>The fact that persistent storage has to happen at the same place to be reliable is only incidental, not a defining attribute of a relational database.
Persistence happens to processes who's state is expensive to reconstruct. Measurements of the world requires time travel to recover and so are quite expensive to acquire! "Incidental" implies a weaker correlation than warranted.
As an aside, ~100% of data models are incapable of modelling contradictary or ambiguous measurements and fail to adequately model alternative normalization and integration into whatever model of "truth" you pick. And these systems entirely fall over when any part of these tacit and underspecified constraints change.
I definitely agree that models are difficult and contextual. I do think that minimizing distances and entropy does reduce the difficulty of doing migrations and refactoring of the data model. Doing migrations across spread out, distributed data is much more difficult.
Persistence happens to processes who's state is expensive to reconstruct. Measurements of the world requires time travel to recover and so are quite expensive to acquire! "Incidental" implies a weaker correlation than warranted.
As an aside, ~100% of data models are incapable of modelling contradictary or ambiguous measurements and fail to adequately model alternative normalization and integration into whatever model of "truth" you pick. And these systems entirely fall over when any part of these tacit and underspecified constraints change.