Most outages these days are also caused by "devops errors" that're much more likely to happen when working with complex infrastructure stacks. So there's real value in keeping things simple.