An Aurora MySQL cluster with one read replica slows down when concurrent client connections exceed 200 (typical load ~180, with spikes past 200). The team wants the database to autoscale as demand varies. Which change will meet this goal?

Attaching an Aurora Auto Scaling policy with a target of 195 DatabaseConnections is the correct solution. Aurora Auto Scaling can automatically add or remove Aurora Replicas based on metrics like DatabaseConnections, ensuring performance is maintained during traffic spikes. Setting the target to 195 connections allows the system to proactively scale out before the 200-connection threshold is consistently breached. Migrating to an Aurora multi-master cluster provides high availability and multiple write nodes but doesn't inherently autoscale read capacity based on connection count. Switching to serverless mode dynamically is not a feature of Aurora; Aurora Serverless is a distinct deployment model, not a dynamic scaling option for provisioned clusters. Scaling up the instance class of the existing replica would increase its individual capacity but wouldn't provide the dynamic, elastic scaling needed for fluctuating concurrent connections.