> The failure modes of boring technology are well-understood
Well-understood failure modes can't get you to resilience and are not all that well-understood to begin with. I guess it's just a false justification people use to stick to familiar broken old technology and broken ways or maybe to avoid relying on people with rare expertise, I don't know. If things were as simple as this we would just handle all the failure modes of "boring technology" in software and get perfectly working systems. But we can't, all the unknown unknowns and not understood failure modes are still there in boring technology, and reliability is a "do backups" afterthought. Systems really have to be designed for reliability if reliability is important and currently software reliability is too cutting edge to be boring.
> I guess it's just a false justification people use to stick to familiar broken old technology and broken ways or maybe to avoid relying on people with rare expertise, I don't know.
This sounds a bit nose in the air to me.
> Systems really have to be designed for reliability if reliability is important and currently software reliability is too cutting edge to be boring.
And yet, people have posted examples of "boring" software running for over a decade.
Depends what you mean... it is often easy to wrap fallible dependency calls with exception logic, and even workarounds sometimes, as long as the logic and failure are clear.
It's mostly a matter of choosing boring software, because it's well documented, both in terms of (semi)-official docs, and things like SO answers, vs. newer software that might have modern advantages and also more undocumented or untested edge cases to be fixed.
Well-understood failure modes can't get you to resilience and are not all that well-understood to begin with. I guess it's just a false justification people use to stick to familiar broken old technology and broken ways or maybe to avoid relying on people with rare expertise, I don't know. If things were as simple as this we would just handle all the failure modes of "boring technology" in software and get perfectly working systems. But we can't, all the unknown unknowns and not understood failure modes are still there in boring technology, and reliability is a "do backups" afterthought. Systems really have to be designed for reliability if reliability is important and currently software reliability is too cutting edge to be boring.