> And use several providers if you have multiple domains in case one implodes. I'd add only on refinement: if possible, use domains with different top-level (n

> And use several providers if you have multiple domains in case one implodes. I'd add only on refinement: if possible, use domains with different top-level (national) authorities. Sadly, this kind of operational wisdom is rapidly being lost by the emphasis on hiring "DevOps" engineers, with most of the emphasis on the "Dev" part, since they're often just coders against cloud APIs, with much less value placed on traditional sysadmin (or what ended up being called operations engineering during and after the dot-com boom) experience. Now, of course, that makes sense not to hire a full-time 100% sysadmin if what you (think you) need is a 5% (or less) sysadmin. Also, most startups, and probably even larger businesses, are going to be easily lulled into complacency by the uptime records of the larger vendors (or really just AWS), when following the best practices for technical reliability. "Infrastructure as code" is supposed to alleviate the human/administrative risk, and I'm actually convinced that it does to a very large degree (just at a huge cost in markup/profit for Amazon for that infrastructure).