The new Slimed Down version of Kubernetes - K3S

In this episode, we discuss the new Kubernetes distribution K3S. This super lightweight distribution is great for Edge, IOT, ARM, and more. Join us as we take a look a the new distribution.

In this episode, we discuss the new Kubernetes distribution K3S. This super lightweight distribution is great for Edge, IOT, ARM, and more. Join us as we take a look a the new distribution.

Episode Transcription


Today we're gonna talk about K3S, K3'S or K Three S however you wanna call it. It's so new we haven't even coined the term yet, so it's open for discussion. If you don't know what K3S, it's the new Kubernetes distribution by Rancher Apps. Rancher has done an amazing job and taken K8S or Kubernetes and subtracted five and made a K3S obviously. But their take on this distribution is not adding more features but taking away and Kubernetes is by default a massive beast. It's getting more and more complex to manage and what Rancher's doing is trying to slim it down and make the edge case or actually the use cases more specific. Kelsey Hightower actually chimed in on this on Twitter, and he said he really likes how Rancher is taking a new look at running low resource computing environments by removing features and not adding features.

When I talk about removing features, K3S is actually removing the legacy of non-default features, so there's a lot of stuff hanging out there that we just don't need right? So the removal of a lot of these things, alpha features, if you're an alpha tester that's great, but in production, we don't really need it or for IOT and Edge. Also, they're removing docker, so they've opted to use container D, which is lighter weight, we don't need all these commands in there.

When we remove all these components, we actually make the K3S binary only 40 megabytes. That's right, 40 megabytes. Now when K3S is running, it only requires 512 megabytes of RAM. Now think about that for a second, a full Kubernetes distribution that is running with 512 megabytes of RAM. This is amazing. Now you can already envision use cases for this for Edge, computing Edge, Cloud Edge, IOT, Raspberry Pi, so this is actually optimized for Raspberry Pi which is great. We're gonna start building our own Raspberry Pi cluster for K3S and start playing around with it and exploring some of the IOT and Edge cases with it, because it's so lightweight, it's so easy to use. I mean, you basically take two commands to get this going, you basically download the binary and then you go, K3S start. You just get up and running and you're going. It's that easy.

It's a new take on how we look at Kubernetes just by taking things away, specifying things exactly the use case, in this case, we want to optimize Kubernetes as much as possible and the next thing we want to really make this just a resource, we want to look at this from a totally different aspect. We don't want to look at it from all of the huge features and things like this, we want to specify that we just want a small Kubernetes cluster running that we can actually interact with and put a couple more containers on.

Now Rancher sees this as an opportunity for some of their customers. I guess they've been asked by their customers, mostly in the retail and Teleco spaces to come up with some lightweight options that they can run at the Edge. Retail operations, they say they want to ... These retail operations were asking, "Hey we would like to run a small Kubernetes cluster in every retail store we have, just to run some jobs and some Edge cases." They don't need a giant Kubernetes cluster running in every store, they just want something small to run a couple of processes.

The same thing with Teleco's, they want to let the Edge to do some of their Edge processing, so this is brilliant. Now taking all this into consideration, Rancher was able to certify and make this a fully compliant Kubernetes distribution. So this is actually on C and CF when you look at Kubernetes distributions, K3S are already there. I mean, it's unbelievable in this short amount of time that it was only released about a week and a half ago, two weeks ago, already has 4700 stars on GitHub and it's increasing in popularity.

From our aspect we see tons of use cases for our customer, 'cause we're focusing on IOT, we're looking at some of the use cases are on Edge, so as we go forward, I'll let you know how it goes because we're starting to use this and explore some of the possibilities here. But it excites us how small this is, how really it doesn't take a lot of resources, however I can still run it on a giant Amazon instance if I want. But the idea is, I want something small, I don't want this giant beast and it's really looking at it from like a docker aspect. I want to install docker and I want all the components and I want it to be lightweight and running.

It's a great take, I'm interested to see where the project goes. If you are using it or have any feedback, please let us know, 'cause we'd love to hear about what you're doing with K3S and we'll keep you posted on what we're gonna do.
That's all for today, go and check out K3S, k3s.io.

I'll put it in the show notes, if not, well have a good day and we'll see you next episode.
Brian Christner