CompanyOctober 15, 2021

3 Hot Takes on KubeCon NA: Community, Edge Computing, and Multi-Cluster

Patrick McFadin
Patrick McFadinVP, Developer Relations & Cassandra Committer
3 Hot Takes on KubeCon NA: Community, Edge Computing, and Multi-Cluster

The dust is still settling on this week’s KubeCon North America 2021, but it’s never too early to share some quick impressions about this gathering of the top minds in leading open source and cloud-native communities (we’ll follow up soon with in-depth coverage).

The DoK community thrives

As an official pre-event before KubeCon started, the Data on Kubernetes day kicked things off. If you haven't seen what's happening in the Data on Kubernetes community, it's time for you to check it out. The energy level is off the charts. The talks covered everything from running databases to configuring storage and generally just how to successfully run your entire stack in Kubernetes. If you missed any of the presentations, the playlist will be on YouTube soon. Part of our upcoming follow-up will be to give you some of the highlights. 

I think it is safe to categorize the DoK community into two parts right now. First are people that are already deploying databases, streaming, and analytics on Kubernetes®. They are looking for a community to work with to help better their skills and share best practices. A new survey shows that a lot of organizations are already doing stateful workloads in Kubernetes. The other part of the community is the newcomers that bring a lot of curiosity and want to get it right the first time. We're approaching the tipping point where enough people who share how they’re doing data on Kubernetes can accelerate people just starting. As new projects are implemented there will be amazing positive feedback created. It’s the concept of standing on the shoulders of giants making user communities great.

The buzz: Edge computing

Typically at each conference, I find an underlying theme that emerges from presentations and buzz in the hallways. KubeCon is no exception, and in this case, there seemed to be two distinct themes. The first was something that caught me by surprise: edge computing and Kubernetes. 

As workloads spread around the globe, large data centers and regions for cloud providers don’t get close enough to where the infrastructure needs to be. The term hyperlocal also pertains to infrastructure, and using the same control plane is appealing. You may think this is simply for IoT, but full-blown applications are now running on the edge. Projects like KubeEdge and OpenShift are making it easier to implement. Edge-friendly Kubernetes distributions, like k3s and k0s, are now becoming mainstream. These are lightweight distributions specifically built for limited-resource hardware. Think of deploying Kubernetes on a Raspberry Pi (yes, you can). More validation of how Kubernetes is taking over how we run infrastructure. It's no longer simply on-prem and cloud but now we have edge in the mix. I’m sure we’ll see Kubernetes running on a space station soon. Can’t wait for that talk. 

Multi-cluster is a thing

The other theme was multi-cluster Kubernetes. We‘re finally able to admit that we run applications in more than one data center. The buzz on this topic wasn’t as surprising. Having this ability is a big need in the Kubernetes community. There are now significant efforts to bring multi-cluster Kubernetes mainstream. The concepts of hybrid-cloud and multi-cloud have been used regularly. Still, one of the unexpected drivers of multi-cluster is how well it allows organizations to be cloud-agnostic, and not be tied to one cloud provider when deploying an application. A deployment in Kubernetes creates a virtual data center that rents the compute, network, and storage of any cloud provider. Even if you use only one cloud provider, you might want to run your application in multiple regions. 

Using Kubernetes outside of one cluster has been almost impossible without a lot of work. The Special Interest Group (SIG) work that's being done in the Kubernetes project is making this a primary focus. Projects such as KubeFed and SIG multi-cluster are ones to watch. Before KubeCon, the K8ssandra project announced an operator that allowed multi-datacenter Apache Cassandra with a single control plane. I think this is an idea whose time has come. 

Those are some quick thoughts about KubeCon North America 2021. Watch out in the next week or two, and we will offer more in-depth discussions about the talks you may have missed. There were a lot of presentations, but we’re here to help you curate the Data on Kubernetes journey. Hopefully, we will see you in the booth next year!

Want to catch up on some of the articles we published just before KubeCon? Here are a few for you to check out:



One-stop Data API for Production GenAI

Astra DB gives JavaScript developers a complete data API and out-of-the-box integrations that make it easier to build production RAG apps with high relevancy and low latency.