What is Replication Lag?

Replication lag is the measurable time difference between when a write transaction commits on a primary database and when that same change is applied to one or more read replicas. It is a natural consequence of asynchronous replication, where the primary does not wait for replicas to acknowledge the write before returning success to the client. The lag can be expressed in seconds or as a count of transactions queued for replay.

Replication lag arises from several factors. Network latency between the primary and replica adds a base delay. The replica's own processing capacity, especially under heavy read load, can slow the application of incoming changes. Large transactions, such as bulk updates or schema changes, take longer to replay and increase lag. In systems like MySQL's asynchronous replication or PostgreSQL's streaming replication, the lag is typically sub-second in well-tuned environments but can grow to seconds or minutes during bursts of write activity or replica resource contention.

In the database stack, replication lag is a key concern for applications that require strong read-after-write consistency. If an application writes to the primary and immediately reads from a replica, it may see stale data if the lag is nonzero. Engineers monitor lag using tools like SHOW SLAVE STATUS in MySQL or pg_stat_replication in PostgreSQL. Mitigation strategies include using synchronous replication (which reduces lag to zero at the cost of write latency), routing critical reads to the primary, or implementing session-level consistency guarantees.