William Morgan on Linkerd and Buoyant

This is Text Drunk tv. Hey everyone, welcome back here. The Text Drunk tv.

We're happy to have you, uh, join us this next guest. He is been on Techron TV many times in the past, but I don't know if we've had him on. I bet it's gotta be a year, maybe more even.

Um, his name's William Morgan. William is the founder and CEO of a company called, and I always, maybe I mispronounce it. Is it Buoyant, William?

Yep. Yeah, that's right. You got it.

Okay. I got it right. William is the founder and CEO of Buoyant.

If you, you're not, I'm gonna let William give you the whole history of it, but basically, William and the folks at Buoyant were the people behind Linker D one of the most successful projects in the, uh, CNCF Pantheon, which is, I think now like 180 something. Projects are the belief. Um, but also at the same time that they nurtured and helped Laker d become successful.

They had the dual task of making Buoyant a successful, uh, startup, you know, for profit startup. And, and those two things don't always go together. We're gonna talk about that and we're gonna talk about William's background.

But William, first of all, welcome back, man. It's been too long. It's great to have you on.

Thank you, Alan. It's great to be here. Yeah, it's been too long.

Yep. Um, you know what, before we jump into the whole buoyant linker D thing, let's talk a little bit about you, you know, how, how did you come to found and become CEO of Buoyant as well as kind of breathe life into this lake or D project? Well, I knew this is what I wanted to do from the moment I was born.

Lay ahead of me. Yeah, yeah. No, I, I came into this in kind of a circuitous route.

So I, and it's funny, with the kind of modern AI craze, that's actually where I started 20 years ago doing AI and natural language processing. And I went to school for that. And at the time it was not very easy to make progress, and it was not very easy to make products that were really useful.

And I think that kind of was a little disappointing. So, um, I joined, you know, a couple startups. I ended up joining Twitter in the early days of Twitter back when things were good, But it was still Twitter.

Yeah, yeah, that's right. And, and at Twitter, I kind of moved from that world, the world of natural language processing and AI to the world of infrastructure mostly. 'cause Twitter was having these really interesting and, uh, kind of very immediate infrastructure problems.

And so at Twitter we went through this very early transformation, and you see the echoes of this, you know, now with Kubernetes and Docker and microservices and all that stuff. We didn't really have any of those words. We didn't have any of the, that technology, but the ideas were the same.

And so when I left Twitter and I started buoyant with one of my colleagues from Twitter, the idea was let's take those ideas that we, and, and the open source technology that we had just explored, and let's bring that to the rest of the world. And that really was the, the genesis of Buoyant. Yeah.

I mean, with Twitter, a lot of it would, and, and not only Twitter, but a lot of the hyperscalers as we come to call them, you know, uh, it, it really was a question at scale, doing things at a scale that was kind of unimaginable, or, and you had to have, you know, these immensely scale scalable solutions. And, and quite frankly, a lot of people, it's almost counterintuitive, I think William, but a lot of people don't realize it, that it was the need for that scalability that drove a lot of the open source innovation that we saw during that period, because people realized, look, the the way to achieve scalability was utilizing open source. And I, you know, again, I say counterintuitively because a lot of people I thought would say, oh, no, that open source is great for toys and experiments, but you can't run mission critical hyperscale on open source.

Well, here we are. Right. Um, and, and so it's interesting how that turned out.

Um, let's talk a little bit about Buoyant and its relationship to Linker D and how that's evolved over the last years. Yeah, certainly. So, you know, it's funny because most of the time I'm either wearing one of two hats.

I'm either wearing my buoyant hat and I'm talking about the company, or I'm wearing my Linky hat and I'm talking about the project. And I typically try, you know, or historically I've tried to keep those two things really separate, you know, and, and did my best to represent linky, the open source, you know, CNCF project, kind of independent of Buoyant. And one of the things that we've done over the past year and a half, kind of explicitly, has been to merge those two things together.

And kind of the, the, the rationale for that, or the reason why is because like every other open source project, especially in the infrastructure world, that, you know, we're all going through this transformation. You know, we're all kind of in the post zero interest rate, you know, policy world, the post ERP world of, okay, funding is, you know, is not free flowing like it was before. Suddenly the requirement to have a real functioning business becomes, you know, very immediate and critical.

And a lot of open source projects that in the, in the past could kind of rely on the fact that, hey, there was always gonna be VC funding. There's always gonna be big company money. Suddenly they're all struggling.

And Linker d to be Frank, was no exception. So Linker d was sure in a spot, you know, even 18 months ago where we were like, everything that we're doing here is, is a paid endeavor. So, you know, the, the world of open source has evolved in many ways since, since I first got started.

Um, one of those ways is that most open source, certainly in the CNCF landscape is not Night Sun Weekends volunteers, it's paid, right? Every maintainer is paid. And arguably that's good.

That's what we want. So, you know, we've Explored this on webinars, we've recorded on interviews as well as our techron gang. Yeah.

The real question is, are they paid enough, right? Yeah. Yeah.

To work. No. Right.

I mean, it's, you know, people don't get rich and retire being a maintainer of an open source project, right? Right. Um, that's Notoriously thankless work, right?

Yeah. Comes in and they demand things and they run away. And Community says, what were you thinking and what are you doing?

And why didn't you do this? And where's Michael? There's no doubt about it.

William, there might be some folks out here who are saying, wait a sec, this linker d I've heard of it. I'm not a hundred percent sure. How would you describe Linker D to them?

Yeah, great, great question. So, linker D is a service mesh, which means that if you have a Kubernetes cluster, one of the nice things about Kubernetes, one of its real advantages is that it is designed as a, as a building block. So on top of Kubernetes, you start adding other objects, other items, and service mesh is one of those items that turns out to be very critical.

So what it does is it mediates all the communication within your cluster, all the network communication within your cluster and across clusters. And it gives you a bunch of capabilities that linkerd itself, uh, sorry, that Kubernetes itself doesn't provide things like encrypting all the data in transit, things like making cross cluster communication, something that's transparent to the applications. You can have, uh, services that span clusters.

Um, things like giving you a uniform layer of visibility, um, and observability into how services are performing, what's their success rate, what's their latency. So kind of a, it's a platform tool that you use when you're building a serious Kubernetes platform. And the evolution over the past 10 years, you know, of service mesh.

So linker has been around for almost 10 years, is in the early days. It's like, what is this thing? And do I need it?

And now, and now I think everyone's kind of come to terms with the fact that yes, a service mesh like Linkerd is a critical part of building any high scale or high reliability Kubernetes platform. You need to solve those problems. Absolutely.

Now, William, I, I know it's on YouTube about open source, right? I've been involved in open source for 25, yeah. 25 or more years.

Um, what people out there may not realize is that when an organization like the CNC at Lennox Foundation takes over the management, it's actually the ownership, if you will, of the ip of the, of the, uh, open source project. The licensing is under their discretion as well, right? And as you said, we, we've seen the whole VC supporting tech world kind of flip on its head, not, you know, there's a pendulum that swings.

I've seen it several times in my career. Um, but certainly, you know, VC funding dried up and, you know, they only wanted the adults at the table who were making money, no time for science experiments, right? And if you weren't making, if you're not profitable or getting profitable, you can have a really hard time attracting investment.

Some companies, I'm not naming names, took the path of, okay, well we're gonna change our licensing, right? We're gonna change our open source licensing so that you can't use it for what you really want to use it for without paying us. Now, linker d is A-C-N-C-F project could not change their licensing, but they're, the goodness is there's more than one way to skin the cat of making a business model out of an open source project.

Why don't, if you end up, don't say anything that's going to get us in trouble, but, uh, talk to us about open source Is Dead. Yeah, exactly. That's just what I need.

Go ahead. Alan Hates open Source. Yeah.

Alan hates open source. Um, talk to us about the patch. You guys talk.

Yeah. Yeah. So for, for me and there, there's so much to, you know, to compress into such a small conversation.

Um, I, I tend to be very pragmatic about this. So when I talk about open source, you know, there's a lot of emotions that can be involved. There's a lot of kind of idealistic principles.

For me, I try and be as, um, you know, as, as realistic as possible. And I try not to attach kind of like a, an emotional component to any of these decisions, even though I have been living, breathing, creating and consuming open source, you know, for, you know, maybe almost as long as you have Alan. Um, but so for me, it was a really pragmatic question, which was like, Hey, I have to pay the linker d maintainer, linker D's a great project.

People are using it everywhere. I have to pay the linker d maintainers. And somehow, you know, and regardless of VC funding or whatever, you know, that's, that's all just kind of describing the environment, but somehow I have to pay the maintainers.

And what's more, you know, the, the, the thing that we were finding, which was really scary to me is, you know, one of the ways that we paid the maintainers was we sold support. So we would say, okay, you know, we will, you can wake us up at three in the morning, like, here, you know, sign up and we'll, you support. And we had a good number of customers who signed up.

And then when things got tight, they all went away. They said, you know what? Querty works.

We don't actually need you. We haven't called you for, you know, for a year. Why would we continue paying for that?

And that was a big wake up call for me as the CEO, obviously, because the money went away. But also because what it made me realize was we had set ourselves up in this downward spiral where the only way we would make money in that situation, the only way was by making liberty worse, because Liberty was good. They didn't need the commercial offering.

And that was a huge wake up call, you know? And I said, okay, whatever we're doing, if we want maintainers to continue to get paid, which we do so they can advance the project, we need to flip that on its head so that when we do something good, when we make Ty good, we get paid. And when we make linker bad, we don't get paid.

Not the opposite. Right? And so that kind of led us down this path, and you're right, you know, the CNCF, you know, has certain limitations on what you can do in terms of licensing.

I think, you know, there's options where we could have fought against that if we really wanted to, but for me, it was a pragmatic decision. And what we ended up doing, which I think was quite, uh, elegant, was we said, open, you know, the license doesn't change. Linker d remains totally open source, but the stable release packages, those things that we build, that you consume, that you pull onto your clusters, those we are gonna do as part of Buoyant, not as part of Ty, which means you're gonna have to pay for those.

We tried to do it in the easiest, simplest way possible. We put it in exclusion. If you're a small startup, it's totally free.

You know, things like that to, to really kind of like, you know, keep the, the magic of this community alive. And, but ultimately our goal was if you are bringing Linkerd into an environment where you are making a business, you're building a serious business on top of this, then you need to somehow feed some of that value back to the project so that we can pay the maintainers. And that was a big gamble.

This happened in February. It was a big gamble on our part, but it seemed like the right thing to do. And you know, now eight or so months later, very happy to report it works.

So, yeah. And we're, we're actually, you know, it's been amazing for the project. We've added maintainers, we've added velocity.

We're shipping features that have been on the backlog for years. It just feels so good to have This. No, no.

It elevates the entire community. I mean, it really, it really does. I mean, I'll, I'll tell you, William, this reminds me of, so back 2000, 2001 guy named Marty Rush, I don't know if you've ever heard of Marty.

He, uh, working for the Gov US government in his spare time creates what back then was called an intrusion detection service. An ID IDs called Snort, funny name, big was the logo Go. Yeah.

And, um, was fantastic. It became the defacto standard. So much so that almost every commercial IDS out there was actually using Snort under the covers with a pretty interface, including a company I helped co-found.

Mm-Hmm. Well, it's still secure in 2001. And, um, well we did something a little different.

We took an IDS and turned it into an IPS. So rather than just detecting, we actually were blocking very clunky Back then you had to send a signal to the firewall and the firewall. Mm-Hmm.

Would block it. It wasn't in line by the time that did it. The, the, the attack was over and done with.

Anyway, but that's a another story. But Marty and the source, the company, the commercial company was called Sourcefire. And, and what he did is he couldn't take Snort out of open source.

We didn't have Linux Foundation in that stuff then. Sure. But he didn't want to take snort out of open source.

But snort, much like what you are talking about, snort needed signatures. 'cause it was, it was a signature matching is how it worked. Right.

So he said the, the razor was still open source. The signatures were the razor blades. They weren't open open source and you had to subscribe to the feed for the signatures.

Yeah. Yeah. And if you did it, he still released them, but they were released, I forget if it was seven days or 30 days delay.

Yeah. So if it wasn't real important to you, you wanna wait 30 days to get ahead. But if security was important to you, and security tends to be an important thing, most times you would pay the subscription.

Yeah. I had another friend, Ron Gula, who started a company called Tenable Network Security. It's called Tenable now.

Nessus Open Source Vulnerability Scanner. I don't know if you've ever heard of that one. Mm-Hmm.

Everybody used that. 7. Nessus was no longer open source.

You could still use it for free, but you didn't get the latest Nale scripts. And so you had to buy Tenable. 6 billion to Cisco.

Ron took Tenable public and made much money as Marty did. Uh, they're both fragile and light. Um, but two different paths from monetizing.

Yeah. The open source, uh, con you, it sounds like you took more of the source fire path than the Nessus path, or do I got 'cause kind of somewhere in between, because there is no kind of equivalent of the, the signatures, you know, there's no razor blade equivalent for us. Right.

Um, but I think in all cases, you know, the, there's so much, you know, like I said, there's so much emotion and kind of like ideology associated with, with open source, and it's very easy to get pulled into strange directions. I think ultimately the, the companies and the projects that are successful are the ones that are really focused on are we solving actual problems for people? And if we are, then how do we do more of that?

And I think if you can stay focused on that, then all the rest just kind of falls away as noise. You can ignore the rest of that. Your only goal is how do I solve problems for people and everything else will fall away.

And I think that's kind of the attitude that we adopted. You know, of course, you know, it's, we, we try and do it as nice and as friendly as as we can, but ultimately our job is to solve real problems that real people are having. And when we do that, we need the value that's being created to fund, you know, right back into the project.

So we can do more of the same, You know, in both of those situations. I mentioned to you, the outcry from the open source community was hysterical. Right?

I mean, there were some people who said, Hey, you're entitled to make a buck. Right? Someone's gotta pay for this.

You know, Ron said, you have free is in freedom and free is in beer, can't always have free beer. Um, but that being said, that too passed. Right?

And that, and, and someone once told me a guy's much smarter than me. I so, so sold my first company to him, said, the ultimate value of what you are doing is what someone's willing to put their hands in their pocket and pay you for. If they're not willing to put their hands in their pocket and pay you for it, they don't truly value it.

And I think that's the lesson of linker dem employant, you know? Right. If if they're willing to put their hands in their pocket, then you're creating something that's valuable.

Right. And that's how the market works. Man.

I, I can't, I don't know a different way of explaining it. And, you know, and, and for something like linker d or almost every open source project, I think that, you know, the, the playing field is, is level by design. So it's, you know, everything that's Apache V two licensed, you can start a company around today.

You can do whatever you want with it effectively, you know, with like very limited exceptions. So, you know, it kind of like the moral authority to tell someone, well no, you're doing something wrong. You should be doing something bad.

You know, if you think that's really true, then go do it. If you really think that what they're doing is wrong, then go do the alternative and see how it goes. Absolutely.

And, and look and open source, you can, right. You could fork. Right.

There will forks of, of Nessus that not anywhere near as successful as many Forks are, but every once in a while you get a fork of Hudson that becomes Jenkins. Mm-Hmm mm-Hmm. So go figure.

Anyway, we're Open tofu, you know, Tofu. Now you know, though, you know, I'm not getting involved. Um, separate, separate that for next time.

Yeah. Uh, next time. William, it's great having you back on and having this, I'm really, really happy to hear that things are working out for you guys over a buoyant with Laker d um, I assume I'll see you maybe a cube card.

I'll be there. Yeah. I'll, I'll be attached like a barnacle to the Linky booth in the Kuka or in the project pavilion.

So if anyone is there who's listening in, please swing by and say hi. Hey, we got a cool breakfast going on Tuesday morning, early before the festivities start right by the right by the conference. I'd love for you to come up, grab some, it's a nice breakfast.

No. Okay. No nothing.

It's just coffee, food. And sit around BSing. So If you got coffee, I'll be then It's good coffee.

All right. William Morgan, founder and CEO of Buoyant here on Techstrong tv. We're gonna take a break.

We're back with Mark.