get the joke when someone says, will you like a. Yes. 

Matt Yonkovit: Yeah. Yeah. Or doormat. A doormat. A doormat. Yeah. I do get that. Yeah. . So, um, Matt, um, just tell us a little bit about yourself. 

Introduce yourself real quick. 

[00:00:48] Getting to know Matt Barker and Jetstack

Matt Barker: So, I'm Matt Barker. I am co-founder and our president of Jett Stack, which is f.

Effectively a Kubernetes company started off the back of the open sourcing of Kubernetes. My background, my entire career history has been in open source software vendors. Okay. A very early employee at Canonical, the guys behind a Buntu mm-hmm. , uh, to the point where I was interviewed by Mark Shuttleworth and started working with him when he was working out of his front, front room.

Mm-hmm. And, uh, then I was an early employed MongoDB, uh, and that's before, before starting, starting Jett Stack. So that. Eight years ago now. Uh, we were bootstrapped, self-funded, and then were acquired a few years ago. Okay. 

[00:01:30] The Early Days of Canonical: Selling Free

Avi Press: A very interesting like, series Yes. Of, uh, progressions in your career. I'm really curious.

So by the time, by the time you started Jett Stack, can you maybe like compare and contrast con the way that you learned about commercializing open source? Yes. As the way canonical saw it versus the way that. 

Matt Barker: That was a large part of what, what gave me the confidence to start a company around open sources that I'd seen a few different approaches to the commercial side of opensource and they were both wildly different cuz I think when Canonical started and Tu was created the it was at, it was at a time when if you could get eyeballs and could get users, you'd automatically assumed you would be able to monetize it and.

That wasn't as easy as I might have thought it would've been. And for a while, I would say probably our greatest source of revenue was a Buntu T-shirts. 

Avi Press: Whoa, . Whoa. 

Matt Yonkovit: That doesn't surprise me. . I mean it, as weird as it sound, 

Matt Barker: it doesn't surprise me, but honestly, that was the first. The first year or so because we, we just, mark had his grand vision and obviously Red Hat had got some traction, but Red Hat even at the time was a tiny company relatively to, to, to the other, um, sort of IT companies or BMS that were out there.

And I think it was all around. I think Mark had great visions of maybe an, an alternative to, to like the Apple ecosystem, but Open and Bri, and, and obviously you, you, you must be aware of the name, a buntu meaning humanity to others in humanity to all mm-hmm. . So I had this great mission, uh, led and on day one I was introduced to concept of, uh, one of our goals as being, uh, squashing bug number one, which was, um, beating m.

So it was, it was all of like, it was very much like a, it was a, it was a, it was a movement. It was a passion. It was, uh, quite evangelical, quite quite evangelical. Mm-hmm. . And a lot of the people that joined, uh, canonical early days were huge open source advocates. Mm-hmm. like, like they were, like, it was, it was written into their very d n a and and soul.

And so, you know, we had people join the company who'd been likely us 12. Roommate, you know, we had a lot of the, the early sort of contributors to the, to the, you know, working with Storeman around licensing and then, uh, early contributors to Linux, Kernel. So we had, like, that was kind of the world that I was, I was kind of, uh, introduced to.

Here's me as a, as a grad. As a grad, uh, who'd just come off three years, uh, doing sales at university, thinking how am I gonna make money from, from, uh, from this, this, and, uh, not really working. For a long time, how to kind of do that. Mm-hmm. , uh, when, when, like I said, a lot of the only thing that was kind of selling relatively easy was the, uh, was the merchandise

Avi Press: Wow. Yeah. 

Matt Yonkovit: That's really crazy. Well, I think a lot of open source projects end up through that phase though, right? Where it's like they can't sell a commercial thing, but they end up getting either donations or. You know, sell the stickers, but very 

Avi Press: few of them get to the kind of scale, um, 

Matt Yonkovit: canonical. Yeah, yeah.

That's very true. 

Avi Press: That's very true. The first Linux show I ever used Daily Yes. Was, uh, what was it going to? Uh, yeah. Which, uh, , 

Matt Barker: there was a breakthrough moment at Canonical, and that was when I, I was trying to sell services around the operating system itself, and I realized actually the value to. Was not in the operating system, it was in the services around it.

Mm-hmm. . So how it make it easier to manage inside an organization. So, so, I mean, I'm not even joking. There were, there were people that I were calling to try and sell support to who were end users. I mean, I might have even called you and tried to sell and a Ubuntu support contract in the early days. But when, when, when Avanti started to take off, when he started to see it permeate through, The, um, companies, all of a sudden you'd have, you know, you'd go from having one person using it on a desk desktop inside that company who was maybe a Sys admin to then an Ubuntu server to then all of a sudden you might have, uh, 20 Ubuntu servers running at that time in an early version of aws.

You know, this is, this is at the time when AWS started to, to come out, and then you'd have it in an on-premise environment, uh, in a, in, in your data center. And all of a sudden they'd be like, how do I do configuration management? This and it starts to become very difficult. Cause before they were treating it like, Street, treat their servers like my pets.

And uh, it's like you have to, to go feed it and water it and update it and manage it and do configuration management. When you've got 10, 20, 30 servers, you have to do configuration management. And we came across this pro, we, we create this pro product called landscape and uh, and that was configuration management for two on a.

You know, on a larger scale. And that really, I, that was the first time I had a product that I found relatively easy to sell the value of. Hmm. Um, before that, just selling support. I mean, you can make, you can make a certain amount of money from it. Yeah. But it wasn't gonna go very far unless you had something that was gonna add real value on top of.

Um, on top of the, um, the operating system itself. 

[00:06:17] Support as a product, retention challenges

Avi Press: So once you've had something that was a little bit more, you know, concrete in terms of the product that was being sold mm-hmm. , did you find that that affected, uh, support and kind of its relationship to that? Like were was, was support being kind of like a tack on to that product?

Did that, did that, yeah. How did that affect kind of view of your sales, the revenue, 

Matt Barker: that kind of thing? So I, I had a year or two without having any. Any add-on products to sell and I, I only had the ability to sell, uh, support and it was at a time when legal departments inside those companies were so concerned about open source licenses that they wanted to pay for indemnity.

Mm-hmm. . Mm-hmm. on, on the on, on the, on the distro. So I could sell a contract that was backed by a canonical that had indemnification, uh, built within it. So those were the two pieces of value was support and indemnification. Remember, this is, this is coming from a backdrop of a time when you had all of the patent trolls, um, threatening legal, like Oh yeah.

Uh, action. And then you had Steve bk, uh, referring to Linux as a cancer mm-hmm. and how it was gonna break 47 or whatever it was of the Microsoft patent. So you had a lot of people very concerned about the impact of running the software inside their organization. So actually I did sell those contracts almost as like an insurance.

But what you tended to find is that once a customer realized that, you know, actually this is, this is fine. You know, nothing's like going wrong, we're not getting sued, and uh, you know, now we're starting to ramp up on our, our training around limits. People know how to use it and manage it and deploy it.

That the value of the support contract we're starting to re reduce. So actually I found it very difficult to renew those contracts maybe after a year or two. 

Matt Yonkovit: Yeah, there's a big renewal problem. Yeah. With a lot of open source because you're competing against free. Exactly. 

Avi Press: Yeah. And you have to, 

Matt Yonkovit: I mean, as bad as it sounds, if you're not having problems, yeah.

The value continues to diminish. So the better software you create, the more unlikely it is. Someone will continue a support contract because they'll look and they'll go, I only used it once in. Two years, why am I gonna pay X amount of 

Avi Press: dollars? But I would think that like at the very big cu like for the very big, you know, enterprise customers, like you just need those guarantees regardless of if you're making use of them though.

Right. You would be 

Matt Barker: surprised . Yeah. Yeah. No, you would be, I think you would be surprised. Yeah. I mean, I, I, that's what I assumed or hoped, but it didn't always work out that way. Mm-hmm. , when you're paying a couple hundred thousand dollars a year for a support contract, you're not getting sued and maybe you open one support ticket and you know, you, but you basically, the, the cost of that support ticket is $200,000.

It starts to look a bit strange to someone who might wanna spend that budget on hiring a new person to do a lot more with the, with, with, with, um, with the budget that they were given. That's what, that's, I certainly, that's what I found. It became a little bit easier when we started to sell to more of the banks because they're highly regulated and at that time, um, I think mostly it's the case that if you're running any piece of software regulation says that you have to have it supported.

So in that case, it's, it's slightly easier. , but that's, that's not always the case. And even in banks, there's still plenty of up source software that's, that's, um, deployed and run without it being technically being, being supported. 

Matt Yonkovit: Yeah. We used to get, like, I ran, um, uh, all of the post sales, like, you know, customer success, support, manage service, you know, so I, I ran that.

Yeah. Um, and we used to get, you know, companies that would come along and they would say like, okay, we opened up two tickets. We need you to cut. 50% our bill because we didn't use it as much, and if not, then we're gonna, you know Yeah. Bounce or, um, because it is open source, there's lots of alternatives.

Right. You know, the threaten to go to, you know, yeah. Centos, or, you know, go to Debian indirectly without, you know. 

Avi Press: Yeah. Right. It's so interesting cuz like, I've talked to people like at Oracle and they kind of look at the support contract's, like, oh yeah, this is basically pure profit for us, it's just like a 15% gross margin on everything.

Matt Yonkovit: But that's because they bake in. The costs. Like you have to have a 20%, um, support fee and you can't get any updates without support. 

Avi Press: I just want, yeah, not being able to, oh, okay. 

Matt Barker: So that was the difference as well. We didn't, we didn't charge for updates. Yeah. And I think I, I might, again, I'm not entirely sure how the Red Hat all works, but I'm pretty sure that, that it was slightly different where you're effectively.

You're paying, you're paying the, the, the totally open source at the time version was Fed Door. Fedora. Fedora. Yeah. And then Red Hat Enterprise Linux was effectively open source. But you'd be paying for the, for the, for the maintenance or the, the update and support 

Avi Press: fixes. 

Matt Yonkovit: Yeah. And support too. Like you get support.

Yeah. But yeah, because it's the Red Hat Network. Yes. So what it would do is you're, mm-hmm. , when you would register for the repositories in the Red Hat Network, it would then, like, you know, , take your username password and as soon as your support contract expired, that username password no longer works. You could get the repo Poof.

Yeah. . 

Matt Barker: Yeah. And so that's how they did it. Got it. Yes. Yeah. So, so that was, yeah. So that was a, that was a bit, that was a bit different. Um, so I was desperate basically for something to sell on top. And landscape for me did that. Uh, and it really helped. And, but I still had those annoying customers. It was like, oh, this is open source.

You should be paying us for these, for the updates, for the pictures, or can I just pay for a one-off ticket for support? And I was like, you know, I'm not gonna make my number on, you know, one ticket and no, I, I can't pay you for, but that's, but it was just a very different, it was a very different attitude in those days.

It was completely different. , like these days, open source is like normal and you know, you wouldn't start a business without using open source, but genuinely in those days you were fighting a huge Oh yeah. In every single customer I spoke to. Yeah. And it 

Matt Yonkovit: involved a lot of legal Yeah, it did. Yeah. But how did you go from that though, to Monga?

Matt Barker: Well, yeah, so I, I, um, I assumed, uh, so, so canonical was great. I had brilliant time at Canonical and, um, I had a great time in engaging customers and selling to enterprises. Uh, among, uh, at, uh, economical, the thing that I didn't really understand or buy into or I, I wanted more focus on was how to, to sell software to companies and enterprises, whereas actually with the, with the, um, the, the Ubuntu phone and then Ubuntu to ones for storage and then, you know, like, you know, the, the, the talk of an Ubuntu tv and so.

For me, it just wasn't something that I was that interested in. I was more interested in the, in the business to business side. So I was looking for, I actually, well, 

[00:12:21] Trying to go to the propietary world from open source

Matt Barker: I actually had a, I had a, there, there was a little bit of a gap between, uh, canonical and MongoDB because I, I, I was like, well, what, I wonder what's like in the proprietary world?

So I, I actually picked the most successful proprietary company in the uk and I was like, if I go there and just learn what proprietary is as just a test almost, and I can learn. What they do. Well, maybe I'll learn something more. Cuz I was in intended to stay in the open source com, uh, like world, but I was like, let's just see what a proprietary company is.

I ended up choosing Autonomy, which was, uh, the company that just got acquired by hp. Um, and so I was there. I went there for about, uh, a couple months, two or three months, I realized how awful it was and then left and, and went straight to Mongo. Uh, Mongo be after that autonomy, by the way, is the one that just got, um, Uh, HP ended up down writing like 8 billion worth of the 11 billion of their payment.

Wow. And the CFO ended up being extradited and is now in prison in the US as far as, as far as I'm aware. Oh, whoa. Yeah. But, uh, so there's a, there's a big story with, with autonomy, but I was like, that was, that was so, so strange. And I, I got to learn a lot about how. Priority sales model and, and sort of like the way they thought about software was slightly different.

[00:13:28] Comparing MongoDB & Canonical

Matt Barker: And I was like, well, I could go back to Canonical cause I loved it there so much. But there were, there were a few, uh, open source companies that were hiring in the uk. One of them was MongoDB, uh, other than Canonical. And I thought, well, it's database. I'd love to learn about database. And they had a slightly different approach.

It felt a lot more commercially. Canonical, which was a lot more, uh, open source, uh, focused. And I thought, well, maybe this is the, maybe this is the midpoint, mid midpoint between like an extremely proprietary, uh, focused company and one that's very open source focused. And longer B would be where I would learn the, the middle path.

Then genuinely, that's where I felt like I did. Okay. 

Matt Yonkovit: Yeah. Yeah. I mean, and I, I think that it is, uh, you know, just a different philosophy because, you know, Mongo was really setting out to commercialize first, right. And the CEO came out and said, you know, this was part of our strategy or go to market strategy, where it sounds like from a canonical it was more like, you know, we're gonna believe in this.

And it's more of that, you know, um, evangelist type 

Matt Barker: thing. Yeah, yeah, absolutely. And I assumed that everyone had the same idea about open source, that. Mark and canonical and Ubuntu did in the fact that it was, it was, um, it was like a precious and to be sort of like really like, thought of as something that was, um, you know, like it was, it was about freedom and, and sort of engagement and, and community.

And, uh, very early do, early on, uh, Mongo db I was in restaurants or bars with the, the founders and I was talking to 'em about open source and it literally shocked me a little bit to hear that they saw it as a, just a really good distribution mechanism and. That was, uh, that was quite shocking for me back in 2012 or 2013 or whenever it was.

But I, it also like, made me think, well, who am I to say that's wrong? Um, I'm a, I'm at least open-minded to sort of seeing whether, whether it works or not. And bear in mind mon to be at this point still, still was like, had a an AGPL. License, I believe. And, um, it had a quite an active community around it.

So like from my standpoint, it was, it's just a more commercial version of a, of a, of a, of a, of a, of an open source, uh, community, uh, and project. And so I, I brought in entirely to it, but I, I realized quite early on that it was, um, Yeah, I was in, I wasn't, you know, I was in a very different type of company because the types of people that they were hiring were much more commercial in, in fo in, in, in focus.

So they were hiring really experienced sales salespeople, um, really experienced, um, people who'd scaled and, and built very successfully, um, valuable companies. Whereas, uh, canonical was much more about hiring the, the most innovative. Most forward thinking, open source people who are doing really interesting stuff with, and I'm not saying that Mongo didn't have those people.

They did, but they were much more focused on just the database and how, how is that gonna help us reach this point commercially. Yeah. 

Avi Press: So I guess, uh, moving towards Jets stack then, um, Which I think, if I understand correctly, is much more towards the canonical side than it is towards 

Matt Barker: the Mongo side. Is that, is that fair?

Yeah. Okay. Yeah, I think 

[00:16:27] Bootstrapping Jetstack with Services

Avi Press: it is. Yeah. I'm curious, so what were the aspects of the, the Mongo side that you think you, you like were the most valuable to you as you were kind of both getting started and kind of growing the 

Matt Barker: business? Yeah, I think having seen. Different approaches to monetizing open source, uh, and also deciding that I didn't want to take outside investment because I'd had friends who'd been badly burnt by that.

I thought that I would start with the tried and tested model of selling services around open source. Okay. To begin with. And that would enable me to hopefully make enough money to hire somebody and then hopefully enough money to be able to then invest profits into experimenting with open source projects ourselves.

And then at some point that there would be, uh, enough traction in the open source projects that we created that I might then get the chance to create a product strategy or a paid for a proprie. So like a, either a landscape from Canonical World or a Mongo B enterprise feature, you know, around automation or backup, whatever it is around the, the project.

I did have a plan that it was gonna start with services. Hopefully be successful enough with that to create some opensource projects and then hopefully at some point then be able to experiment with, uh, with products. But it wasn't, I I, you know, I didn't, mark had, you know, he, he, he was a, he, he'd sold, uh, his previous company thought it was 750 million and, um, MongoDB uh, with, um, you know, Maxus and who was, who was the CEO at the time, they'd just got investment of, you know, a couple hundred million.

So they clearly had, you know, the money to be able to do something slightly different. Whereas I knew I was gonna have to go through this long journey of kind of testing, but I, I'd had the experience of selling all around all these different models to know that if I took, took this one small. With, with, um, with the company.

This way, I, I might be able to get to the next step, to the next step, to next. So I was always hoping to sort of test out the various diff different models, but, uh, I was, it was just gonna take a bit longer, but I saw Kubernetes as a perfect ve vehicle for that because a little bit like landscape was the value on top of Ubuntu early days.

I saw Kubernetes as being the value on top of the docker and container ecosystem. So Docker was gonna create the market, I thought, but the value was gonna be created for companies with the, with the automation and the wrapper around it. And I likened Kubernetes in my head to landscape, um, at a buntu. And then, uh, with some of the MongoDB enterprise features around the, the open source database.

Hmm. Like 

Matt Yonkovit: fascinating. Yeah. Yeah. 

[00:18:50] Moving from services to product

Matt Yonkovit: And, um, so with, with, um, Jetstack now, so is it, you know, where, where are you on kind of that, that, that curve? When you said you started with services and, you know, you kind of started to evolve. 


Matt Barker: Tell us a little bit about that. So we grew, uh, out of services, the, we, we ended up, um, I was how, how am we gonna, how I'm gonna find customers because at the time, Kubernetes by had only just been open sourced.

And if you were using containers, you were actually likely to be using Mesos or Mesosphere. Mm-hmm. . Right. The air orchestration. Uh, and then obviously Docker was the hottest company in town. So Kubecon was much smaller than Docker Con. So I, so the, the early innovators in the space where it DockerCon rather than at, at Kubecon by the cncf, so Swarm, which they created to be an orchestrator for Docker.

Mm-hmm. was actually. Probably one of the, probably had had most of the mind share at least. So in some ways Jets Stack was, one of the, was was, was sort of, um, was sort of, We weren't really, um, a part of the mainstream thinking in those early days by focusing on Kubernetes. So, so we were a bit unique in, in that regard.

And, um, and, um, I spent a lot of time kind of researching who might be using Kubernetes. And there was, for about a period of about a year, I actually personally tracked every mention of Kubernetes on Twitter and every coup mention of Kubernetes in every job. Worldwide. And then I would personally then contact that person and try to find out what they were doing with the software.

And, um, I was trying to, trying to find, figure out sort of, you know, what, what, why, why, why using Kubernetes, like what's the, what, what do you see is the value in it? And I just built this great like little network, but I couldn't find anyone to actually pay for any services because no one was actually deploying at that point.

The only people that I could find to actually pay us were independent software vendors. That felt like it had enough potential to become something that would become very valuable and we, they would, they would effectively pay us to create connectors for Kubernetes, for, for their existing success. Yeah.

Matt Yonkovit: Kind of like that partnership 

Matt Barker: thing. Exactly. Yeah. So that's where, that's where we started. And then the other thing that we did was I went to the, to the source and, and build a relationship with Google, because I would go to the Google like office. That just Google Cloud are just kind of, was just kind of starting up to try and compete against Amazon.

Part of them open sourcing. Kubernetes was obviously to try and compete a little bit against, um, the incumbent, which was a w s, and we would go in and actually educate the Google team on what Kubernetes was. And, um, that was fantastic for us because they were like, what is Kubernetes? I was like, well, you just opensource it, you tell me.

But they genuinely didn't, they didn't know because it was so, it was so new. And, uh, there was a number of times that Googler said, don't start a company around Kubernetes. Do it around our data services. Said, you're much more, much, much lower. They all say that. Yeah. Yeah. And I was, I was, I was quite scared.

And then other people said, don't start a company around Kubernetes, because serverless will just end up taking over everything. So, you know, what's the point of focusing on a container orchestra or contact a cluster manager? 

[00:21:53] Convinced Kubernetes was the right space

Avi Press: So like, what gave you the conviction, uh, to push through all of 

Matt Barker: that? Well, there was, there was a couple of things.

The first thing was that, um, I'd managed to get in contact with and meet the. So Jo, Joe Vader, uh, he was a, and I, I thought he was brilliant and he, just, by speaking to him and understanding what his vision is and where he was going, that gave me a lot of confidence. The fact that it'd been in Google and been running a version of it with Borg for 10 years, and they were starting 2 billion containers.

A week at the time and I was like, well, clearly, you know, this technology can do something pretty. And the fact that it was apparently running stateful services as well as stateless. And I was like, if you can run stateful services inside in Borg, in in Google, then we're gonna, we're gonna be getting some, this is gonna be pretty valuable.

Cause I looked at all the valuations of the. The database companies. I was like, well, if you could put one of these databases inside a container inside Kubernetes, then pretty sure Kubernetes is gonna be quite valued. So that gave me the conviction. But the, the thing that really triggered it for me was when we created a little open source project and the founder, one of the other founders, Brendan Burns, set a tweet saying, this is the best example of a, of an AON I could ever have imagined.

For Kubernetes and was like, well, we're obviously doing something right. And then you had people like Kelsey Hightower who came in and just started talking about Kubernetes and saying how it's a future. And before you know it, you knew it. All of a sudden it just, it just all clicked. And, and the, so I, I felt it was the, it was the individuals that I trusted in, in the community as well as, you know, the, the, the, I guess the inside knowledge I had from, from speaking to Joe and understanding how this technology was used at Google just made me realize that it was a, it was a very good, a very good bet.


Matt Yonkovit: Yes. 

Avi Press: I'm sorry, I've got so many follow up questions. Well, 

Matt Yonkovit: so here's the thing. Matt has a panel to go to and he needs to leave in like about 10 minutes. Okay. So, yeah. So we do have a time limit here. We do, we do. So, you know, we might wanna schedule a follow up chat with Matt if Couple do. Yeah. You know, we would love to chat more with you.

Matt Barker: Um, you know, let's, let's do another 10 50 minutes. Yeah, I'm, that's good. Good. Great. 

Matt Yonkovit: So 

Avi Press: now you follow up question. 

[00:23:54] Betting on truely open software instead of open source controled by 1 company

Avi Press: Yeah. Okay. So, I. I am trying to imagine myself in your position. I, I would be, I What made you want to stick to specific, like, you know, if, if Mesos was kind of the popular thing at that time?

Yeah. I'm. Imagining myself in your situation, I would've been compelled to even try to hedge bets and like try to build for both potentially. Or seeing how you can kind of, you know, be in multiple ecosystems at a time. And I think there are a lot of other open source founders that also kind of have to hitch their wagon to one ecosystem or another.

How are you thinking about kind of going all in on one versus diversifying? 

Matt Barker: If you'd have been my advisor at the time, I might have ended up being . Maybe I should've. Well, 

Avi Press: you would've been 

Matt Barker: wrong though, right? If you were listening to this. Yeah. But, but genuinely I did. I did actually look at, at Mesos and, uh, there were a couple of customers at MongoDB who'd, um, who'd experimented with it, and they re really liked it.

But then when I started scratching the surface, I, I, I, um, , I heard from some of the users and they were saying it's incredibly powerful, but it's taken us a year or two years to get to a point where we've, we've actually started to, to get ahead around how to use it and manage it. And it's so complicated.

And, um, and just, it was just listening to user stories and just hearing that it was, it was almost so difficult to set up that I was like, I'm not sure I can really be, be bothered. And you, you might say Kubernetes is a complicated setup, which it is, but it was actually probably easier. Than, than than Mesos or Mesos sphere.

The point being 

Avi Press: orchestration, just a complicated problem in general. Like there's not going to be a simple, you know, 

Matt Barker: kind of one size fits all. I'm probably wrong with this as well, but I think I, I saw Mesos , I, I saw Mesos being very closely connected to Mesosphere as a. As a company. Mm-hmm. , uh, so a little bit more tied to the, to the, to the, to the company.

Oh. Whereas Google had just given or put Kubernetes into the CNCF as part of the Linux Foundation. So it felt to me like there was more likely to be a bigger ecosystem that grew around Kubernetes than there would around Mesosphere. So, as much as I thought Meso Sphere might be successful, I, I felt like there was gonna be more space for me as a, uh, as a, as a founder in an open source project to be in one that was a bit more.

Uh, you know, a bit more protected from commercial interests and that's why, um, Kubernetes was also a good, a good, um, good choice. Yeah, that's a really 

Matt Yonkovit: good point because I think that when you look at building an open source company and you're building it around another open source ecosystem mm-hmm.

right? So you're building open source software on top of open source software, on top of open source software. Um, as we all are, you know, picking vendors that are, or projects that are single vendor controlled. Is difficult because it does kind of like sway a lot of the development efforts and a lot of the features and it, and I think it does, you know, potentially, you know, provide some risk, you know, because, 

Avi Press: but Macs was in the Apache Foundation too at the time, 

Matt Yonkovit: so it's not so, but e even so with being in the Apache Foundation, that's great and you know, whatnot.

But what you see, you start to see. E even with the Apache projects that are out there today, if you look at the core contributors, the majority of core contributors for Cassandra are in, you know, a single company or, you know, uh, you know, Kafka or, you know, like, so yeah. So you, you end up having. A lot of politics behind the scenes.

Mm-hmm. , but it still has a very slanted way. And if those developers are focused on enterprise features for their own open core version. Yep. It, yeah. 

Matt Barker: So I mean, you could make the argument that Google is, that is that commercial entity behind it, but I think the difference Google was that they A, donated it and b, It didn't feel to me like they were trying to use Kubernetes to make money from it directly.

Oh. What they cared about was their cloud service. And so that is a fundamental difference. And the, the compute, like rather than selling Kubernetes. Right. So again, I mean, I, I, and, and again, there are people that I was speaking to at the time and the people I've seen here, Who were very conscious of Google's influence in Kubernetes, and they were worried.

And I was like, yeah, I can kind of see that. But I also feel like this is a fundamental thing they've given to the community because they're trying to compete against Amazon and they're willing to give it up. And I think, and, and so it was a competitive thing against Amazon. But, but, and they, and they knew what they were doing in that.

And so therefore, I felt like it was, it was enough dissociated from Google that it was. It was gonna create its own ecosystem. Which, which it did because pretty soon afterwards, you know, you had, you had Amazon like creating an Amazon service around at Microsoft doing the same thing. Cuz Bren and Brendan actually being a founder who ended up being, uh, brought in by Microsoft, gave me a bit of confidence because I was like, Microsoft would not be investing in, uh, hiring one of the Kubernetes founders and trying to create a Kubernetes service unless they also had the confidence it was gonna, it was gonna be, be a standalone, um, project that could, that could create its own ecosystem around it.

So, But it wasn't by any, by any means, an easy question to answer. Put it that. 

[00:28:43] Measuring growth and adoption

Avi Press: Okay. Uh, yeah, with last question, um, so this is a little bit shifting gears a little bit, but, um, I guess we talked offline a little bit about the fact that Jets Stack had done kind of quite a lot of work, uh, to, you know, just generally in measuring metrics around Yeah.

Um, open source usage, and we talk about metrics a lot on this podcast, and I'm really curious when Jets Stack was first really getting off the ground. What were, what were you all looking at in terms of what you were trying to really measure and optimize, or 

Matt Barker: were you At the time we weren't really measuring much at all.

Like the only thing that I could measure it by without knowing much about how you would look at the, at the metrics at all was. Um, number of people coming up to me at conference talking about our project , which was quite, quite high, right? And the, the passionate, they were very passionate around it. So there was a, I, I distinctly remember a, a group of, um, guys from Finland coming over saying, you save so much time.

Here's a $200 bottle of vodka. We have to drink it together now. Thank you so much. Well, that's a good sign. But that and GitHub stars and literally it was stars. It wasn't even anything, anything else. Um, and, um, That was, in some ways that was, and then obviously, uh, number of people mentioning on Twitter, the number of people joining our Slack channel.

Um, the number of people from the CNCF that were contacting me about it, it, it was very messy and I didn't have any really good sense of it. Um, so yeah, it was, it was stars. It was my own subjective feelings on how many people contacted me about it and, and how. How they, uh, how they talked about it and responded to it.

And then, um, yeah, and then, and then the, the really exciting thing was when it started to show up in companies that we were talking to, and that was really exciting. But that, my, my concept and thinking around it changed immeasurably over time, uh, especially after we acquired because, um, we, uh, I looked in 2019 as to like whether we should take a product.

Uh, path and, uh, take funding around building products and trying to scale, um, that way or to keep with services, um, and being acquired, uh, for effectively an opportunity to, to build a product around Set Manager meant that we had to do a lot more work into looking at the metrics. And, and it's only really in the past two or three years, I remember a former site manager's been around since 2015 or 16, uh, in Cube, Lego.

It's only the past couple of years that I've really. Understood how the importance of metrics and actually how to go about it a bit, bit of a better way. And it's, um, it's, it's very illuminating and it's incredibly powerful and important. So anyone who's listening, you know, I would say it's something you should be thinking about sooner rather than later.

Matt Yonkovit: I, if there is there one metric now that you kind of, for, for you as a business at your stage kind of trumps 

Matt Barker: all. No. And like we, we, we, we, we struggle to actually, uh, to, to, we struggle to, to sort of boil it down. So actually what we've done is we've, we, we, we measure a, a lot of things. And then there are a few in our KPIs that we, that we keep, we keep track of.

Okay. And so I think, uh, it's, um, it's, it's percent is contributions. It's, it's still GI hub stars, it's growth of the Slack channel. Um, And it's, uh, Docker pulls. Mm. Okay. Those are the ones that we kind of like look at and, and sort of, yeah. Can tri triangulate into like a, like a, a like a health score basically.

Avi Press: Yeah. Yeah. Okay. That's a great set for people to, to look at. Definitely. Yeah. Absolutely. 

[00:32:04] Rapid Fire

Avi Press: And, um, we are about out of time. I feel like we had talked for many, many hours about this. Wait, we have 

Matt Yonkovit: two minutes. To do our rapid fire rounds. Okay. 

Matt Barker: Oh no's true. , 

Avi Press: you're gonna 

Matt Yonkovit: close me out before rapid fire. 

Matt Barker: You're, you're gonna try and catch me out here on something?

Avi Press: No, well, no. No. Like fun lightning. Okay. 

Matt Yonkovit: Well actually the first one is one I normally ask if you're gonna go back in time, you're gonna get a DeLorean and go back in time and meet your younger self, what's the one thing you're gonna tell 'em? Uh, one piece of advice, . 

Matt Barker: I think it would, it would be more, Everything's gonna be fine because I, I think fine, I think I have a, I have a tendency to overthink, I think most, a lot of founders do.

Oh, me too. And you, and you're kind of like, you're looking at every sort of possible avenue, and then you often go down the, the rat hole of like, oh, this is, this is gonna be much worse than it is. So just like trust. Everything's gonna be, everything's gonna work out. So I think even my future self, if I, if I came back to now, I'd be saying the same thing,

Fair enough. Yeah, 

Matt Yonkovit: fair enough. Now I normally ask, what's your first Lennox distro? But I don't have to ask that anymore. Right? Because that's It is. Yeah, yeah, yeah. Ask that way. Me too. So what is your favorite open source tool right now, outside 

Matt Barker: of your own? Well, I mean, I'm not, uh, I'm not actually, uh, an engineer by background.

I don't, I don't use technology a lot, but the one that I really like the look of, uh, and I'm seeing a lot of traction around is, uh, is Nix os. Absolutely. Yeah. 

Matt Yonkovit: Okay. Alright. Fair enough, fair enough. And, um, when you go up on stage, right, what's the theme music that they play or the walkup music to, to send you up to the top 

Matt Barker: of the.

Well, I mean, it's, I know it's not very upbeat, but I'm a massive Smiths fan, so I'm gonna go with a Smiths. Okay, well, 

Matt Yonkovit: fair enough. We're, we're gonna build a playlist from all our guests where we answer that question. . Um, so 

Matt Barker: favorite restaurant here? Favorite restaurant? God, that's a good question. Um, you cannot go wrong with.

There's, I I, I've got so many, there's so many restaurants have to one. This is so difficult to, like, where do I go? Okay. I'm gonna say one great all round restaurant. You go at any time your kids like it, your family's happy with it. Your parents are happy with it. Nandos, everyone likes a good Nandos. Have you heard of Vanos?

Avi Press: What is it? We're American . 

Matt Barker: It's um, it's basically a Portuguese style chicken, chicken restaurant that Oh. But, but it's done like fast food. And you said that Oh, alright. Yeah, 

Matt Yonkovit: yeah, we've got some of that. Let's just called up different, obviously. Yeah, I got it. 

Matt Barker: Allos. 

Matt Yonkovit: Matt, thank you. Oh yeah. Thank you for that.

We appreciate it. Um, you know, we, we, we'd 

Matt Barker: love having you on. I really enjoyed it. And, um, one of the terms you'll find out of, you know, you might use if you're in London, is let's go for Chiqui Nandos. So maybe we could go for Chiqui Nandos. That will be amazing. . 

Matt Yonkovit: Fair enough, fair enough. Uh, thank you everyone for watching and don't bring the light and subscribe.

Welcome to the Matt Cave: at the State of Open Con
Getting to know Matt Barker and Jetstack
The Early Days of Canonical: Selling Free
Support as a product, retention challenges
Trying to go to the propietary world from open source
Comparing MongoDB & Canonical
Bootstrapping Jetstack with Services
Moving from services to product
Convinced Kubernetes was the right space
Betting on truely open software instead of open source controled by 1 company
Measuring growth and adoption
Rapid Fire