Been a while, but we are happy to announce that we are trying to restart our Meetups. For this one we have Jrme talking about not so common aspects of Kubernetes. We would like to thank John from the Bay Area Amazon Webservices Meetup group to help us with the venue on short notice.
Understanding Kubernetes by building our cluster one piece at a time
In this talk, we will show how to set up a simplified, artisanal, hand-crafted, small-batch, Kubernetes cluster, without using external tools. It will definitely not be production-ready or secure or compliant, but it will be a great way to understand how Kubernetes works, or to prepare for the Certified Kubernetes Administrator exam from the CNCF, for instance.
Running Kubernetes without nodes: why & how
On a traditional Kubernetes cluster, our applications run in containers. A container is part of a pod, & a pod runs on a node. The Kubernetes scheduler allocates pods to nodes to optimize resource utilization, but even with the best scheduler in the world, our cluster is almost always too big (and we're then wasting resources) or too small (and we have to evict lower priority workloads). Enter solutions like Virtual Kubelet or Milpa, which provision resources on the fly when we need them. This means that we pay only for what we use, while being able to use all the capacity offered by our cloud provider. We'll explain how this works, with plenty of demos.
Bio
Jrme Petazzoni was part of the team that built, scaled, & operated the dotCloud PAAS, before it became Docker. He worked seven years at the famous container company, wearing various hats. When he's not busy with computers, he collects musical instruments. He can arguably play the theme of Zelda on a dozen of them.