Giant Robots Smashing Into Other Giant Robots

489: CTO Lunches with Kendall Miller

August 24th, 2023

Kendall Miller is the Co-Founder and COO of CTO Lunches, a network of engineering leaders to get trusted advice and connections.

The first half of the conversation with host Victoria Guido and special guest host, Joe Ferris, CTO of thoughtbot revolves around the use, adoption, and growth of Kubernetes within the technology industry. The discussion explores Kubernetes' history, influence, and its comparison with other platforms like Heroku and WordPress, emphasizing its adaptability and potential.

The second half focuses on more practical aspects of Kubernetes, including its adoption and scalability. It centers on the appropriateness of adopting Kubernetes for different projects and how it can future-proof infrastructure. The importance of translating technical language into business speak is emphasized to influence executives and others in the decision-making process and Kendall also discuss communication and empathy in tech, particularly the skill of framing questions and understanding others' emotional states.


Become a Sponsor of Giant Robots!


VICTORIA: This is the Giant Robots Smashing Into Other Giant Robots Podcast, where we explore the design, development, and business of great products. I'm your host, Victoria Guido. And with me today is Kendall Miller, Co-Founder and COO of CTO Lunches, a network of engineering leaders to get trusted advice and connections. Kendall, thank you for joining me.

KENDALL: Thanks for having me. I'm excited.

VICTORIA: And today, we have a special guest host, Joe Ferris, CTO of thoughtbot. Joe, thank you for joining us.

JOE: Hello there. Thank you for having me.

KENDALL: Hi, Joe. Thanks for being here. It's exciting.

VICTORIA: Yes. It's so exciting. I think this is going to be a great episode. So, Kendall, I met you at a San Diego CTO lunch recently, and I know that's not the only thing that you do. So, you're also an advisor, a board member, and CXO. So, maybe tell us a little bit more about your background.

KENDALL: Gosh, my background is complicated. I've been involved in tech for a very long time. In college, I worked for a company that started Twitter about five years too soon, and then worked in the nonprofit space in China for ten years, then came back, got back involved in tech.

Today, I'm usually the business guy. So, when technical founders start technical products and want help turning them into successful technical businesses, that's when they call me. So, I have the technical background. I have never been paid to write code, which is probably a good thing. But I can hang in the technical conversations for the most part, but I'm much more interested in the business side and the people leadership side of business. So that tends to be where I play. Every organization hires me to do something different.

VICTORIA: Thank you for that. And I'm just curious about the CTO Lunches. Just tell me a little bit more about that. And what's the idea behind it that led you to co-found it?

KENDALL: CTO Lunches has actually been around for about eight years. And I didn't start the initial incarnation of it. It was two people that got us started, and I was trying to hire one of them; one thing led to another. Actually, originally, they did not want me to join. I think, at the time, my title was COO at a company that I was working with.

About six months later, I took over engineering as VP of engineering, and then they're like, you can join the group now. We're less strict about that [laughs] now. Although it is highly focused on senior engineering leaders, it's not exclusively CTOs. But the group's been in place for a very long time, just intended as a place to network, have conversation with people who are in that senior-most technical position at technical organization.

So, the CTO role is a lonely role. CTOs get fired all the time. There's not a technical person at the company that doesn't think they can do the job better than them. So, the CTO is always getting feedback. You're doing this wrong. The trade-offs you're making are wrong. This isn't going where it should be going. We should automate that. Why haven't we automated that? We should switch to this other tool. I've used it before; it's 100 times better. Joe, let me know if I'm getting any of this wrong. But that's the experience that I've had.

Having a place where people can get together and, you know, half the time just complain to each other, hey, this is hard, is really why the networking group exists. So, it's a listserv. And there are local lunches that started in Boulder, Colorado. It's gotten pretty global.

About a year ago, a little over a year ago, I was talking with one of the people who'd gotten it started. I've been involved in the Denver chapter for most of those eight years. And I was suggesting to him that he change a few things about it, to monetize it so that he could invest in it further. And he came back a few months later and said, "I want to take your advice and do this, but I want you to come do it with me."

So, we founded the company officially...I think in December is when paperwork went into place. And we started investing in it a little bit more heavily. I was living in Europe last year, so we went and put on lunches in Paris, and Lisbon, and London, and, gosh, all over the place. I'm sure I'm missing some, Amsterdam. But there's been chapters all over the U.S. and a couple of other parts of the world for a long time.

VICTORIA: That reflects my experience attending a CTO lunch. It's just very casual, like, just get together and eat food and talk about what you've worked on recently, issues you're having, just get ideas and make some friends. So, I really appreciated the group, and I'm going to personally plug the San Diego Chapter has picked up again. And we're meeting next Friday down in Del Mar. And we're going to be meeting on the last Friday of every month through October. So, I'm super excited to be a part of the group.

And Joe, yeah, I'm curious about your perspective. As a CTO with thoughtbot, just what are your thoughts about that kind of thing?

KENDALL: Yeah. How right am I about how lonely you are, Joe?

JOE: [laughs] You know, I've been lonelier since we went remote. I used to work in the office, and I was a CTO, but also, I had lunch with people, which was nice. So, I'm lonelier. But yeah, I think everybody needs a group like that, like, senior developer therapy just to talk about your woes together, drown your sorrows.

KENDALL: Well, I think years ago, I heard that CTOs are the most fired C-level executive.

JOE: You're making me nervous now.

KENDALL: [laughs] You've been there a long time, Joe. I know you've been there a long time. If you haven't been fired yet, you probably got a little while longer in you. This will be really awkward if it's published and you've already been fired.

VICTORIA: We can always edit that out afterwards.


KENDALL: Yeah, no, I think it is a particularly lonely position. And, again, I think a lot of it is the average engineer in a technical company doesn't look at the COO or the CFO or even the CEO and think I could do that. But they're all looking at the CTO and thinking, what does that person do that I can't do?

It's ridiculous because most of them would make terrible CTOs because it does require some of the business sense. Or, you know, right out of the gate, they might make terrible CTOs. It actually is quite a skill to be the most technical person and speak the business language. I mean, am I right about that, Joe? Like, was that hard for you to learn?

JOE: Yeah, I definitely, my background is also technical. I have a background in consulting. So, I always did a lot of metaprogramming, if you will. But making that transition to thinking about organizations that way, thinking about how all the other pieces play into it, was a pretty big step for me, even before I became a CTO as a consultant.

KENDALL: Well, because you can't just chase the newest, hottest technology. You have to make business trade-offs. And not everything can be resume-driven development, right? Even if that technology over there is newer or hotter, it doesn't mean you have a business model that supports it. And it doesn't mean that migrating to it can be done, right?

JOE: Yeah. I mean, even beyond choosing technologies, just choosing where to invest in your software stack, like, what needs to be reworked, what doesn't, and trying to explain those trade-offs, I think, is a rare skill. Being able to explain why something would be harder than something else when you're working with the leadership to prioritize a backlog it's a puzzle.

KENDALL: Well, and I think when I'm in an executive conversation, and the CTO says, "Here's the thing that I think is the best decision technically, and I think it's the wrong decision for the business because of X, Y, or Z," I'm always super impressed, right? Like, this is the right technical solution for what we want. However, we shouldn't pursue that for business reasons right now. Maybe we can in six months, but right now, we need to prioritize this other thing. I don't know, that's always when I feel like, oh, this person knows what they're doing.

JOE: There's nothing more dangerous to software than a bored developer. [laughs] One nice thing about being a consultant is that I don't have to invent problems to solve with technology at my company because sooner or later, I'll run across a company that has those problems, and I'll get to use that technology.

But I think a lot of people are mostly happy...they might be happy in their role. They might be happy with our team. But they're very interested in whatever is hot right now, like machine learning, AI. And so, suddenly, that surreptitiously makes its way into the tech stack. And then, years later, it's somebody's problem to maintain.

KENDALL: [laughs] Well, I have a specific memory of a firm in New York City that was, you know, this is relevant to y'all as thoughtbot is that, you know, at least historically, it was, to me, the premier Ruby on Rails consulting shop. I think that's still largely y'alls focus. Am I right about that?

JOE: We still do a ton of Rails, yeah.

KENDALL: Okay. Well, so this organization was all Ruby on Rails. It was a big organization. They had a very large customer base. And they hired a new CTO who came in, told everybody in the company they were stupid, laid off 70% of the engineering organization, and told the CEO he was going to completely rewrite the product from scratch in .NET, and he could do it in three weeks. And I'm pretty sure the business went under about three months later [laughs] because that was just so outrageously nuts to me.

JOE: It's too bad he laid everybody off beforehand. I've been in that situation where somebody tells me, "I'm going to rewrite this. It'll be ready in three weeks." And I could fight with them and try and convince them they're wrong. But I feel like somebody who's approaching that with that attitude they're missing all of the nuance and context that would make it possible to explain to them why it's not going to work. And so, it's easier to just say, "You know, take the three weeks. I'll talk to you in three weeks." But if you've already laid off your development team, that's hard [laughs] to recover from.

KENDALL: That's exactly right.

VICTORIA: There's got to be a name for that kind of CTO who just wants to come in and blow everything up [laughs]. Yeah, so you spend a lot of time talking to different CTOs and doing this social networking aspect. I'm wondering if there's, like, patterns that you see. You've mentioned already one about just, like, the most often getting fired. [laughs] But what are the patterns you see, like, in challenges, and then what makes someone successful in that CTO role?

KENDALL: Well, oh gosh, I have so many thoughts about this. First of all, I run into a couple of different categories of CTOs. There's a lot of people who come to CTO Lunches who are small company CTOs. I mean, it makes sense that there's a lot more small company CTOs than there are big company CTOs. But the small company CTO who maybe it's their first gig in the role or they're a serial CTO. There's the fractional CTOs that come that are doing it across several different organizations at the same time, and then there's the big company CTO who shows up. And honestly, all of their problems are very different.

The thing that they have in common is even at a very large organization, in that position, they can make a decision that causes the company to go under. So, there is a significant amount of volatility in the amount of power that they wield. So, what's interesting about that is not everybody understands that.

And so, first of all, there's the kind of CTO that just doesn't get that, and that doesn't matter if they're fractional, or a small company CTO, or a big company CTO. If they don't understand that, they're going to cause significant problems, right? Like the person I just mentioned who said, "I can just re-platform this in three weeks in .NET." There's that.

I mean, I think, as with any senior leadership position, the comfort with volatility, the ability to know what to communicate down versus across and versus up, and then the ability to speak the business language. For everybody, the CFO's job is to communicate the financial needs alongside of the business leads, right?

If the CFO's sole goal is to cut costs or make sure we're running as lean as possible, they're a bad CFO. But they're not as good of a CFO as the CFO who can say, "Hey, we're underspending right here. And I can look at the numbers and know we should invest more there. How can we invest more there and invest it well?" And it's the same thing for a technology executive to be able to look at the business context and communicate it back.

And there are so many CTOs that I've worked with who they're the most technical person in the room, and they know it. And as a result, they're just a jerk to everyone around them, like, everything you did here was wrong. You know, that's where they fail. And so, if they can communicate the business needs, navigate the volatility, and support a team that's going to make decisions that aren't always the same decision they're going to make, they're going to be successful. Honestly, there's very, very few CTOs that I've met like that. People who are excited to meet you at work, excited to see you succeed, excited to see that you went and built a thing is great.

I mean, the reason I was VP of engineering is the CTO that I was working with at the's a terrible story. There was an engineer who had seen something that we were doing on repeat all the time and, in his spare time, spent about 40 hours outside of work, not during work hours, automating this task that we were doing regularly. And it was related to standing up a whole bunch of things in our standard infrastructure.

He brings it to the CTO and says, "Look what I built." And the CTO, instead of saying, "Hey, this is incredible. Thank you. This is going to save us a bunch of time. Let's iterate on it. Here's some things I'd like to tweak. Can we bring it in this direction? Can we..." you know, whatever, said, "Why is this in Python? It should be in Ansible," something like that. I can't remember. And the engineer literally burst into tears. [laughs]

JOE: Oh my God.

KENDALL: [laughs] Well, I mean, yeah, it was like; literally, that's why the CTO stopped managing people that day. There's a lot of examples that I have like that. Joe, I appreciate that your response is, "Oh my God." Because I think there's a lot of people who'd be like, wait, what was wrong with that? Shouldn't it have been in Ansible?

JOE: [laughs] Yeah, I've seen CTOs come into primarily two groups. One is the CTO who just tells, you know, like, they make the decisions, and they tell everybody what to do. They obviously don't have all of the information because you can't be in every room all the time. And the other is the CTO, who just wants to be one of the team members and doesn't make any decisions and tries to get people to make decisions collectively on their own without any particular guidance or structure. And finding that middle spot of, like, not just saying, "Hey, everything's in Ansible," allowing for the creativity and initiative, but also coalescing the group into a single direction, I think, is what makes a good CTO.

KENDALL: Well, yeah, because the CTO does have to say no, sometimes, right? Like, the best product, people say, "No." Good CTOs say, "No." There is some amount of, hey, I need you to come to me with trade-offs about this. Why are you going to make that decision? And I'm sorry, you still didn't convince me, right? Like, I mean, those are appropriate things to say. But yeah, I'm with you on that. You said they fall into two categories. But you really mean the third and that middle ground. Is it easy for you to walk that middle ground, Joe?

JOE: I wouldn't say it's easy. [laughs]

KENDALL: Yeah. Well, I'm always nervous to say something. I'm doing well because I know there's a report out there that can point at every time I failed at it, right? So...


Are you an entrepreneur or start-up founder looking to gain confidence in the way forward for your idea? At thoughtbot, we know you’re tight on time and investment, which is why we’ve created targeted 1-hour remote workshops to help you develop a concrete plan for your product’s next steps.

Over four interactive sessions, we work with you on research, product design sprint, critical path, and presentation prep so that you and your team are better equipped with the skills and knowledge for success.

Find out how we can help you move the needle at:

VICTORIA: Yeah, what I'm getting from what you're saying, too, is this communication ability and not just, like, to communicate clearly but with a high level of empathy. So, if you say like, "Well, why is it in Python and not Ansible?" is different than being like, "What makes Python the best solution here?" Like, it's a different way to frame the question that could put someone on the defensive that just really requires, like, a high level of emotional intelligence.

And also, if they've just worked, like, an 80-hour week, [laughs] I probably would maybe choose a different time to bring those questions up and notice that they have been burning the candle at both ends and prioritize getting them some rest. So, speaking of, like, communication and getting prioritization for [inaudible 15:34], especially on, like, infrastructure teams, maybe we could talk a little bit about Kubernetes, like, when that comes up as an appropriate solution, and how you talk about it with the business.

KENDALL: My background with Kubernetes is long because a company that I still work with, Fairwinds, used to be called ReactiveOps, has been in the Kubernetes space for a very long time. I think we were one of the very first companies working with Kubernetes. It was coming up that people were running into the limits of something like Heroku, right? And I think it's Kelsey Hightower who said every company wants a PaaS. They just want the Paas that they built themselves. And that's really accurate.

And I think Kubernetes isn't quite a framework for building your own PaaS or isn't quite a foundation where I think of a foundation for a house. Instead, it's more like rebar and cement and somebody saying, "Good luck, buddy." You know, you still have to know how to put the rebar and cement together to even make the foundation, but it is the building blocks that help get you to a custom-built PaaS. And it's become something that a lot of people have landed on as, you know, the broadly accepted way to build cloud-native infrastructure.

The reason I've been in the Kubernetes space and the space that I see Kubernetes still filling is we need to standardize on something. We can choose a cloud provider's PaaS. We can choose a third-party PaaS, or we can standardize on something like Kubernetes. And even though we're not going to migrate from AWS to Azure, the flexibility that Kubernetes gives us as a broadly adopted pattern is going to give us some ability to be future-proofed in our infrastructure in a way that previous stacks were not, you know, it was Puppet, and it was Ansible. And it was SaltStack. And it was all Terraform all the time.

I'm not saying those things don't exist anymore. I'm saying Kubernetes kind of has won that battle. Joe, since you're here and I know y'all are doing some Kubernetes work now at thoughtbot, I'm curious if you agree with that characterization.

JOE: Yeah, I think that's true. I think it's the center for people to coalesce around. Like, for an effort in the industry to move forward, there needs to be some common language, some common ground. And I think Kubernetes struck the right balance of being abstract. So, you can use it in different environments but still making some decisions, so you don't have to make them all.

And so, like, all of the things you had to do with containers like figuring out what your data solution is going to be, what your networking solution is going to be, Kubernetes didn't even really make those decisions. [laughs] They just made a platform where those decisions can be made in a common way. And that allowed the community and the ecosystem to grow.

KENDALL: I mean, I think of it a lot like WordPress; you know, WordPress is hated by many. When WordPress came out, it was hot, right? And it was PHP, which everybody was super excited about at the time. Kubernetes is going to reach a point where it's as long in the tooth and terrible as people think WordPress is, but it has become the standard. And the advantage of the standard is you can use the not standard. You can go build a website in Jekyll instead of WordPress, and there's going to be some things that are nicer about Jekyll.

But because WordPress is so broadly adopted, there's a plugin for everything. And I think that's where Kubernetes sits is because it's become so widely adopted everybody's building for it. Everybody's adapting for it. If you run into a problem, you're going to find somebody else out there who has that problem.

In fact, I think of one organization that I know that was on HashiCorp's Nomad. And they said, "Actually, we think Nomad has better technology through and through. But we think we're the only company at this size and scale using Nomad. And so, when we run into a problem, we can't Google for it. There's no such thing as a plugin that exists to solve this. Nobody has ever run into this before on Nomad. But there's 100 companies dealing with the same problem in Kubernetes, and there's ten solutions." And I think that's the power that it brings.

VICTORIA: So, it's not just a trend that CTOs are moving towards, you think.

KENDALL: I mean, I think it's already won the battle and the hockey stick of adoption. We're still right at the very bottom of that tick-up because it takes people a long time to adapt new technology like this, especially in their infrastructure. It's a big migration, to move. So, I don't think it's the widely adopted infrastructure technology even yet. I think a lot of the biggest organizations are still running on things that predate Kubernetes. But I think it has won the battle, and it is winning the battle and is going to be the thing going forward, so yeah.

JOE: I think it also has a lot of room to grow still. Like, there are other technologies that I used previously, like Docker, and they were a big step up from some of the things I was doing at the time. But you quickly hit the ceiling, or it was, like, I don't know where to go with this next. I don't know what else is going to happen.

Whereas with Kubernetes, there are so many directions it can go in. Like, the serverless Kubernetes offerings that are starting to pop up are extremely interesting, where, you know, you don't actually maintain a cluster or anything. You just deploy things to this ethereal cluster that always exists. And so, that sort of combination of platform as a service, function as a service, Kubernetes, as that evolves, I think there are a lot of exciting things that have yet to come in the Kubernetes space.

KENDALL: Well, so say more about that, Joe, because I've been going to KubeCon for a very long time, maybe...I don't know if it's 2016 or so when I first went. And it felt for a number of years...maybe those first four-ish years it was always the people at KubeCon were the, like, big dreamers and thinkers and, like, we're here to change the future of cloud infrastructure. And this is going places, and we're excited to be here and be a part of it. And here's what I'm going to do that changes the next thing.

And I feel like now if I go to KubeCon, it's a lot of people from, you know, IBM and some big bank that are, like, deep sigh, well, I have to adopt Kubernetes. I need to know what the vendors are. What do you guys do, and how does this work? Can you please teach it to me? Because I'm being told by my boss, I have to do it.

I don't see that excitement around Kubernetes anymore. The excitement I see is all around further up the stack, you know, things like Wasm, WebAssembly, or eBPF, the networking things and tracing things that are possible. Maybe that's further down the stack. I guess it depends on how you think about it, but different part of the stack. So, I'm curious, touching on the serverless components of Kubernetes; sure, I get that. And I do think, increasingly, the PaaSs of the future are all going to be Kubernetes-based, whether that's exposed or not.

But where are the places that you think it's still going to go? Because I feel like it's already gotten boring, maybe in a positive way. But I don't see the excitement around it like I saw a few years back. And I'm curious what else you think is going to happen.

JOE: Yeah, I mean, I don't think I disagree. I think Kubernetes itself, the core concept, is, like, it's still changing. But you're right that the excitement about Kubernetes existing has gone down because it's been there for a while. But I feel like the ecosystem is still growing pretty rapidly. Like, the things you mentioned, like Wasm and Istio, and all the tools in that ecosystem that continue to grow, is where I think the interesting things will happen. Like, it's created this new lower-level layer of abstraction that makes it possible to build concepts and technology that could not have existed before.

KENDALL: Yeah, well, and I'm, you know, talking to people who are working really hard at making short-run ephemeral workloads work better on things like GPUs for the sake of AI, right? Like, I mean, there is some really interesting things happening, and people are doing this in Kubernetes. So, I get that. I agree with that.

It is interesting that Kubernetes has become sort of the stable thing, and now it's about who can build the interesting add-ons. It's almost like, okay, we've built Half-Life. What is Counter-Strike going to look like? You know. That's a terrible (I'm aging myself.) example. But still.

VICTORIA: I think it's interesting, I mean, to look at the size of the market for platform engineering right now. In 2022, was 4.8 billion, and it's estimated to be in 10 years $41 billion. So, there is this emerging trend of different platform engineering products, different abstractions on top of Kubernetes. And I wonder what advice you would have for a technical founder who's looking to build and solve some of these interesting issues in Kubernetes and create a business around it.

KENDALL: Well, okay, let me clarify that question. Are you thinking, I'm a startup, and I need to build my infrastructure, and I'm going to choose Kubernetes. What advice do I need? Or are you thinking, I am founder, and I want to go build on the Kubernetes ecosystem. What advice do you have?

VICTORIA: Now I want to know the answer to both. But my question was the second one to start.

KENDALL: One of the things that is hard about the Kubernetes ecosystem is there's not a ton of companies that have made a whole bunch of money in Kubernetes because, as I said, I still think we're actually really early in the adoption curve.

The kinds of companies that have adopted Kubernetes are the kinds of companies that don't spend lots and lots of money on an infrastructure. [laughs] They're the kinds of companies that are fast-moving, early adopters, or, you know, those first followers, and so they're under $100 million companies for the most part. Where the JP Morgans and Chase are running Kubernetes somewhere in their stack, but they haven't adopted it across the stack to need the biggest, best tools about it.

So, the first piece of advice that I'd give is, be a little wary. It's still very early to the market. Maybe now is the time to build the thing. When ReactiveOps pivoted to Kubernetes, I think it was six months of having conversations with companies who were just, like, so excited about it, and this is definitely what we want to do. But nobody was doing it yet. You know, it was, we have, like, six solid months of just excitement and nobody actually pulling the trigger. And, you know, we were a little too early to that market. And that was just the people adopting it.

So, I think there is some nervousness that cloud-native solutions the only people who are really making money in Kubernetes are named Amazon, Google, and Microsoft because it's the cloud providers that are making a ton off of it. Now, there's Rancher. There is StackPointCloud. There's a few others that have had big exits in this space. But I don't think it's actually as big of a booming economy as a lot of people think, in part because EKS is an incredibly amazing product.

Like, eight years ago, the thing people paid us the most to do at ReactiveOps was just stand up Kubernetes because it was so stinking hard to just get it up and working. And now you click some buttons. Anybody can go do that. So, it's changed a lot, right? And I think be wary when you're entering that ecosystem.

And then, my advice to the founder that's not building on the ecosystem but just looking to adopt a technology that's going to be a future-proofed infrastructure is just adopt one of the cloud-native platforms. And there are a whole bunch of sort of default best-in-class add-ons out there that you need to throw in. Don't adopt too many because then you have to maintain them forever. That's the easiest way to get started. You can figure out all the rest of it later. But if you go use EKS, or GKE, AKS, you can get started pretty easily and build something that is going to be future-proofed. I don't know, Joe; I'm curious if you disagree with any of that.

JOE: Well, I think it's interesting to think about who's making money in Kubernetes. Like, I think there might not be as many companies who are doing only Kubernetes and Kubernetes-focused products that are massively successful. But I think because it has had a good amount of adoption and because it's easier to work with something that's standardized, it has helped companies sell things that they wanted to sell anyway. Like, all the Datadog, all the Scalas, the logging companies, they all have Kubernetes add-ons.

And now everybody is paying Datadog [laughs] to have a dashboard for their Kubernetes cluster. I think they're making more money than they would have been without targeting the market. And so, I think that's really...if you want to get into the market, it's not, like, I'm going to build a Kubernetes product. It's if I'm building operations and an infrastructure product, I should definitely have it work with Kubernetes, and people will want to click and install it.

KENDALL: So, to be clear, you know, one of the companies that I work with is called Axiom, and they play in the same, you know, monitoring, observability space as Datadog does. And part of what makes Kubernetes interesting in that space is in a microservice environment; there's so much happening. Where are problems being caused? We don't live in a day where I can just run my code, and it tells me that there's an unexpected semicolon on line 23, right? Like, that still happens. You're still doing those things.

But this microservice talking to that microservice is where things tend to break down. Did I communicate this correctly? What was sent? What was received? Where did it break down? What was the latency? And if you were doing things in the old way back when you were standing it up with, say, Ansible, or Puppet, or something like that, and you were orchestrating all of these cloud virtual machines, you had to really work hard to instrument the tracing and logging and everything involved in order to track what was going on.

Whereas that's one of the magic things about Kubernetes is with a few of the add-ons or some of the things out of the box with Kubernetes, it's a couple of clicks to get so, so much of the data and have insight into where things are going and what's going wrong. And so, I 100% agree with that.

Kubernetes is generating a tremendous amount of data. And if you're a data company, it's really nice to have all that come in, and it helps them make money, helps the user of Kubernetes in that situation understand where problems are happening and breaking down. Yeah, there's definitely some network effects of what Kubernetes is doing in that. I completely agree.

JOE: I think there are also some interesting companies, like, where they make...Emissary, Ambassador, and they have that sort of dual --

KENDALL: Komodor, is that --

JOE: Yeah, maybe. They have open source, but then they have a product.

KENDALL: You're thinking of Ambassador Labs.

JOE: Yeah. Ambassador Labs, yeah. I guess I don't really know how much money they're making. But I think that's a really interesting concept as people who make open-source things then make a well-supported product built around it.

KENDALL: Sure. What's interesting is, I think in the VC world, at least right now, and it may pick up again, but post-Silicon Valley Bank nearly caving in, I think that the VC tolerance for, yeah, just go get a billion open-source adopters, and we'll figure out how to monetize later I think that the tolerance for that is a lot lower than it was even six months ago.

JOE: Yeah, I think you have to have a dual model right from the beginning now.

KENDALL: Yeah. Agreed.

VICTORIA: You got to figure out how to make money on Kubernetes before you can. [laughs]

KENDALL: You know, minor detail. That's why I think services companies in this space still have a lot going for it. Because in order to even be able to sell software to a company using Kubernetes, you half the time have to go stand up Kubernetes for them because it is still that hard for so many people to really adopt it.

VICTORIA: Yeah. And maybe, like, talking more about, like, when it is the right decision to start on Kubernetes because I think the question I get sometimes is just, is it overkill? Is it too much for what we're building? Especially, like, if you're building a brand-new product, you're not even sure if it's going to get adopted that widely.

KENDALL: I mean, and I'm [laughs] curious your thought on this, Joe, but there's a good argument to be made that Heroku was enough for the vast majority of founders early on. But the thing is, Kubernetes isn't as hard as it used to be. Going and clicking a couple of buttons on GKE and deploying something into Kubernetes with GKE Autopilot running it's not as easy as Heroku, but it's not wildly far off. And it does substantially future-proof you.

So, when is it too early? I'm not sure it's ever too early if you have an intention of scaling if you're planning on running some kind of legacy workload, like, things that are going to be stateful. Or maybe WordPress, for example, you don't probably need to deploy your WordPress blog onto Kubernetes. You can do that in your cPanel on Bluehost. I don't actually know if Bluehost even exists anymore, but I assume it's still a thing. I don't know, what would you say, Joe?

JOE: I agree with that. I think it's a hard first pill to swallow. But I think the reality is that it's very easy to underestimate the infrastructure needs of even an early product. Like, it doesn't really matter what you're building. You're still going to have things like secrets management. You're still going to have to worry about networking. They just don't go away. There's no way you have a product without them.

And so, rather than slowly solving all those problems from scratch on a platform that isn't designed for it, I think it's easier to just bite the bullet and use one of the managed solutions, especially, as you said, I think it's getting easier and easier. The activation energy from going from credit card to Kubernetes cluster is just getting lower.

KENDALL: And so, the role of the CTO is just getting easier and easier because they can just adopt the one technology, and it's obviously Kubernetes. And it's obviously Rust, right? [laughter] Yeah, no, I'm with you. And I think if you find somebody who knows Kubernetes inside and out, it's really not going to take them long to get started.

VICTORIA: Yeah, once again, change management is the biggest challenge for any new innovation coming into adoption. So, I'm curious to talk more about the influence that you need and how you influence others to come around to these types of ideas, like, in the executive suite and with the leadership of a company, especially on these types of topics, which can feel maybe a little abstract for people.

KENDALL: How you influence them specifically to use Kubernetes, or just how you talk with them about technology adoption in general? Or what are you asking?

VICTORIA: Yeah, like, how do I get people to not just turn their ears off when I say the word Kubernetes? [laughs]

KENDALL: Yeah, I mean, I I think that's where it's the technologist's job and the role of the CTO to translate these things into business speak. And that's why I'm using words like future-proofing your infrastructure is because there are companies that...I know one company that made a conscious decision that they were going to try to re-platform every single year, and that is not a good idea or sustainable for the vast majority [laughs] of companies. In fact, I can't think of a single situation where that makes sense.

But if you can say to the CFO, "Hey, it's going to cost us a little bit more right now. It's going to save us substantially in the long term because this is the thing that's winning. And if we go standardize on Heroku right now, every company does eventually have to migrate off of Heroku. They either go out of business, or they get too big for it." That's the kind of thing that needs to be communicated in order to get people to adopt it.

They don't care what the word is. They don't care if you're saying Kubernetes; you know, most CFOs understand it about as well as my mom does. My mom tries to bring it up in conversation because she's heard me use it. And she thinks it makes her sound smart, which maybe it does in the right climate.

VICTORIA: My partner does the same thing. He says DevOps and Kubernetes all the time. I'm like; you don't know what you're talking about. [laughter]

JOE: Those words do not come up in my house.

KENDALL: One of my kids asked me to explain Kubernetes. And I do a whole talk, particularly at organizations where understanding Kubernetes is essential to the salespeople's role. And I give a whole talk about the background of how we got here from deploying on some servers in our back room. And, you know, what's different about the cloud, what containerization did, et cetera. And I have this long explanation. And I remember taking a deep breath and saying to my kids, "Do you really want to hear this?" And I had one son say, "Yes, absolutely." And my wife and three of the other kids all stood up and said, "No way," and left the room.

So, when somebody asks me, "What do you do?" Actually, one of the key relationships I built with some of the early people at GCP when we were partnering closely with them was a person that I met, and I asked, "What do you do for a living?" And he said, "I can tell you, but it's not going to mean anything to you." And I was like, "That's what I say to people." And it turned out he was in charge of, you know, Kubernetes partnerships for Google. I can explain to you what it means and why it's important. But you're not going to be happy that I spent that time explaining it to you.

VICTORIA: [laughs] That sounds awesome, though. It sounds like you built a server rack just to demo to your children what it was.

KENDALL: No, no. I just talked back through the history of...that company that I mentioned that built Twitter about five years too early; we had a, you know, we had a server rack in the...literally physically in our closet that was serving up our product at the time.

VICTORIA: Probably the best demo I ever saw was at Google headquarters in Herndon, and someone had built...They had 3D-printed a little mini server rack that they had put Raspberry Pis onto, and then they had Kubernetes deployed on it. And they did an automatic failover of a node to just demo how it works and had little lights that went with it. It was pretty fun. So maybe you should get one for yourself. [laughter] It's a fun project.

KENDALL: They remember the things that it enables. They don't remember what it does. And so, when I say so, and so is a client that's using this technology, then they get real excited because they're like, "My dad makes that work." And I'm like, well, okay, that's kind of a stretch, but you get the idea.

VICTORIA: Yeah, you got to lean into that kind of reputation in your house.

KENDALL: That's right.

VICTORIA: And you're like, yes, that's correct.

KENDALL: That's right. [laughs]

VICTORIA: I do make Kubernetes. I make all the clouds work, yeah.

KENDALL: Actually, my most common explanation is Kubernetes is the plumbing of the internet. Unless you're a plumber, you don't care about the pipes. You just want your shit to flush when you use the toilet. You want the things to load when you click your buttons. You don't actually care what's going on behind the scenes, but this is what's orchestrating it increasingly across the internet.

VICTORIA: So far, we've called Kubernetes WordPress or the toilet. [laughs]

KENDALL: The plumbing.


VICTORIA: You are really good at selling it.


KENDALL: Hey, if you want to build a nice, clean city, you need good plumbing. You might not care what the pipes are made of, but you need good plumbing. [laughs]

VICTORIA: Works for me. On that note -- [laughs]

KENDALL: Yeah. Right? Right?

VICTORIA: That's [inaudible 36:41] on a high note. Is there anything else that you'd like to promote?

KENDALL: With regards to CTO Lunches, we have a free listserv. There are local lunches. If there isn't a local lunch where you are, it's very lightweight to start up a chapter. We often have folks who are willing to sponsor that first lunch to get you going. We do have a paid tier of CTO Lunches. If you want a small back room Slack channel of people to discuss, I think it's $99 a month.

Yeah, if you're a CTO and/or a senior engineering leader and you want a community of people to process with, be it our free tier or our paid tier, we've got something for you. We're trying to invest in this to build community around it. And it's something we enjoy doing more than almost anything. Come take part.

VICTORIA: You can subscribe to the show and find notes along with a complete transcript for this episode at If you have questions or comments, email us at And you can find me on Twitter @victori_ousg.

This podcast is brought to you by thoughtbot and produced and edited by Mandy Moore. Thanks for listening. See you next time.

ANNOUNCER: This podcast is brought to you by thoughtbot, your expert strategy, design, development, and product management partner. We bring digital products from idea to success and teach you how because we care. Learn more at

Support Giant Robots Smashing Into Other Giant Robots