Recently I’ve been forced to learn more about Kubernetes as a development and deployment tool because of some changes happening at my day job. I started hacking away with minikube and dabbled with kind. Both tools had their strengths and weaknesses, but there were some parts of interacting with them that didn’t quite mirror the real world. With minikube and kind there is no real great way to test an ingress controller like istio or nginx or play with something like external-dns. Certainly both have their use for testing certain types of kube manifests, validating configs, etc. But then there are times you want to test how an application actually scales, not necessarily for a full blown load test, but just to make sure things actually get better with two nodes instead of one. So I took it upon myself to test some of the offerings from the likes of Vultr and Linode, and I landed on Vultr only because of the $2/mo cost difference per node for the cheapest offering. Okay, and maybe also the fact that I’m still sad that Linode, who I’ve been using since 2008 or so, got bought by Akamai.
Some Clarifications and My Goals
First and foremost I want to clarify something. Someone reading this might think my employer is not equipping me to do my job such that I’ve set out to use Vultr or Linode instead of just…
