Eventual Consistency

Eventual consistency is a consistency model in which updates propagate to all replicas over time, so reads may briefly return stale data, but all replicas converge on the same value once updates stop.

Why it matters

It lets distributed systems stay fast and available across regions, because writes do not wait for every copy to confirm. DNS is a familiar example: a changed record spreads as caches expire. The cost is that applications must tolerate stale reads and sometimes resolve conflicting writes, using techniques such as last-write-wins or version vectors.

Example

t0  write x=2 on node A
t1  read x on node B -> 1   (stale)
t2  replication arrives at B
t3  read x on node B -> 2   (converged)

Common mistakes

• Showing users stale data after their own save, without read-your-writes handling.
• Assuming last-write-wins is safe when clocks across servers are not perfectly in sync.
• Applying it to data that needs strict correctness, such as account balances.

Related

• CAP theorem
• Database replication
• DNS propagation
• TTL

This page is general information written by our editorial team and updated when documentation changes. Servers and providers differ, so test changes safely first. See our editorial policy and disclaimer.