Kubernetes, Take the Wheel: Lessons Learned While Containerizing All the Things
Over the past few years, I've been honing the hyperconverged infrastructure I use to run my test lab at Red Hat, with a focus on wringing the most out of the hardware available to me while keeping the whole thing resilient enough to rely on. One of the tricks of mashing together projects such as oVirt virtualization and Gluster storage, among others, is that these components tend to assume that they have resources all to themselves -- a challenge surmountable through virtualization, but with hefty resource overhead. Sounds like a job for... Containerization! Google is said to run everything, including the VMs it rents out as part of its IaaS service, in containers -- why not try a similar strategy for my own needs? In this presentation, I'll talk about the lessons I've learned while adapting my test lab infrastructure to Kubernetes and Docker, including discussion of the limitations and benefits inherent to this sort of stack, and a look at upcoming features and capabilities to keep an eye on in this space.