Apache is built around a parent process, which accepts connections from clients, and one or more child processes. These child processes complete the requests and return the accessed page. The number of child processes changes dynamically, according to the number of connections to the Web server. The number of processes is limited by the value specified here.
The Web server will start with the number of child processes specified here.
The maximum number of child processes. Each client connection uses a child process, so this value indirectly controls how many clients can simultaneously connect to the Web server. If this value is too low, clients will be locked out, so be generous. This value is intended mostly as a safeguard against having a runaway web server crash the entire system.
A child process will exit after serving this number of accesses. It is intended as a failsafe against possible memory leaks. Currently, however, there are none known.
Apache makes sure that there always are a few idle child processes, for handling transient load spikes. If fewer than this number of child processes are idle, more are started.
Apache also makes sure that no more than this number of idle child processes exist. If more exist, some of them are killed.