# CAP Theorem

* Fundamental principle in distributed systems.
* Tradeoff between desirable properties: Consistency, Availability, Partition Tolerance.

## Consistency (C)

* Every read receives the most recent write (or an error).
* This ensures that all nodes in the system have the same data at any given time.

## Availability (A)

* Every request (read or write) receives a response, regardless of whether it is successful or not.
* The system remains operational even in the presence of some node failures.

## Partition Tolerance (P)

* The system continues to function even when there is a network partition (a situation where communication between nodes is disrupted).

## The Theorem

* The CAP theorem states that a distributed system cannot guarantee all three properties simultaneously.
* Can only provide at most two out of the three.
* **Consistency and Availability (CA):**
  * Works well if there are no network partitions.
  * Examples: Traditional relational databases (if deployed without replication).
* **Consistency and Partition Tolerance (CP):**
  * Guarantees consistency even during network partitions but may sacrifice availability (e.g., by rejecting requests during a partition).
  * Examples: Distributed databases like MongoDB (in some configurations).
* **Availability and Partition Tolerance (AP):**
  * Guarantees availability even during network partitions but may sacrifice consistency (e.g., by allowing nodes to return stale or divergent data).
  * Examples: Systems like DynamoDB, Cassandra.

## Practical Implications

* Distributed systems designers must choose which two properties to prioritize based on the application's needs.
* Real-world systems often make trade-offs to provide "good enough" versions of all three, but strict guarantees are subject to CAP's limitations.

## Example

* Consider a banking system:
  * If consistency is prioritized (e.g., accurate account balances), availability might be sacrificed during network partitions.
  * If availability is prioritized (e.g., ensuring that ATMs are always operational), consistency might be relaxed during partition events.