VSHN.timer #37: Extreme Kubernetes

Welcome to another VSHN.timer! Every Monday, 5 links related to Kubernetes, OpenShift, CI / CD, and DevOps; all stuff coming out of our own chat system, making us think, laugh, or simply work better.

This week we’re going to talk about stretching Kubernetes in all directions, testing its capacity and resilience. Ready for a workout?

1. Red Hat is pushing OpenShift to the limits. They recently achieved the impressive milestone of running 500 pods per node. Not just “hello world” kind of pods, but rather actual workloads that teams actually need. We’re big fans of OpenShift at VSHN, from offering APPUiO to our customers, to running CodeReady Containers in our laptops, and this one made us drool. The article contains the full description of the hardware, software, and configuration settings required to do this.

https://www.openshift.com/blog/500_pods_per_node

2. The analogy of pets vs. cattle for cloud service management goes way back, apparently as back as 2006. But now in 2020, as we move up in the abstraction hierarchy, it is now time to start treating Kubernetes clusters as cattle; for example, with Project Syn. Thanks to Kubernetes’ flexibility, having a single team managing many different Kubernetes clusters is not unheard of. Enter Rancher Fleet, a new product aimed specifically for this task, and a natural step for Rancher, given the wild popularity of K3s.

https://rancher.com/blog/2020/fleet-management-kubernetes/

3. Speaking about moving upwards the abstraction hierarchy, Helm is a great example. But Helm begat Helmfiles. Which begs the question, how to best start the development of our own Helm charts and Helmfiles? Paul Czarkowski has distilled a starter kit of very opinionated best practices and templates just for that. A great resource to help us stand on the shoulder of giants.

https://github.com/paulczar/helmfile-starter-kit

4. Whoever has to work concurrently with a myriad of Kubernetes clusters (remember, the new cattle) feels the pain. Switching from one kubectl (or oc) context to another can quickly become a hassle. Hence Kubie by Simon Bernier St-Pierre; a tool created specifically to solve that problem. And now your local Minikube test cluster is just a kubie ctx away.

https://blog.sbstp.ca/introducing-kubie/

5. The tool of the week is pgsync, a tool by Instacart to sync data between PostgreSQL databases, designed for speed, security, flexibility and convenience.

https://github.com/ankane/pgsync

How many clusters do you currently run and/or manage? What other tools or techniques help you and your team do that? Get in touch with us through the form at the bottom of this page, and see you next week for another edition of VSHN.timer.