The correct answer is C because read replicas are designed to handle read traffic, thereby reducing the load on the primary database. This approach is scalable and efficient for read-heavy applications.

• Option A is incorrect as backups are intended for disaster recovery, not for scaling read operations.
• Option B is incorrect because high availability configurations are aimed at ensuring failover capabilities, not improving read performance.
• Option D is incorrect due to the complexity and maintenance overhead associated with managing external replicas, making it less ideal for scaling purposes.

For more details, refer to the GCP documentation on Cloud SQL Replication Options.