Scale-out and RBR : any benchmarks/experiences ?

Hello,

I’m using SBR replication for scale-out with mysql stable (5.0.46), and some triggers on each slaves which build new tables.

I have really good performance with this trick, but at a specific level arround 50% cpu usage, queries are queued and not executed, cpu usage go to 100% …

With RBR, I could use a server which replicate from master, have triggers and which is also a RBR master.

Others slaves with RBR will have less load for replication and will handle more queries, no ?

Greg