Adding another AOR and distributing workload via dynamic routing. More expensive (another region to manage, another set of resources) but provides linear scalability and improved failure isolation.