What's new with Kubernetes 1.22?

Hello, cloud gurus! I'm Nigel Poulton, and this is your monthly update on what’s new with Kubernetes. In this post, we'll take a closer look at Kubernetes 1.22: Reaching New Peaks. We'll dive into new features like etcd 3.5 adoption and the deprecated features to note before making the switch. Plus, we'll bring you five other important K8s-related announcements you won't want to miss.

Want to know more? Read on!

This is the second release of 2021, and the very first from the new elongated release cycle. In July 2021, the release team merged a Kubernetes enhancement proposal, flipping the number of Kubernetes releases per year from four down to three. And this is the first under this new longer cycle.

Well, the release version is 1.22, and its code-named "reaching new peaks."

On the numbers front, it's packing a total of 53 enhancements.

• 13 of these are features graduating to stable, so considered production ready. 
• 24 have moved to beta.
• And we welcome 16 new features as alpha.

But just as important as the new stuff, is the stuff that's going away. We've got three features being deprecated, and 10 previously deprecated features actually being removed.

On the new features front, I think the biggest one for me is the switch to etcd 3.5.

Earlier this year, the release of etcd 3.5 meant major improvements for Kubernetes — especially larger, busier Kubernetes clusters where etcd was often a bottleneck on performance.

At the time, we said that although etcd 3.5 was here, we had to wait for the Kubernetes project to adopt it. Well, they didn't make us wait long, as I'm really pleased that Kubernetes 1.22 will ship with etcd 3.5. And for sure, it's mainly behind-the-scenes improvements. But they span everything from security and logging to much-needed performance improvements.

Watch: Automating Kubernetes Security
In this free, on-demand webinar, learn how to beef up your K8s security with Pod Security Policies. We'll show you how they work and what it looks like to implement them in a real Kubernetes cluster.

Also on the GA or stable front, external credentials plugins are finally stable after being in beta for 10 versions. Windows support for CSI plugins also went GA.

On the alpha front, I think swap memory support and default profiles for seccomp will be important in the future.

On the swap memory front, of course it's only in alpha, but work is in progress to support cluster nodes with swap memory enabled.

And then on the seccomp front: seccomp is obviously a Linux technology for improving security, and work is in progress for creating default seccomp profiles that will give us better security (out of the box) than the current approach of leaving the front door wide open.  Again, it's only alpha for now. But it is definitely a step in the right direction.

Last but not least (and with potentially serious impacts if you don't plan for it) 10 deprecated features are finally removed and will obviously no longer work.

The full list is shown on the Kubernetes website. But for me, Ingress and Custom Resource Definitions are big ones, as these are pretty widely used.  So, if you're using any of these, you need to plan, plan, and plan before you flip the switch to Kubernetes 1.22.

In other significant news that you don't want to miss... Linkerd, the popular and SIMPLE service mesh, graduated the CNCF. I liked one of their taglines: "A victory for simplicity in a space notorious for complexity." As service meshes become more and more an integral component of production Kubernetes deployments, Linkerd is most definitely worth a look. Especially if you’re looking for simplicity.

Staying in the Service Mesh space, Istio announced version 1.11, with features and improvements including the CNI plugin going to beta to help solve some security requirements around init containers. And beta support of the external control plane, where the istio control plane can be hosted in an external cluster.

In the public cloud space, Google announced GKE loadbalancer uptime checks. And I love stuff like this as it's all vital for running Kubernetes in production: For example, monitoring uptime of services is mega important in production environments, so Google adding dedicated uptime checks for loadbalancers in GKE environments is a great step in a positive direction.

Over to the networking space, and we've said in previous episodes that eBPF is really starting to rock the networking world, and some folks are playing with potentially game-changing integrations with Kubernetes.

We're talking stuff like a programmable network with intelligence to understand Kubernetes and optimize for it.

Anyway, a bunch of major industry players just announced the eBPF foundation to drive this kind of stuff forward. The foundation will live within the Linux Foundation. And it will drive development and adoption of eBPF but also organize and host events.

That’s it for this month’s edition of Kubernetes This Month. Stay safe, and I’ll see you all again next month — same Kube time, same Kube place.

• Which Kubernetes Certification Path Should I Take?
• Which Kubernetes distribution is right for you?
• AKS vs EKS vs GKE: Managed Kubernetes services compared
• Watch: Kubernetes + Azure, the HashiCorp way

Want to keep up with all things Kubernetes? You can follow Nigel on Twitter or keep up with him here. Subscribe to A Cloud Guru on YouTube for regular news updates, analysis, and assorted awesomeness. You can also like ACG on Facebook, follow ACG on Twitter, or join the conversation on Discord!

Learn faster. Move faster. Get started with ACG and transform your career with courses and real hands-on labs in AWS, Microsoft Azure, Google Cloud, and beyond.