essentially it forks before a request comes in, the master process pushes requests to an idle "pre-forked" worker process. Works pretty well under Linux with pretty light processes - pretty badly under windows nt where processes are much fatter than threads.
The main process tries to maintain a pool of idle workers so there is no need to wait for a fork before an incomming request is handled.
Comments
What is a "pre-fork worker model"? I hear that term get thrown around, but I'm not sure what it means.
The Apache documentation has a nice overview:
http://httpd.apache.org/docs/2.2/mod/prefork.html
essentially it forks before a request comes in, the master process pushes requests to an idle "pre-forked" worker process. Works pretty well under Linux with pretty light processes - pretty badly under windows nt where processes are much fatter than threads.
The main process tries to maintain a pool of idle workers so there is no need to wait for a fork before an incomming request is handled.
One "master" process that keeps a pool of worker processes created by forking itself.