A financial services company hosts a critical online transaction processing (OLTP) database server with a zero-tolerance policy for data loss (RPO of zero). The disaster recovery site is in a separate geographic region connected by a high-latency WAN link. The primary business requirement is to ensure that no committed transactions are lost if the primary site fails. Which disaster recovery replication method should a server administrator implement to meet this requirement, despite the potential performance impact on the primary site?
The correct answer is synchronous replication. Synchronous replication guarantees data consistency between the primary and secondary sites by writing data to both locations before the transaction is committed and acknowledged. This method ensures a Recovery Point Objective (RPO) of zero, meaning no data is lost upon a primary site failure. The question specifically notes the tolerance for a performance impact, which is a known trade-off of synchronous replication over high-latency WAN links.
Asynchronous replication writes data to the primary site first and then copies it to the secondary site, which creates a potential for data loss (a non-zero RPO) if a failure occurs before replication completes.
Application-consistent snapshot replication creates point-in-time copies that are consistent for applications, but the replication of these snapshots is periodic, resulting in a non-zero RPO.
Bidirectional replication is used for active-active environments where two sites replicate to each other, which is a different architecture than the primary-secondary disaster recovery scenario described.