What is the point of sending timeouts to a different queue and host in nservicebus?

I notice the default config/recommenda开发者_开发问答tion of the timoutmanager is running it in its own host and pointing to its own queue.

Is it simply for performance?

The timeout manager can manage timeouts for many different nsb endpoint hence the need to run in a separate process. This will also give you fault tolerance for your timeouts