How meterpreter handles requests?

[junkoi2004 at gmail.com (Jun Koi)]

Hi,

I am still very new to metasploit, so may have some silly questions.
Sorry for that.

I am wondering how meterpreter handles requests sent to it from
client. Exactly how it interacts with the framework, so it can handle
requests.

Here is what I understand: The target has a bug, and can be exploited
remotely. We use meterpreter as a payload, and let the target execute
this palyload. When it executes, meterpreter registeres some "server
extensions" which has some commands (as explained in the paper
"meterpreter.pdf").

Now the client sends a command request to target. A question: does
this use a separate channel to send request, or it uses the same
channel with "normal" traffic? If so, how can the target know that
this is a command for meterpreter, but not for "normal" processing, so
it can execute the corresponding handler?

That is it for now. All the helps are very much appreciated.

Thanks,
J