> YES! This is exactly what we should have had been able to do in K8S swiftly, but I would like to suggest one additional goal: to get a highly-distributed, fau

> YES! This is exactly what we should have had been able to do in K8S swiftly, but I would like to suggest one additional goal: to get a highly-distributed, fault tolerant computer cluster over different network without a hassle (as a mean to spread risk) In practice this is going to be tricky unless your services are completely stateless. For 80% of people that's going to be true, but if you have customers with large datasets, (I'm thinking mostly media) you do not want to be schlepping those things between clouds or between cloud and on prem or even between on prems.