It also sounded unmanageable the first time I heard the idea of consistently replicating databases across clusters in multiple availability zones, but ultimately the vast majority of tech companies nonetheless decide it's worth going through all that trouble to avoid existential risk in the event of a SPOF.

I'm curious about the calculus that makes this tradeoff non-worthwhile when it comes to squirreling away an emergency payroll fund in a backup bank. What about that is unmanageable?