Ubuntu – Why is Apache running so many processes? Excessive RAM here

Apache by default on most Linux distros uses a preforking worker module. This spawns multiple processes, which each handle some of the requests.

Apache keeps a number of processes ready and waiting to reduce latency/overhead of forking a new process - these are the extra lines you see in ps.

The processes you see don't directly map 1:1 to users/requests.

http://httpd.apache.org/docs/2.2/mod/prefork.html covers this in a lot more detail.

To check which worker module you are using on Ubuntu/Debian, run apache2 -V and look for the APACHE_MPM_DIR line. Or run dpkg -l | grep apache - each MPM is a different package.

Do you have a lot of virtual hosts? Are you using mod_balance?

We had a system where apache2 was serving as a front-end to a mod_perl/apache13 backend. Apache2 was expending an absurd amount of memory per connection (30-40mb UNSHARED, and we easily had 100's of simultaneous connections). It turns out the way we setup virtual hosts + load balancing wasn't really optimal in an apache2 system: Each one of our 200+ clients are assigned a dedicated port on our apache13 backend, and the front-end uses name-based virtual hosting, and load-balances to one of 4 backend nodes, direct to the 'client port'. Every client required it's own 'balancer' section, on top of it's own vhost section in apache2. nginx requires seperate balancer configurations too, but it seems to handle it far more optimally. We went from 500-600mb total for apache2, to 40-50mb tops for nginx.

Nginx saved us from having to wildly expand our hardware, and we've never looked back. You may want to consider using php-fcgi + nginx or lighttpd in your setup, you may see much better figures for memory usage.