When the MTA starts, it starts one or more router threads to process its "in progress" (gwinprog) queue. As messages arrive in the "in progress" queue, it starts additional routers as needed, within parameters you can set.
By default, the MTA starts another router thread if more than 50 messages are waiting in the "in progress" queue. Use the /queuethreshold switch in the MTA startup file to adjust how backed up the "in progress" queue is allowed to become.
Set /queuethreshold to a lower number if you want the MTA to start an additional router more quickly. Set /queuethreshold to a higher number to conserve resources and keep the MTA from starting an additional router thread even if messages are backed up in the "in progress" queue.
By default, the MTA will continue to start additional router threads to processes messages in the "in progress" queue as long as message traffic demands it, until as many as 16 router threads are running. Use the /maxrouters switch in the MTA startup file to control the number of router threads the MTA can start.
Set /maxrouters to a lower number to conserve resources and keep the MTA from starting more than the specified maximum number of router threads.
By default, after the MTA starts a router thread, it keeps it running, up to the maximum number specified by the /maxrouters switch. In a system where short bursts of heavy message traffic are followed by extended lulls, idle router threads could be consuming resources that would be better used by other processes. Use the /maxidlerouters switch in the MTA startup file to determine how many idle router threads are allowed to remain running. The default is 16 idle router threads.
Set /maxidlerouters to a lower number if you want the MTA to terminate idle router threads more quickly. Set /maxidlerouters to a higher number if you want the MTA to keep more idle router threads ready to process incoming message traffic.