Replication Lag
The delay between a write hitting the primary and showing up on a replica.
Replication lag is how far behind a replica is from the primary database — the window during which a committed write is not yet visible on the replica. It is the practical face of eventual consistency in a primary-replica setup.
Worked example: a user updates their profile (the write goes to the primary) then reloads and reads from a replica lagging 200ms behind, so they briefly see the old profile — the “read-your-writes” problem. Gotcha: lag grows under write bursts or slow replicas, so critical read-after-write paths must route to the primary or wait for the replica to catch up; monitoring lag matters because a silently growing lag means stale reads everywhere.