A web app currently serves from a single Asia region, must use a native JDBC driver, optimize for cost now and global presence later. Which database approach fits initial cost focus and later multi-region needs?

Cloud Spanner is the best choice because it offers a native JDBC driver and can be configured as a single-region instance for initial cost savings. Crucially, it allows seamless conversion to a multi-region configuration later without application re-architecture or downtime, directly addressing the requirement for future global presence. Cloud SQL for PostgreSQL, while offering a JDBC driver, does not inherently support seamless multi-region expansion in the same way Spanner does. Migrating from Cloud SQL to Bigtable would involve significant application changes, as Bigtable is a NoSQL wide-column store and not a relational database, failing the "native JDBC driver" and "optimize for cost now" requirements for a relational database. Migrating from Cloud SQL zonal to HA is an upgrade within Cloud SQL but doesn't address the multi-region requirement for global presence.