The Mirror Activator system is highly fault-tolerant. It is designed to avoid any single point of failure, and to provide automatic and graceful recovery from any system failure.
During normal Mirror Activator system operation, the disk replication system mirrors the primary database log devices synchronously, which guarantees zero data loss—that is, no data for a completed transaction will be lost.
Device-based queues that the Replication Server maintains guarantee transaction delivery to the standby database, even if network failures occur at the standby site, or if the standby database itself fails.
The Mirror Activator system maintains a queue ID in both the Mirror Replication Agent and the Replication Server, which it uses to keep track of the most recent confirmed transaction (successfully replicated) in the standby database. The queue ID allows the Mirror Activator system to automatically recover from system faults, and it prevents either the Mirror Replication Agent or the Replication Server component from becoming a single point of failure.
