What is CAP Theorem?

The CAP theorem, also called Brewer's theorem, was introduced by computer scientist Eric Brewer in 2000 as a conjecture about the fundamental trade-offs in distributed systems. It was formally proved by Seth Gilbert and Nancy Lynch in 2002. The theorem describes three properties: consistency, meaning every read receives the most recent write or an error; availability, meaning every request receives a non-error response, without guarantee that it contains the most recent write; and partition tolerance, meaning the system continues to operate despite arbitrary message loss or network failures between nodes. According to the theorem, a distributed system can achieve at most two of these three properties simultaneously.

When a network partition occurs, meaning messages between nodes are lost or delayed, the system must choose between maintaining consistency and maintaining availability. If it chooses consistency, it must reject requests that cannot be verified against the latest state, thereby sacrificing availability. If it chooses availability, it may respond with stale data, sacrificing consistency. In practice, network partitions are inevitable, so distributed systems are usually designed as either CP (consistent and partition tolerant) or AP (available and partition tolerant). A system that claims to be CA is only possible if partitions never happen, which is unrealistic in a real-world distributed network.

Many modern distributed databases and services are architected with explicit CAP trade-offs. For example, Apache HBase and ZooKeeper favor CP by refusing writes during a partition to ensure consistency. Apache Cassandra and Amazon DynamoDB favor AP by allowing writes and reads from any replica even if some replicas are unreachable, accepting eventual consistency. The CAP theorem continues to guide system design, though later work such as the PACELC theorem extended the model to include trade-offs even in the absence of partitions.