I dropped midway through the post. It reads as a classic engineering justification for X vs A, B or C (the same justification can be made for anything vs anythi

I dropped midway through the post. It reads as a classic engineering justification for X vs A, B or C (the same justification can be made for anything vs anything). I found it interesting that for someone that doesn't use Kubernetes they spend a lot of time describing it. There are benefits in using mainstream solutions like Kubernetes. I've spent over a decade building distributed systems from Hadoop to Mesos and Kubernetes and have seen the pains of datacenter and AWS, Azure and GCP and all I can say is GKE works great so I don't buy the simpler in-house argument. Once you start properly integrating concerns that need to be end-2-end integrated, the complexity explodes and you end up with a partially working system. I do believe that serverless will likely make Kubernetes and similar-level tech irrelevant for most users, however only part of a rich ecosystem that provides all other concerns.