Railway should support global databases with multi-region replicas and automatic routing to the nearest node. This would allow services to maintain low-latency, stateful connections while ensuring high availability and failover support.

I have an api service and database in Singapore. If I want to improve latency in the US by having the service there there the latency increases as my database is in a different region.

The HA clusters seem to only be used for failover and does not work as read replicas that automatically direct reads to the correct node.