By 
MicroK8s is a lightweight and easy to use Kubernetes distribution designed to run in resource-constrained environments such as IoT and edge devices. As Canonical is eyeing enterprise use-cases it’s making Microk8s more resilient by adding high availability capabilities to it.
Microk8s already has the clustering feature; with a single command, users can join multiple MicroK8s nodes in a cluster. With HA, as soon as users join three or more nodes, they get the Kubernetes control plane distributed across these nodes. If they join more nodes they get all the API services of Kubernetes available on all nodes and the control plane is still distributed on these nodes.
To make it work Canonical has swapped etcd, the Kubernetes data store, with Dqlite, which Alex Chalkias, Product Manager at Canonical, says is the most popular distributed SQLite database out there.
With HA, Microk8s can now power more industrial use cases, where users can have multiple sensors and IoT appliances running in a factory floor. “With HA and Dqlite, we ensure to have the best reliable solution for Kubernetes where the users don’t really have to deal with all the configuration and the complexity that Kubernetes brings,” he said. “It’s the evolution of MicroK8s.”
MicroK8s HA is part of the complete Canonical solution for edge machine provisioning and provides VMs, containers and storage on-demand.
[expander_maker]
Here is the automated transcript of the discussion
Alex K:
MicroK8s started two years ago as a sort of an easy Kubernetes that can help people get their hands dirty, familiarize themselves, kind of a developer’s companion. But we felt that we wanted to sort of have a shift towards more enterprise use cases, production use cases. And having a resilient Kubernetes is effectively what drives enterprise adoption. And so this is why we brought HA.
Swapnil:
Hi, this is your host Swapnil Bhartiya. Welcome to another episode of TFiR Insights. MicroK8s is a powerful lightweight Kubernetes distribution with a very small disk and memory footprint without any compromises that you would expect from a full-fledged Kubernetes distribution. And with the explosion of edge computing and IoT devices, I think MicroK8s is ideal to cater to such use cases.
Swapnil:
This month, Canonical announced high availability MicroK8s. To talk about what is high availability MicroK8s, and what are the use cases that are ideal for it, today, we have with us on our show, Alex Chalkias, Product Manager at Canonical. Alex, first of all, welcome to the show.
Alex K:
Thanks for having me once again.
Swapnil:
Please tell me about high availability MicroK8s. How is it different from regular MicroK8s?
Alex K:
MicroK8s started two years ago as sort of an easy Kubernetes that can help people get their hands dirty, familiarize themselves, kind of a developer’s companion. And we had seen a lot of traction or there’s some great positive feedback, people loving the experience, how easy it was, how small it is as well.
Alex K:
But we felt that we wanted to sort of have a shift towards more enterprise use cases, production use cases. And having a resilient Kubernetes is effectively what drives enterprise adoption. And so this is why we brought HA.
Alex K:
So in HA, what we do is that we’ve already had the clustering feature. So you could join… with a single command, you could join multiple MicroK8s nodes in a cluster. With HA, as soon as you join three or more nodes, you get the Kubernetes control plane distributed across these nodes. If you join more nodes, you get all the API services of Kubernetes available on all nodes. And the control plane is still distributed on three nodes.
Alex K:
How this works in MicroK8s is that we have effectively swapped etcd, the Kubernetes data store with Dqlite. So Dqlite is probably the most popular distributed SQLite database out there. And that’s what’s powering the HA feature.
Swapnil:
Is it the evolution of MicroK8s or it’s just one more variants of MicroK8s?
Alex K:
No, it’s definitely the same product. So the same distribution of Kubernetes MicroK8s, but it’s really an evolution of it. So as I said before, we started as a developer-friendly Kubernetes, easy to get on onboard. But now we’re also looking towards more production use cases like edge with the Micro Clouds as we recently introduced the term, which is bringing compute storage and networking closer to the end user, closer to the data source.
Alex K:
So we want to have MicroK8s part of that edge solution. And we want to ensure that people can rely on their Kubernetes with MicroK8s for all their production use cases. There’s also a really interesting AI/ML story with Kubeflow, which is enabling users to create their machine learning pipelines on top of MicroK8s. That’s also a big driver of production use for MicroK8s. So yeah, with HA, we effectively provide a zero-ops autonomous, highly available Kubernetes to address this.
Swapnil:
What would be the ideal use cases or what are the use cases you’re looking at with MicroK8s? Also, you also have Core which is focused on IoT devices. Ubuntu Core, which is also lightweight [inaudible 00:00:05:00]. So how does those two technology work? Do they compliment each other? Talk about the whole positioning for edge computing or IoT space.
Alex K:
Edge computing, I already briefly introduced. So we want to have MicroK8s as part of a complete solution that could provide… that could do edge machine provisioning and provide VMs on-demand and containers on-demand and storage on-demand. So MicroK8s will address the container part of that story.
Alex K:
Then for IoT, of course, we have a lot of applications with MicroK8s, starting from Raspberry Pi since we both support ARM and Intel architecture. So that’s really neat to come with the lightweight packaging that can allow you to run MicroK8s in sort of resource-constraint environments.
Alex K:
But now with HA, with the resilience that is added to MicroK8s, we can also have MicroK8s powering more industrial use case, where you would have multiple sensors and IoT appliances running in your factory, for example. And with HA and Dqlite, we ensure to have the best reliable solution for Kubernetes and adding to the simplicity where the users don’t really have to deal with all the configuration and the complexity that Kubernetes brings.
Swapnil:
If you can give some use case example, maybe.
Alex K:
We have some solutions on retail stores. So to have the super popular Chick-fil-A example where they were running Kubernetes on the back of their stores, we have similar customer use cases. There are other use cases like Telcos, where they want to embed MicroK8s to their sort of Telco mini edge stack, which is super important.
Alex K:
We love working with Telcos. They have all these high-performance requirements, usually, that drive really the products to its limits. So it’s a good live test to see that it’s really performant and it’s really robust. Yeah, I would say between Telco and retail, that’s probably now the bulk of our customer base. But we also have discussions with finance and medical and automotive and the Industry 4.0 companies. So yeah, I hope next time I could sort of give you some specifics on the customer stories.
Swapnil:
Alex, thank you so much for taking time out of your schedule and talk about, not only MicroK8s high availability, but also how it is going to serve the exploding edge computing and IoT space. And I would love to talk to you again, because I’m pretty sure that you will have a lot of things to talk about in future. So once again-
Alex K:
Absolutely.
Swapnil:
… thank you.
Alex K:
Again, thanks for having me.
[/expander_maker]
