You can use whatever scheduling mechanism you want, really, all it has to do is make a database connection and execute a proc.  Personally, I usually use SQL jobs and DTS packages, as they're built in and pretty easy to use.

It's usually a process that "wakes up" every X amount of time and checks for new records to be processed (in the table being updated by the trigger).  For example, you could send an email for each row, update a web page, whatever (DTS is a good way to do these sorts of things if you're not a SQL guru).

This makes sense only in after triggers, really.  If the process needs to occur before the table with the trigger is updated then this won't work.  Rarely would an external process actually have to occur BEFORE the insert or update occurs.