If you want to monitor that amount of hosts, you really should think this over *very* well and do some proper research/planning, set proper requirements etc before jumping straight to ideas and how you think it should be implemented exactly.
So far my "Captain Obvious"-behavior.
More technical, I guess you don't want to use multiple servers to check if a client is really-really-really down. There are other possible solutions, like aggregated passive checks, active checks, client-side custom checks, scripting, etc. Still, you need a proper plan before you start building. Swinging into one direction as first step (without thinking it over properly) can cost you!
My basic advice: get help from a Zabbix professional, preferably certified for large Zabbix deployments.
If you have more questions, just let us know