For a small relational dataset in a single location requiring point-in-time recovery, which Google Cloud option is the most cost-effective?

Cloud SQL (MySQL) with binary logging enabled is the most cost-effective option for a small relational dataset in a single location requiring point-in-time recovery. Binary logging captures all data modifications, allowing restoration to any specific point in time. Cloud SQL is a managed service, simplifying database administration. Creating a failover replica in Cloud SQL provides high availability, but it's not directly for point-in-time recovery and adds cost. Cloud Spanner, whether with a 2-node instance or multi-regional, is a globally distributed, highly scalable, and highly available database service designed for much larger, mission-critical workloads with stringent consistency requirements. Its advanced features and distributed architecture make it significantly more expensive and overkill for a "small relational dataset in a single location" scenario.