CAP theorem

The CAP theorem states that a distributed data store cannot guarantee consistency, availability, and partition tolerance simultaneously when a network partition occurs.

Consistency (CAP theorem): In CAP, consistency means every read receives the latest successful write or an error. This is the specific consistency guarantee that conflicts with availability during a partition.
Network partition: A communication failure that splits a network into groups unable to exchange messages across the split. A partition creates the condition in which CAP's competing guarantees cannot all hold.
CP system: A distributed system that preserves consistency during a partition by withholding some responses or operations. It sacrifices availability when necessary to avoid inconsistent results.
Eric Brewer: A computer scientist known for work on distributed systems and for formulating Brewer's conjecture. Brewer presented the conjecture that became known as the CAP theorem.
FLP impossibility result: A theorem showing that deterministic consensus cannot be guaranteed in a fully asynchronous system with even one possible crash failure. FLP concerns consensus termination under asynchrony, not CAP's three-way data-store guarantee.
Availability (CAP theorem): In CAP, availability means every request to a non-failing node receives a response, though it may be stale. CAP availability requires responses even when some nodes cannot communicate.
Asynchronous system: A computational model in which message delays and process speeds have no known upper bounds. Unbounded delays make a failed link indistinguishable from a very slow one.
AP system: A distributed system that continues responding during a partition while allowing replicas to diverge temporarily. It favors availability and reconciles conflicting state after communication returns.
Seth Gilbert: A computer scientist whose work includes formal results on distributed algorithms and impossibility. Gilbert co-authored the formal proof of the CAP result.
Two Generals' Problem: An impossibility result showing that two parties communicating over an unreliable channel cannot achieve common knowledge of coordinated action. It illustrates a different limit on coordination over unreliable communication.

Linked from 2 pages