It took me a while to write this update. A lot of is going on in my private life and it is hard to find some free time to write. But, I enjoy writing and sharing. There are some great and important things I want to share in this monthly update.
A few weeks ago I created Cloud Native meetup group in Sarajevo. In Sarajevo we don't have any DevOps or Cloud Native related meetups. But, I know a lot of people who are doing the similar job like I do, and I know almost all local IT companies who are ready to support this kind of meetings. Even some universities are happy to have some workshops organized for their students. Also, the companies are struggling to find people interested in DevOps in general. So, I decided to start this venture and to get more people interested. CNCF was happy to support us with the swag:
We are preparing our first meetup and I hope that we will become the first and official Cloud Native meetup in this area.
Also, I covered the KubeCon and CloudNativeCon in Copenhagen by providing 10 must watch presentations. It was really hard to pick top ten, but there is my list.
While working on one project I was looking for a way to reschedule pods. Default scheduler won't do this, but I found a promising piece of software. Kubernetes descheduler checks for pods and evicts them based on defined policies. I wrote a blog post about it, meet a Kubernetes descheduler.
And I decided to write a blog post about important part of any Kubernetes cluster, autoscaling. This post covers AWS, but other cloud providers are no different. The tool that will help you with autoscaling is Cluster Autoscaler, which will watch Kubernetes cluster and on insufficient resources scale up or scale down when you have underutilized nodes. In short, it will save you some operational cost.
Helm is a CNCF Project
Some users are not happy with the Helm. But, I think that Helm is pretty good and it could be even better in the future. Also, you might notice that all my blog posts are using Helm to install something on Kubernetes. The reason is simple. It is easy and it works just fine. Did I mention the Kubeapps? Yeah, there is a Helm chart for most popular open source software and creating custom charts to pack your applications is easy.
I started to explore Kubernetes operators last year while deciding how to deploy Elasticsearch cluster. I like the idea that we can easily embed a human knowledge into an operator. Operator Framework for building Kubernetes applications is finally here! The Operator Framework includes SDK, Lifecycle Management, and Metering. This came out as a surprise, but I was really happy to hear it. Stateless applications can leverage the Lifecycle Management features without writing any code by using generic Operator, for example Helm operator, but with stateful applications, Operator shines. If you are new to the operators check my post about managing Elasticsearch cluster with an operator.
Kubernetes the Hard Way - Updated
Kubernetes setting up tutorial, Kubernetes the hard way by Kelsey Hightower now includes the new version 1.10.2 and gVisor support. This is probably the best tutorial out there, great for learning or improving your Kubernetes skills.
Istio v0.8.0 Released
You shouldn't ignore what is happening in the service mesh world. Actually, my second most popular post is about Kubernetes Service Mesh. New Istio LTS release is available and includes new or updated features around networking, security (introducing Citadel), telemetry and installation. The Istio has a rich set of features, but you don't need to install all if not needed. With customized Helm installation you can choose only what you need.