Hi All,

I have been looking at tuning the Apache Config, and while i have read alot, im a little confused and was looking for clarity on a couple of things.

Context.

I have an application that serves XML request/response over a farm of say 10 application servers.
The load is balanced over this farm by 2 Apache webservers (Apache/2.2.3)

This is a very high throughput application, with millions of requests per day.

Load Testing the application shows that per application server, 20 concurrent requests is what each worker can deal with optimally.

Im using the worker MPM.

Question.

Can i limit my requests based on workers? eg; i want a max of 20 concurrent requests per worker in Apache.

The current configuration, and pages from Apache seem to be based on total number of max connections and child processes. Eg: 10 appservers x 20 connections = allow 200 connections and load balance.
My issue is that if one server fails, then the load goes from 20 to 22.22 per server, leading to performance degradtion, which may backup, cause more failures etc.

Many thanks,