Eventual Consistency
Summary: A consistency guarantee where all replicas will eventually converge to the same state if no new updates are made for a certain period.
Sources: chapter5
Last updated: 2026-04-15
Eventual consistency is a common model in distributed systems, particularly those using asynchronous-replication. It prioritizes availability and low latency over strong, immediate consistency (source: chapter5, p. 161).
Characteristics
- Replication Lag: There is typically a delay between when a write is accepted by the leader and when it’s available on all followers.
- Convergence: All replicas are guaranteed to reach the same state eventually, but the time it takes to reach that state is not specified.
- Read Consistency: Users might read different values from different replicas depending on the replication lag (source: chapter5).
Stronger Guarantees
While eventual consistency is common, applications often require stronger guarantees, such as:
- read-after-write-consistency: Ensuring users see their own updates immediately.
- monotonic-reads: Preventing users from seeing data move backward in time.
- consistent-prefix-reads: Maintaining causality in sequences of writes (source: chapter5, p. 163).