#019 – Kubernetes for Humans Podcast with Donnie Berkholz (Platify Insights)

Itiel Shwartz: Hello everyone, and welcome to another episode of the Kubernetes for Humans podcast. My name is Itiel Shwartz, and today with me on the show, we have Donnie. Donnie, can you please introduce yourself?

Donnie Berkholz: Hey, I’m Donnie Berkholz. I’m currently the founder and chief analyst of an independent analyst firm called Platify Insights. We focus on technology platforms in general, but specifically areas like cloud-native computing, generative AI, and platform engineering.

Itiel Shwartz: That sounds super relevant, and I think it will be interesting to hear your predictions at the end since you do this for a living. But let’s start with your background. How did you get into the tech space, into containers and Kubernetes, and then we’ll take it from there.

Donnie Berkholz: Sounds good. I have a non-traditional background. I didn’t get a CS degree and follow the typical path into software engineering. I went into science and thought I was going to be a scientist doing biomedical research. I got my bachelor’s degree in Biochemistry and my Ph.D. in Biochemistry and Biophysics. I was doing something called x-ray crystallography.

Itiel Shwartz: What went wrong?

Donnie Berkholz: I think it’s more like what went right. As I was doing science, two things happened. First, I had to learn how to use Linux to do simulations of how molecules might move, test different shapes they would take, and how a drug might dock into a protein to create its effects in the body. I had to get pretty technical, familiar with the command line, and with working with Linux systems. That turned into a passion because I found out that I loved it. I started using it and joining communities. The first community I joined was FreeBSD, and later I joined Gentoo Linux. I started with Red Hat back before Red Hat Enterprise Linux.

I joined the Gentoo community and found a passion for having everything in my control. It got to the point where I started contributing, became an open-source developer, and spent about 13 years contributing to the Gentoo Linux distribution, which is still around today. They just announced the other day that they’ve now got binary packages, so you don’t have to compile it all from source yourself anymore. It’s now comparable to Arch in some ways.

During the day, my science work became more computational. First, it was working with Linux, then writing code to manage libraries and simulations. Eventually, I felt like I was almost a software engineer with the job title of a scientist. I decided that I enjoyed the means more than the ends. I was having more fun doing the technology than the science. I looked at my mentors in science, and they spent a lot of their time trying to get grants to do the work, which wasn’t fun. Doing the work is fun; getting the money is not. Whether you’re in science trying to get grants or a CEO of a startup trying to get your next round of funding, that’s not the fun part. The fun part is doing the thing, creating the impact.

Long story short, I moved from that background to spend some time as an industry analyst at companies like RedMonk and 451 Research, focusing on software development, DevOps, and platforms. I spent some time as an end-user, leading a DevOps transformation at a large travel tech company, Carlson Wagonlit Travel (CWT), which had an office in Tel Aviv. When COVID hit, and travel got boring, I moved on to be a VP of Products at Docker, then at Percona, an open-source databases company with Kubernetes operators. Most recently, I decided to start my own company.

Itiel Shwartz: Why start your own company? Maybe share more about how you got into Kubernetes, the company’s stage, and your thoughts back then. Then we can go into the future.

Donnie Berkholz: I started tracking containers early on at RedMonk and 451 Research, around 2012 to 2015, when Docker was heating up. By 2016-2017, Kubernetes started picking up steam as a potential alternative for running things at scale. As an industry analyst, I was paying close attention, talking to people about why they used containers, the value they were getting, and how to quantify that value so developers, engineers, and operators could convince their executives that it was worth migrating something over to Kubernetes.

Later, when I moved to the enterprise side at CWT, I experienced it firsthand. We had an internal “anarchy” where each product group was free to do whatever was necessary to achieve business outcomes. There was no top-down mandate. Instead, it was like empowered product teams across product organizations. Each team adopted different technologies based on their needs, knowledge, and willingness to dive deep. Some used Docker, others used Amazon ECS or ran their own Kubernetes clusters. It was a spectrum of technologies with varying levels of interest and expertise.

Itiel Shwartz: That makes total sense. You’ve been in both individual contributor and managerial roles. What do you think has been the biggest change in the landscape over the last five or six years, particularly with the cloud and orchestrator wars? Kubernetes seems to be the last one standing, but how did it feel back then, and why did Kubernetes win in the end?

Donnie Berkholz: If you rewind six or seven years ago, it was very unclear what would happen. We had Mesosphere, Kubernetes, Docker Swarm, and Docker had two different versions of Swarm, which got confusing for people. At the time, it was unclear what would win out. This fragmentation is common when there’s a lot of innovation in a space. People were choosing all these options because they seemed the same, or there wasn’t effective differentiation. Over time, this innovation phase led to consolidation, where it became clear what the obvious solution was.

During that time, vendors jumped in, trying to sell their solutions. DevOps was the dominant term for the 2010s, so everyone was DevOps-washing their products. As things evolved, the roles, technologies, and communities had to evolve too. At CWT, with multiple container solutions and teams choosing the right tool for their jobs, it wasn’t always clear what the right choice was. Sometimes, the right tool was the one with the lowest barrier to entry, which might be the tool the team was most familiar with. If you’re an AWS shop, it might be easier to pick another AWS solution than to try something entirely different.

Itiel Shwartz: Choosing the right distribution or architecture for both infrastructure and teams is one of the hardest decisions. We see a lot of companies migrating to Kubernetes, but no one is sure they’re doing it right. What would you suggest to a company at that stage? Should they choose Kubernetes, ECS, or something else? What are the key parameters to consider?

Donnie Berkholz: The first question is always whether Kubernetes is the right answer for this application. It’s crucial to avoid investing time and effort into something that isn’t well-suited to your workload. If you have a static workload with minimal variability in traffic and it’s fine to take it down occasionally for upgrades, Kubernetes might not be necessary. But if you need minimal downtime even during configuration changes and upgrades, Kubernetes could be a good choice.

You also want to consider if you can maintain the in-house skill set to run Kubernetes effectively at scale. If you can’t, you might want to use a managed service provider or stick with simpler solutions like a hosted container service or running things yourself in EC2. There’s also the question of whether Kubernetes will bring the benefits you need without adding unnecessary complexity. If you’re running a monolithic application, Kubernetes may not provide much value.

Itiel Shwartz: That makes complete sense. Let’s talk about the future. What are the biggest trends you’re seeing, or what’s your boldest prediction?

Donnie Berkholz: The obvious trend is generative AI and its intersection with Kubernetes. We’re seeing people wanting to train their large language models (LLMs) inside Kubernetes. Kubernetes itself will need to adapt to handle these workloads effectively. Generative AI workloads share similarities with high-performance computing, which could benefit Kubernetes adoption in scientific communities, academic research, and areas like defense.

In addition, there’s continued growth in platform engineering. But like DevOps, it’s becoming whitewashed by some vendors, losing some of its meaning. To me, platform engineering means providing self-service to developers, thinking about the platform as a product, and applying product operations approaches to running the platform at scale. Many vendors offer technology that supports platform engineering, but it’s also about people and processes, not just technology.

Itiel Shwartz: Anything else you want to share with the world before we wrap up?

Donnie Berkholz: Thanks for listening. If you have any questions, feel free to reach out at platifyinsights.com.

Itiel Shwartz: We’ll put it in the show notes. Thanks a lot, Donnie, for being a great guest, and good luck!

Donnie Berkholz: Thanks, appreciate it. Have a great rest of your day.

[Music]

Donnie Berkholz, Ph.D., is the founder & chief analyst at Platify Insights, the only analyst firm for technology platforms. Before that, he was SVP Product Management at Percona, the unbiased open-source database company, where he led product management, design, and community.

Previously he served as VP Products & Advisor at Docker, where he built a developer-focused product strategy with a product-led growth model, with his contributions resulting in a 4x year-over-year increase in ARR. His extensive experience includes acting as Executive in Residence at Scale Venture Partners, leading the DevOps transformation at travel-tech leader CWT (fka Carlson Wagonlit Travel), advising upon development, DevOps and data science at industry-analyst firms 451 Research (acquired by S&P Global) and RedMonk, and leading more than 250 open-source contributors at Gentoo Linux. He also spent more than a decade in computational & experimental drug discovery and biochemistry, including a fellowship at Mayo Clinic.

Donnie holds a Ph.D. in biochemistry and biophysics from Oregon State University, where he specialized in computational structural biology, and dual B.S. and B.A. degrees in biochemistry and chemistry from the University of Richmond.

Itiel Shwartz is CTO and co-founder of Komodor, a company building the next-gen Kubernetes management platform for Engineers.

Worked at eBay, Forter, and Rookout as the first developer.

Backend & Infra developer turned ‘DevOps’, an avid public speaker who loves talking about infrastructure, Kubernetes, Python observability, and the evolution of R&D culture.  He is also the host of the Kubernetes for Humans Podcast. 

Please note: This transcript was generated using automatic transcription software. While we strive for accuracy, there may be slight discrepancies between the text and the audio. For the most precise understanding, we recommend listening to the podcast episode