The classic use of fork() is in client-server applications. The server just listens for new inbound connections and when one is detected, it accepts the connection to this new client and makes a "copy of itself" to handle whatever business this client wants. This "new copy" is known as a child process. The code is simplified as the server doesn't know how many "children" it has and each child deals with one client. Roughly speaking, this "start new copy of myself" is fork(). On Windows, this doesn't quite work this way and details get complicated, but same idea applies. There are a lot of "pesky details" to building good, robust client server applications.

One basic thing is: don't use system() or exec() if you can call a Perl module or function to do what you want. For example, you wouldn't want to call a shell command like 'ls' on Unix or 'dir' on Windows to list a directory.

Any way in programming fork is not a dining utensil, it is more like a "fork in the road".