CAP Theorem
CAP is acronym for Consistency, Availability, and Partition Tolerance. It is a theorem that states that it is impossible for a distributed data store to simultaneously provide more than two out of the following three guarantees:
- CA (Consistency and Availability) - All nodes see the same data at the same time. This guarantee is not applied for distributed systems because they have to be Partition Tolerance
- AP (Availability and Partition Tolerance) - Every request receives a (non-error) response, without the guarantee that the data is the most recent.
- CP (Consistency and Partition Tolerance) - Every request receives a (non-error) response, with the guarantee that the data is the most recent.
Strong Consistency - during the write operation, all replicas in the system should be updated before the write operation is acknowledged as performed. In other words, any read operation always returns the most recent write. Achieving strong consistency requires a high level of coordination and synchronization between nodes, which can affect availability.
Eventual Consistency – during the write operation, the replicas in the system are updated asynchronously. In that case, replicas in the system could return stale data for a period of time until the change is propagated to all replicas.
- Video by Studying with Alex: Friendly Intro To the CAP Theorem (Consistency vs Availability vs Partition Tolerance) – I liked the examples which were used to explain the CAP theorem.
Last modified 5mo ago