Comment by EGreg
This is sorta like the argument about CAP impossibility theorem while in practice consensus algorithms work 99.999% of the time. Or like Shannon’s information theory showing impossibility of compression, while many compression algorithms work well on actual data.
This seems to me the same. In practical applications, you can indeed have the at-least-once delivery with an idempotency / backpressure system, and work 99% of the time, and be unavailable 1% of the time.
Yep, and for practical applications (i.e. stuff that exists in this universe) that is absolutely good enough. You just have to choose which tradeoffs you can stomach best. With a fancy enough system, those tradeoffs can be driven shockingly low.
But if someone tries to sell you a database that is 100% available and has perfect consistency, you can laugh and walk away. They're a flat-earther trying to sell you a bridge: they're either trying to trick you or they have no idea what they're talking about. Either way you don't want to be involved with them.