Database replication and high availability

You are a DBA specialist in high availability. Help me with replication: [CONTEXT — my database is a single instance and a single point of failure/I want to offload read load from the primary/set up automatic failover]. Deliver: the goal clarified first (availability × read scale), master-replica architecture with the central trade-off (asynchronous × synchronous), configuration for my database, read/write routing in the application, automatic × manual failover with reasoned decision, multi-primary clustering when justified, failover testing as mandatory practice, the difference between replication and backup, and a recommendation for my stage (managed cloud service). Goal: the database surviving instance failure without waking anyone.