Eventual Consistency
In plain terms
A model where data copies converge to the same value, just not instantly.
Going deeper
In distributed systems prioritizing availability, updates propagate over time, so different replicas may briefly disagree before settling. It trades immediate accuracy for scale and uptime.
Under the hood
Tunable consistency (quorum reads/writes) and conflict resolution (last-write-wins, CRDTs, vector clocks) determine how and when replicas converge.
Examples
DNS propagation, DynamoDB, Cassandra, shopping-cart merges, S3 (now strongly consistent).
Think of it like...
Like a rumor spreading through an office: not everyone hears it at the same moment, but eventually the whole floor knows the same story.