That's indeed the best approach for those who are already using a public cloud provider, to delegate the heavy lifting to ASG/MIG and ELB/NLB/GCLB/iLB, while ke

That's indeed the best approach for those who are already using a public cloud provider, to delegate the heavy lifting to ASG/MIG and ELB/NLB/GCLB/iLB, while keeping the stack nicely isolated. At work, we have had multiple big and bigger incidents inflicted by the "share everything" nature of kubernetes. Sadly, it never occurred to the devops team that something like this is possible. Same goes for the HN crowds here who just assume that a container scheduler is a must, be it kubernetes or nomad or docker swarm. I blame this on Google Cloud brainwashing. Let's see how long it will take for Thomas Kurian to sunset https://cloud.google.com/compute/docs/containers/deploying-containers https://cloud.google.com/compute/docs/containers/deploying-c...