“Most of the time, most people don’t need most of the power of bitemporality”
Rather a strange thing for an author of a bitemporal database to say but, pragmatically, it’s often true - you could certainly be forgiven for eschewing it in favour of something simpler, especially when you’re just trying to get a new system off the ground.
But, as Kent Beck outlines in his recent blog “Eventual Business Consistency”: soon, the time-oriented requirements roll in.
When they do, bitemporality provides us with a useful, rigorously defined mental model to help reason about these time-oriented problems, the tools to implement it and, by approaching problems with this in mind, we become aware of edge cases we may have otherwise missed.
We believe that, overall, it’s a mistake to cast it aside.
Naturally, though, we’d like to have our cake and eat it: we want the ease and performance of a traditional update-in-place database for everyday transactions and queries, but the safety net of a bitemporal database when we need it - this is our primary success criterion for XTDB.
As always, we’re keen to hear your thoughts - see you in the comments section!