Multi-region database replication: our journey from single to global

we're expanding our services globally, specifically into europe and asia, and our current single-region database setup is becoming a major bottleneck due to latency. users in singapore are experiencing 250ms+ round trips just for basic data access, which is unacceptable. we've explored basic read replicas in other regions, but that doesn't solve the write latency or data residency challenges. we're looking at options like cockroachdb or yugabytedb for true multi-region consistency, but that's a significant architectural shift and operational commitment. another idea is using edge functions combined with a regional db, but that adds complexity for data synchronization. how have others successfully moved from a single-region to a globally distributed database architecture? what were the biggest hurdles you faced, especially regarding consistency, latency, and data residency requirements?