Which change will meet these requirements?

Explanations:

Adding an Aurora Replica increases read capacity by distributing read requests across multiple nodes, which can reduce RAM consumption and improve select latency during traffic surges. This approach also minimizes downtime and does not affect data integrity.

Converting to a multi-master setup can improve write availability but not necessarily optimize read performance or memory usage. Multi-master also introduces additional complexity and might require downtime for conversion.

Creating a new DB cluster from a snapshot with larger instances could improve performance but requires downtime for the transition, as it involves moving to a new cluster. This does not meet the requirement of minimizing downtime.

Increasing disk storage capacity would only affect storage constraints, not RAM or read latency. Since the issue is related to memory consumption and read performance, this option does not address the root cause.