PACELC
An extension of CAP stating that during a partition you trade availability against consistency, and else — in normal operation — you trade latency against consistency.
CAP describes what happens during a partition, which is a rare event. PACELC adds the part that governs every millisecond of normal operation, and that is why it is the more useful formulation for day-to-day design.
Read it as: if Partition, choose Availability or Consistency; Else, choose Latency or Consistency. The second clause is the one that shapes systems, because strong consistency in normal operation still requires coordination — a quorum acknowledgement, a leader round trip, a consensus commit — and coordination costs latency that scales with the distance between replicas.
This explains designs that CAP alone cannot. A database replicating synchronously across regions is consistent and pays 80 ms per write for it. The same database replicating asynchronously answers in 2 ms and can lose recent writes on failover. Neither is partitioned; both made a choice.
The classification is useful shorthand in design discussions: a system can be PC/EC — consistent always, paying latency — or PA/EL — available and fast, eventually consistent — or PC/EL, which is the most common real configuration and the least discussed: consistent during a partition, but tolerating stale reads from replicas in normal operation for speed.