A second region must satisfy a named requirement#
Regions can reduce user latency, provide a recovery domain, address data residency, or add capacity. They also introduce replication delay, routing, conflicts, data locality, failover, and significant operational cost.
A user updates name = May in Region A; before replication reaches B, a read in B may still return Maya. The design must choose write ownership, conflict behavior, and whether session routing or another mechanism provides read-your-writes.
Failover capacity is not implied by redundancy
40K + 40K = 80K RPS demand
80K - 60K = 20K RPS beyond B's capacityThe surviving region needs headroom, tested scaling, degradation, or shedding. A second region does not automatically make failover viable.
Related practice problems
Use the manifest-backed prompts below to apply this lesson in a complete design.
- Use defined recovery strategies to meet recovery objectivesAWS Well-Architected Framework
Reach this point to mark the lesson as read.