Practice replication lag vocabulary: replica delay from primary, lag spikes during backup windows, lag-triggered read-only mode, and async vs sync replication trade-offs.
0 / 5 completed
1 / 5
What does 'the replica is 30ms behind the primary' mean?
Replication lag of 30ms means writes acknowledged by the primary 30ms ago have not yet been applied to the replica. Reads from the replica may return data that is 30ms stale — a concern for applications requiring read-your-writes consistency.
2 / 5
Why does 'replication lag spike during the backup window'?
Backup jobs compete for disk I/O and CPU with the replication process. During a backup window, the primary may generate write-ahead log entries faster than the replica can apply them, causing lag to spike — and then recover after the backup completes.
3 / 5
A system is configured so that 'lag threshold triggers read-only mode.' What is the purpose of this safeguard?
Some high-consistency systems switch to read-only mode when replication lag exceeds a threshold. This prevents new writes that would widen the divergence between primary and replica, protecting against data loss if a failover occurs while lag is high.
4 / 5
What is the key trade-off between asynchronous and synchronous replication?
Asynchronous replication: the primary acknowledges the write immediately, then replicates in the background — fast but risks losing recent writes if the primary fails. Synchronous replication: the primary waits for at least one replica to confirm the write — no data loss but higher write latency.
5 / 5
An engineer says 'we're seeing 500ms of ___ lag on our read replica.' What word fits?
'Replication lag' is the standard term for the delay between when a write is committed on the primary and when it is applied on a replica. 500ms lag on a read replica means reads from that replica can be up to 500ms stale.