TM is using the bgRFC technology for asynchronous processing. The behaviour of bgRFC can be controlled in different ways, e.g. in terms of timing, as explained in
a previous posting.
Another aspect is the number if triggers processed in parallel. To avoid flooding the system this can be controlled globally, but there is also an (optional) setting per bgRFC destination.
All this is done in the transaction sbgRFCconf.
The global setting is done in the scheduler-Tab
In the example it is set to 500 connections.
There is also an option to control this per bgRFC destination. In the TM-Case this is the destination TM_BGRFC_INBOUND, which is created using the report /SCMTMS/TRIGGER_SETUP_BGRFC or manually.
If nothing is defined specifically, the number of connection is defaulted to 10 per destination. In case you have a high load on this destination, this can be changed here. However, the maximum as defined globally for the scheduler will not be exceeded, of course.
And of course the settings must be balanced with the server performance and users and other processes running on the system.