How would you design a globally distributed system with data consistency trade-offs?

A global system balances consistency, availability, and latency using replication strategies and CAP trade-offs.

A globally distributed architecture uses multi-region deployment with active-active or active-passive replication. Strong consistency requires synchronous replication (higher latency), while eventual consistency uses async replication (better availability). Techniques include quorum reads/writes, conflict resolution strategies, and CRDTs in some cases.