How should you design a highly available Cloud SQL for PostgreSQL instance to support 100 databases, each with 80 tables migrated from on-premises to Google Cloud, ensuring low-latency read and write operations for applications distributed across multiple US regions?
