What fascinates you about K8s?

What fascinates me about K8s? That this project is so big and open source. I like that the code is open and you can look into it. You can see each component with your own eyes, not just in theory. I really like the documentation: extensive explains everything. Kubernetes changes very quickly and you have to stay up to date, but it has super mechanisms with versioning. You can move forward without losing compatibility. I really like the range of possibilities that K8s gives. We can practically automate everything. And at the same time be sure that HA will be provided. This kube-apiserver (and more key components, ed.) is replicated on control plane nodes. Transparency of operators and resource management. This is unified by APIs and objects that are held in etcd. Anyone who is interested in creating a solution based on K8s can create their controllers/operators that allow flexible state management of such a cluster.

What fascinates me is that it (K8s) gives you tools that allow you to interact with the components of the physical nodes and the physical infrastructure. It’s this flexibility that allows you to create environment-specific solutions. I’m thinking of CSI, CNI, Cloud…. just can’t remember what it was called….interface for integration with cloud solutions.

Well, and the whole idea of this technology being based on containerization. And the idea, to create solutions that are stateless and replicated, that are also fault-tolerant. (Thanks to which) we are able to create upgrade canary deployments using K8s mechanisms, which ensure smoothness in the operation of applications and the lowest possible downtime.

Are you fascinated by K8s?

Of course I am.

Why?

The answer is one. Because of its complexity. I’m fascinated by the complexity of this whole environment, but also, how it’s being developed all the time. I haven’t had much to do with it yet, but as I’m getting to know it, it’s interesting to see how many management tools there are directly in K8s. How well provisioning is solved, how super the API is used and the fact that we can gather information with it and control what’s going on. It’s very fascinating how it all works together. Also how the resources are assigned, how if something is missing, how they work together so that everything works and doesn’t crash. All the orchestration makes that it all plays together….

The management tools are interesting, and that you can create your own.

Provisioning is interesting, how much you can know about what’s going on, how to control it, what’s happening to it (the application), how there are some corner cases. There is also good feedback, what is worth controlling and what is worth fixing (improving).

This is not a tool that can be learned in a moment. It requires a lot of knowledge, gathered in practice and in theory, to fully understand it. And something that is difficult to understand is certainly interesting 😉

Are you fascinated by Kuberentes?

I am fascinated by its capabilities.

Why?

(…long silence…) I like the change in approach to running applications relative to previous patterns…. and that the limitations it imposes allow you to create interchangeable modules. You can write a CSI driver and any application that supports it will work, the same is true for networking, service mesh, it doesn’t have too many components by itself, but adding external components is simple-easy. Even these components can be interchanged which is important, e.g. microK8s uses dqlite instead of etcd. The architecture is very nicely done, it allows for virtually any expansion of capabilities. That is, we don’t just have containers themselves, but also, for example, VM management, K8s operators and objects. Well, and I think it’s very well done, because it was done by Google people, after they did their own soft yet without containers, I think Borg, and the architecture is done in such a way that those things I mentioned before allow extensibility, well, and it’s open source. And it’s well thought out.

And how does Tanzu fit into this?

Tanzu expands K8s from all possible angles. And everything that can be done (in K8s ed.) Tanzu does. Going from the bottom up: storage (CSI driver to consume vsphere disks), networking, the whole Antrea integrated with NSX, which allows for network observables, even between K8s’ clusters, well and loadbalancers, you have management, sort of a level up, It’s awesome, because even the cluster supervisor is also K8s and a thing that creates clusters, it’s an operator. So the supervisor cluster itself is something of a K8s of K8s. Easy to manage, upgrade, destroy, it’s automated. To manually stage it to make it comparable is very hard, because it’s technically complicated. You have to know a lot of components and how to connect them. And it stands on the shoulders of giants, namely stable vSphere.

And how would you relate it to Sarkan?

We have the storage done. We don’t have all these components. We don’t have supervisor cluster, but when we will have it, it will be compatible not only with vSphere. But our solution only requires disks and Ubuntu, and as long as these are maintained, it all works out of the box. To have Tanzu then you must have compatible servers and storage. So for now, if you want to have it mainly for storage, then Sarkan will be simpler and cheaper.

And what is so wow?

MP that we write to Flopsar 😀…but with K8s…. this compatibility, as you have these components, the application will work everywhere, on any distribution, as long as it is not somehow messed up, in the sense of specifically adapted.

Are you fascinated by Kuberentes?

Of course I am 🙂

And what fascinates you about it?

…starting with containerization and microservices and moving on to K8s…because sort of the whole topic is fascinating. And K8s gives a lot of support for management. It can be divided into different tiers, which fascinates me. Parts: operator, developer, ecosystem, community. From development to just management, scalability, this reliability in accessing systems, from operator’s point of view: monitoring, optimization even in terms of cost, just how we have scalability, it’s hard to estimate, and with K8s we can optimize it. In terms of scalability, as we have an application that is unevenly loaded, K8s allows, where there is an increased load to handle it…this is a very cool feature of K8s…

I don’t know what else…. it also certainly simplifies application deployment. And here with the support of microservices, we are able to divide the team into smaller teams and we can develop them in parallel….automating various activities has a big plus. And it’s cool that we manage (resources ed.) in a declarative way. We define the requirements of the application and K8s provides us with that….don’t know I think I’ve exhausted…

But what is mega awesome, so much that you would say: “how brilliantly figured out!”?

Hmmmm… … … well, I guess, it will be so general …. But the management of these applications. And operators and security mechanisms.

Are you fascinated by Kuberentes?

Yes.

Why?

Um… Because it’s a cool tool… It makes it possible to… hmmm… it sort of… just works… Provides a lot of cool stuff for both… hmmm… some hobbyist application deployment, as well as for some large enterprise clusters. It lets you not worry about things like loadbalancing and scalability. It takes care of the big stuff for the user. Admittedly, the entry threshold is quite high and you have to learn a lot, but once you grasp the basics, you can easily do deployments. It takes care of rollout and rollbacks, if something doesn’t go right in deployment, K8s undoes it. Updating applications is much more seamless. All in all, too, technology has a way of liking to break down, services and hardware break down. And the K8s’ ability to selfheal allows it to monitor itself, check what’s going on in the app and possibly tries to fix any problems itself. We don’t have to do anything and K8s does it for us…In short, it’s a cool tool simply put.

But is it any awesome?

Well it is awesome…. what to say…. 😀… the thing that blows my mind is that this solution works brilliantly given how complex it is 🙂 …. or the ability to extend its functionality, that we can literally create custom K8s resources and act on them as if they were the default ones…. great thing…. creating native K8s applications that run and communicate with the K8s cluster…. that’s an awesome thing 🙂…

Are you fascinated by Kuberentes?

All in all, I was somehow not very familiar with K8s. Although during my first assignment on Sarkan, the guys let me get acquainted with it and I even told them that I was shocked at how it works. And not having seen it before, I was able to say that it works cool.

Why does it fascinate you?

Hm… …I think it’s cool that we don’t have to personally watch over the containers, the whole applications built from those containers. He does it for us. So that there is no interruption when something breaks, he will stage it again. What else? It’s cool that we can specify the processing power and RAM memory for these containers, and I guess that’s it. It’s nice that such a tool was created and you don’t have to do all this manually.

But what is so really awesome?

Well then on voffice (virtual all-day collaborative work using Teams, because we work remotely – editor’s note)… I said I was shocked, because when I removed the metadata server, and he changed the leader himself, then put up a new server himself…. I didn’t know it worked like that yet, maybe that’s why…. for now, I think so much of what was so awesome…. I guess…