Typically in this type of setup, your central server at HQ won't do anything or will do a very limited set of checks. It just sits there and receives NSCA connections which write service check results to your nagios.cmd file, and sends out service alerts based on those results. I would be more inclined to call your central server the "dummy", as it's not doing much.
Your distributed servers can be compiled minimally. They don't need the web interface or any extra crap. Your distributed nagios processes are the ones actually doing the checking, so they are running and actively checking hosts and then using NSCA to send results back to the central server. In your nagios.cfg in the distributed boxes you set an "OCSP" command. This command is run after every single service check, and it's just a script that passes service check results to send_nsca, which then connects to the central server and sends in the results.
NSCA documentation IS hard to find. Here's a great howto that i used way back when i did it:
nagios.sourceforge.net/download/ ... _Setup.pdf