Simplifying Kubernetes, hands-on learning and on-premise

Bart: Who are you?

Fahd: Hi, I'm Fahad. I'm a partner engineer at Google Cloud for the last three and a half years. I work out of Amsterdam.

Bart: What are three emerging Kubernetes tools that you are keeping an eye on?

Fahd: If you look at tools and this floor behind us, there are so many that are catching my attention. The most important thing I'm looking at nowadays is how we can do things more effectively. Sustainability is one thing. I don't know if you've spoken to the guys at Resync, but they are focusing a lot on making everything more green. The second thing is heavy into observability. I'm always curious about what's getting released in observability. Thirdly, getting Kubernetes to the edge. For example, Google is doing this with GCDE, Google Cloud Distributed Edge, bringing the Kubernetes experience to edge locations. And that's a new world of pain. I come out of the hardware space, and it brings me back to the hardware space. So I think that's my focus this year.

Bart: One of our guests, Matt, argues that GKE is one of the best Kubernetes managed services because a lot of complexity is taken away from you. For example, he mentioned Autopilot as a way to provision the correct node size and optimize your utilization automatically. What's your opinion on Kubernetes managed services? Are they all created equal? Should they offer a more friendly Kubernetes wrapper like GKE, or only the basics without any opinion on how you want your Kubernetes cluster to be?

Fahd: The way I see it is we want to make things easier for customers. And I'm very biased here. So GKE is the best. That's what I've been shouting for years, even before I worked at Google. But in general, when you look at the trajectory of GKE, making things easier with autopilot, and now even looking at GKE enterprise, where we can also absorb other Kubernetes clusters with the GKE wrapper and all the nice things that come with GKE, it's one way to make things nicer, easier, and more homogeneous. I don't think we need to stamp it in a way that when you use GKE, you can only use GKE the GKE way. You can still use all kinds of distributions on Google Cloud depending on what you need. I think the main focus is on the requirements and the needs. I wouldn't fiddle around too much with the details that typically don't matter. Abstracting things away makes things easier for developers. Let's focus on that. And GKE does that very well.

Bart: Another guest, Matthias, suggested the best way to learn Kubernetes is by doing and getting your hands dirty. He then built his own bare metal Kubernetes cluster in his spare time. What's your strategy for learning new Kubernetes tools and features?

Fahd: Learning by doing. Learning by doing, Kelsey Hightower's The Hard Way is a very good one. But you need to ask yourself one important question. Do you want to understand Kubernetes as a whole? Do you want to understand it as a user? If you look at the certification tracks, for example, you could get CKA certified. That means you know the whole stack. But in your day-to-day basis, are you going to use the whole stack or not? I think it's easier if you use something like DigitalOcean or some other distribution, just to get the hang of it and start deploying your first workload. There's the... The Hipster Shop or the microservices deployment you can do yourself really easily and test it and get it up and running. In the past, I interviewed a lot of people and they said, I've never got the opportunity to touch Kubernetes. And I'm like, why do you need the opportunity? You have Minikube on your notebook. It's that simple. So learn by doing. YouTube is a great source for a lot of material. I think that's where you should start if you want to get into this space.

Bart: Matthias also believes that on-premise deployments require proper education and attention, especially regarding managing on-prem architecture versus cloud architecture. After spending a few months building an on-prem Kubernetes cluster, he shared this advice. What's your experience with bare-metal Kubernetes clusters, and how does that differ from using Kubernetes in the cloud? What would you have liked to know before starting Kubernetes on bare metal?

Fahd: In the past, I've also been into consultancy. So I've experienced Kubernetes in the field on bare metal as well. It is hard, very hard, especially if you look at how you would consume it, what kind of resources you need, and mainly what kind of knowledge you need. My take on this is people experience cloud Kubernetes a lot and then think to implement it on-premises. The first thing you need to do is ask, "Do we really need this?" I think that's the most important question. The second thing you need to look at is what are the requirements and what can we do? If you have a hardware estate and you need to run Kubernetes, make your life easier and use something like OpenShift, Rancher, or in our case, Anthos (now called GDCV), which helps you run container workloads on-premises without the hassle. If you have to install machines with an operating system and on top of that install Kubernetes and maintain all of that, it requires a lot of effort, a lot of people, and a lot of knowledge. Mainly, things can go wrong. As a famous German bank once said to me, "Who do I call when Kubernetes is broken?" Well, you cannot call anyone. Hence, you need to have something that has at least some kind of support to run the business. It's not a hobby. It's a production system. So weigh your choices and make the right decisions as you go. Kubernetes is turning 10 years old this year.

Bart: What do you expect in the next 10 years to come? Kubernetes is turning 10 years old this year.

Fahd: I think in the next 10 years, we're going to talk more about Kubernetes. Ten years ago, I expected that nothing was going to happen with this Kubernetes thing, but I don't know what's going to happen. I hope this thing will continue to happen and make me happy.