Distributed systems: choice without choice
When there is too much data for one server, a cluster of several nodes is built. As soon as there is more than one node, network delays and scenarios of “break” of communication between parts of the system appear. The CAP theorem formulates fundamental limitations: in such conditions, it is impossible to simultaneously satisfy all three desires as much as possible - you have to consciously set priorities (for example, for a global social network vs a payment core).