Chosen option is the best approach:
* Single-region data locality: By pointing your automated backups at a region-scoped Cloud- Storage bucket, you ensure all data (including backups) stays in-region.
* RTO ≤-1-day: Automated daily backups let you restore from the most recent snapshot within your 24-hour RTO window.
* Cost minimized: Disabling HA and not spinning up any replicas avoids the extra instance costs.
You pay only for one primary instance plus your small storage charges for ~10-GB of backups.
* Simplicity: Cloud-SQL's built-in automated backups require zero operational overhead beyond the one-time bucket configuration.

It is always more secure to use an internal IP, so removing them doesn't make sense. Eliminate A. You can use Private Google Access when VM instances only have internal IP addresses, so disabling the internal IPs and use Private Google Access doesn't make sense. Eliminate B. Specifying an authorized network when they're on the same subnet doesn't make sense. Eliminate C. A way to improve network security would be to disable external IPs since they're not needed.

Cloud Spanner meets the requirement for strongly consistent transactions through externally consistent reads and writes, which allows you to process OLTP workloads without violating correctness during spikes. The multi-regional configuration provides a five nines availability target through synchronous replication and automatic failover, which aligns with the 99.999 percent requirement. SSD storage underpins low latency performance.
For users on three continents, a global multi-region such as a configuration that spans North America, Europe, and Asia enables sub second reads from nearby replicas using bounded staleness, while transactions remain strongly consistent by committing through the leader. Cloud Spanner also scales horizontally and can use processing units and compute autoscaling so you can expand capacity during 8x surges and reduce it when demand subsides to help keep spend under control.