Consistency & CAP Study Portal

TL;DR

As soon as data lives on more than one machine, a reader can see a copy that has not caught up with the latest write. Consistency models name exactly what a reader is allowed to see, from linearizable (every read sees the latest committed write, as if there were one copy) down to eventual (copies converge if writes stop). CAP says that during a network partition you must choose between answering (availability) and answering correctly (consistency); PACELC adds that even without a partition you trade latency for consistency. Replication topology — single-leader, multi-leader, leaderless — decides which models are cheap. Multi-region makes the trade-off visible, because a cross-region round trip costs 50–150 ms on every consistent write.