A lot of daemons like to run separate processes to handle a lot of the same thing (like HTTP requests, for example), the same with threads, only I know each thread can handle more than one request at a time (event driven).
So why? If one thread, let alone one process, can handle multiple requests (and running a single thread for each request is considered bad), why even have more than one of each? Couldn't you theoretically just pump up the given CPU time of your thread to match that of several normal threads and/or other processes, and just do it all through there?
Why so split up, and yet not completely split up?