This task would enable master to selectively decide slaves that will replicate events at the time of writing the events. This is possible to inject multiple events for the same TRX , one for each destination domain_id using domain filtering on slaves but it makes a big write amplification

Implementation could inject in binlogs a system variable @domain_receiver='1,2" that each slave would check.

Maxscale, Spider or Galera can potentially use this feature for re-sync a failed nodes

• Spider re-sync tables
• Semi-sync to log events only when not sent to binlog servers
• Galera using replication for IST