After using k8s for about a year at work now, the way I understand it is that there’s a really large gap in abstraction between “a bare metal machine or VM with

After using k8s for about a year at work now, the way I understand it is that there’s a really large gap in abstraction between “a bare metal machine or VM with a shell prompt” and “a service that handles requests with some Python code”, and kubernetes helps fill that gap with some sane defaults and useful abstractions for people who run services. If you run your docker app on kubernetes, you get a lot of things for free with the platform (rolling no-downtime deployments, service discovery, auto scaling) that you’d have to set up manually if you were running your service on (say) EC2 instances instead. It can be a headache to learn sometimes, but ultimately saves a lot of effort if your use case fits!