18 May
2013
18 May
'13
9:07 p.m.
I want to set up an io_service-per-thread, with each io_service having a deadline_timer to handle the heartbeat. The only other complicated bit is selecting an io_service to handle an incoming request.
Can't you just "round-robin" them?
It's a bit irksome to have to trade off the scalability to get this extra feature.
Why io_service-per-core would give worse scalability than thread-per-core?