Remove one secondary cluster from the Aurora global database to reduce replication lag blocking writes
A company uses an Amazon Aurora PostgreSQL global database that has two secondary AWS Regions. A DevOps engineer has configured the database parameter group to guarantee an RPO of 60 seconds. Write operations on the primary cluster are occasionally blocked because of the RPO setting. The DevOps engineer needs to reduce the frequency of blocked write operations. Which solution will meet these requirements?
Community Votes
100% of anonymous learners picked answer C. Votes are pick records left by other test-takers — they are not the verified answer.
Community Insight
The blocking is caused by replication lag against the 60-second RPO target across two secondary Regions, so the remedy is to reduce the number of replication targets rather than to change how writes are routed (C). Adding another secondary cluster would increase the load and worsen the problem (A), which is the opposite of what is needed. Enabling write forwarding (B) reroutes writes to a secondary Region, which addresses read scaling and write distribution rather than the primary's replication lag. Configuring synchronous replication (D) would make lag affect commit latency directly, so it would increase blocking rather than reduce it.
Write operations on the primary are blocked because the secondary clusters cannot keep up with the configured 60-second RPO target. Each additional secondary Region adds replicated write load that the primary must wait on, so reducing the number of secondary clusters directly reduces the write latency the primary must tolerate and therefore reduces how often writes are blocked. Removing one of the two secondary clusters leaves the global database with fewer replication targets, which lowers the chance of exceeding the RPO threshold during write bursts.
Adding an additional secondary cluster to the global database (A) — spring21 identified this as counterproductive, because each additional secondary Region adds replicated write traffic that the primary must wait on, which increases the very lag that is blocking writes. Enabling write forwarding for the global database (B) — teo2157 considered it, and while write forwarding allows secondary Regions to handle writes, it addresses write distribution and read scaling rather than reducing the primary's replication lag against the RPO target, so it does not unblock the primary writes described. Configuring synchronous replication (D) — youonebe noted that Aurora global databases replicate asynchronously by default for fast cross-Region replication, and forcing synchronous replication makes commit latency depend on the slowest secondary, which would make the write blocking worse rather than better.
Community Discussion (4 comments)
Comments & Corrections
No comments yet — spotted an error or have a note? Share it below.
Expert Analysis
Why the Answer Is Correct
The write blocking is a direct consequence of the primary unable to replicate fast enough to keep every secondary Region within the 60-second RPO target, so the Aurora engine throttles writes at the primary rather than let replication lag exceed the configured objective. With two secondary Regions, the primary is replicating every write twice, and each additional Region adds write load and therefore additional lag. Removing one of the secondary clusters reduces the number of replication targets, which lowers the replication work the primary must complete for each write and therefore reduces how often the RPO threshold is exceeded and writes are blocked (C). This keeps the global database intact with a secondary Region available for failover while lowering the replication pressure. C is the correct answer.Why the Other Options Are Wrong
A adds an additional secondary cluster to the global database. As spring21 identified, this is counterproductive: each additional secondary Region adds replicated write traffic that the primary must wait on, increasing the replication lag that is causing the blocking, so writes would be blocked more often rather than less. B enables write forwarding for the global database. teo2157 considered this option and noted that write forwarding allows secondary Regions to handle writes, but its purpose is to distribute write workloads and scale reads across Regions, not to reduce replication lag on the primary; the blocking described is caused by the primary's own replication lag against the RPO target, which write forwarding does not address. D configures synchronous replication for the global database. As youonebe noted, Aurora global databases replicate asynchronously by default precisely to provide fast cross-Region replication; forcing synchronous replication makes the primary's commit latency depend on the slowest secondary, so replication lag would directly translate into commit latency and the writes would be blocked more often, which is the opposite of the goal. C is correct.Community Comment Notes
Community was split, C (71) against B (14), D (14), and a small A vote. spring21 supported C, explaining that the goal with a 60-second RPO is to improve replication behaviour so writes are not blocked and that removing a secondary cluster reduces the write contention the primary experiences. DKM agreed, noting that removing a secondary cluster alleviates write contention on the primary and improves overall performance. teo2157 favoured B on the basis that write forwarding allows secondaries to handle writes. youonebe favoured D but its reasoning argues against it, that Aurora global databases replicate asynchronously for fast cross-region replication and that synchronous replication would make lag affect commit latency directly. No alternative received majority support.Official Reference
Related Analysis
Practice All DOP-C02 Questions
Access 85 questions with complete answers and detailed explanations.
View Full DOP-C02 Practice Test →