Models of consistency

Strong consistency

• Any read operation always returns the most recent data item.
• Client never sees stale data
• Usually achieved by forcing a replica not to accept new reads/writes until replica has agreed on latest write

Weak consistency

• Read operations may not see the most updated value

Eventual consistency

• Specific form of weak consistency
• All updates are propagated with time. Eventually, all replicas get the same value (the most recent data)
• When concurrent (parallel) writes happen, the system allows the inconsistent values to enter the system and eventually reconciles the nodes to consistent values
• Examples: Dynamo, Cassandra

Related Notes

• Meaning of consistency in distributed systems
• CAP Theorem

Sources

• (Done) Design A Key-value Store - Highlights