The hard part about working at scale is not that you need to use more efficient algorithms or combine computers to handle the load. It is that all the weird and spooky non-reproducible edge case stuff is happening constantly & will actually demand your attention.
The lower your scale, the more you can get away with using a "high scalability" database. Who cares how concurrent transactions are reconciled when you've never seen one? I would wager that most users aren't actually handling the caveats at application layer. They're just not hitting the weird stuff often enough to ever really consider it.
The lower your scale, the more you can get away with using a "high scalability" database. Who cares how concurrent transactions are reconciled when you've never seen one? I would wager that most users aren't actually handling the caveats at application layer. They're just not hitting the weird stuff often enough to ever really consider it.