More to the specifically, Endpoints run just a small "snippet" of a program (called either an agent or a service or a daemon, depending on your background). The snippet runs just enough Tivoli stuff to let it find and connect to a real Tivoli server. Whenever a process (inventory, distribution, etc) has to be done, all of the REAL code (binaries, executables, methods) are pushed through the small snippet and run locally on the machine.
A managed node is part facilitator of communications (between the endpoint and the TMR) and part local distribution point (it has the real code local so everything doesn't always get pushed all the way down the communication pipe. It also holds a local database of information about the endpoints to which it talks. That inforamation gets relayed back to the TMR as needed.
This is an incredible over-simplication, but it should get you started conceptually. Let me know if it helps or hurts.
Leon
Leon Adato (adatole@yahoo.com)
Measure what is measurable,
And make measurable what is not so.
- Galileo