Take a look at this for more detail. It refers to how Apache handles multiple requests. Preforking, which is the default, starts a number of Apache processes (2 by default here, though I believe one can configure this through httpd.conf). Worker MPM will start a new thread per request, which I would guess, is more memory efficient. Historically, Apache has used prefork, so it's a better-tested model. Threading was only added in 2.0.