Welcome to another monthly update. Last week I was on vacation in Brela, Croatia. Really beautiful place. So, I feel fresh again to start writing. I wrote only three articles in June and one of them was the last monthly update. Definitely not happy with that. I need to write more.
As you already know this blog is running with Ghost. The default editor was a markdown which I love, but they added the new editor which is even better. So, this post is actually the first written with the new Koenig editor.
To start with some great news. We had the first Cloud Native meetup here in Sarajevo.
As an organizer, I'm really happy that I started this. We had a great time, great people, and great discussion at the end. Because that was the first meetup I talked about Cloud Native and Why it Matters. All our presentations will be on GitHub, so feel free to take a look.
Also, I wrote two interesting articles. The first one, Serverless: The New Way of Thinking, got some attention with an interesting discussion on Reddit:
Serverless is the popular topic, so make sure you understand what is it all about.
I figured out that people are not aware of all benefits when using Kubernetes ingress controller, even when running the cluster on the cloud. That being said I wrote a blog post on how to save some money when using Kubernetes ingress controller on AWS. TL;DR - With ingress running in your Kubernetes cluster you will only have one ELB for all your external endpoints.
Some other news:
- One of my old posts was featured in DevOps weekly newsletter - Learn How to Troubleshoot Applications Running on Kubernetes
CNCF Speaker’s Bureau
CNCF Speaker’s Bureau is the new service for speakers where you can find all CNCF meetup organizers, ambassadors, and other community members who are willing to talk at local events.
The CNCF Community Speaker’s Bureau helps connect event organizers with speakers with a variety of expertise within the cloud-native ecosystem.
I'm also available on this list. 😎
AWS EKS and Azure AKS Reached GA
Good news from two most popular public cloud providers, AWS and Azure. Managed Kubernetes is now GA. It is interesting that this happened almost at the same time. Learn more on AWS EKS or Azure AKS. Also, there are plenty of articles which compares those two that you might take a look.
HashiCorp released the new version of Consul, but this time it is a big update. There is a new feature called Connect that turns existing Consul cluster into service mesh solution. With dynamic cloud-based infrastructure and cloud-native applications service mesh is the necessity. You can check my post on Kubernetes Service Mesh where I answered on some general questions that you may have about service mesh.
New functionalities of Consul with Connect are:
- Encrypted Traffic
- Connection Authorization
- Native Integration
- Certificate Management and Rotation
For the full feature list check the official blog post.
The Kubernetes community turned four! I still remember exploring it for the first time and making some contributions in 2014.
But, we also have the new Kubernetes version with some new features and stability improvements. The most notable:
- IPVS instead of iptables for in-cluster service load balancing - this change brings better network throughput, better latency, and higher scalability for the Kubernetes service model. IPVS is not the default, but it is production ready.
- CoreDNS is GA and used by default with kubeadm - it is written in Go with fewer moving parts. Also more flexible with support for custom DNS entries.
- Dynamic kubelet configuration is now beta - for live cluster configuration updates.
- Improvements in Custom Resource Definitions - support for multiple versions.
- General improvements in Container Storage Interface - CSI
- New storage features, like, online resizing of Persistent Volumes, dynamic maximum volume count, and safeguard mechanism that prevents deleting PVs and PVCs used by pods
k8s Spot Rescheduler Updates
I mentioned this project a few months ago. If you are using AWS and want to save some money by utilizing spot instances k8s spot rescheduler is your friend. I wrote initial version of Helm chart and using it currently on one project. There were some issues, but all is fixed now and working just like it should. Thanks, everyone for contributing to this project!