Kubernetes is a distributed container orchestration system and a seed-bed for microservices. Running Kubernetes and running on Kubernetes is two different things, but none of them is as easy as it sounds, but you can change this.
IBM Budapest Lab (previously Ustream) has been running Kubernetes, and some parts of our system are running on Kubernetes almost a year now, on a very diverse infrastructure, and none of them is Google Container Engine. We run some parts of our streaming stack, testing infrastructure and video backup on Kubernetes and this transformation is still in progress.
In this talk you will learn some best practices of setting up and managing Kubernetes cluster on your infrastructure, how to change your applications to become really cloud native and distributing them via Helm charts. We will also cover what are the benefits of running on Kubernetes instead of VM's and bare-metal boxes, and also touch the topic of managing stateful services on top of Kubernetes, and what challenges we faced.
Nandor is a software engineer in fact, but now builds infrastructure at Ustream. He usually writes distributed software. He worked before in telecommunications where he worked on a radio optimizer system for 3G/4G networks among other things. Later on shifted to the FX business at an investment bank. In his free time he contributes to open-source (github.com/bonifaido). Nowadays containers and distributed schedulers are what makes him happy and relaxed. He only feels satisfied when all environments are reproducible and identical.
