I promise you my little server was getting more traffic than that, and since patio11 -- another vocal "turn off KeepAlive" advocate -- personally commented in that thread that turning off KeepAlive directly resolved their site outage, I think I'm going to stick with what I've got.
The server is already using mpm-worker, as well as fcgid carefully tuned to make PHP nice and snappy without eating all of the server's available memory.
Final point: site response times aren't the only metric a server admin should care about. A server admin should also -- and maybe even primarily -- care about concurrent connections. If I can serve thousands of simultaneous connections at 6.87s or hundreds of simultaneous connections at 5.31s, guess which one I choose?
I promise you my little server was getting more traffic than that, and since patio11 -- another vocal "turn off KeepAlive" advocate -- personally commented in that thread that turning off KeepAlive directly resolved their site outage, I think I'm going to stick with what I've got.
The server is already using mpm-worker, as well as fcgid carefully tuned to make PHP nice and snappy without eating all of the server's available memory.
Final point: site response times aren't the only metric a server admin should care about. A server admin should also -- and maybe even primarily -- care about concurrent connections. If I can serve thousands of simultaneous connections at 6.87s or hundreds of simultaneous connections at 5.31s, guess which one I choose?