Gillaume,
Here's the difference between those three option:
New Target on the Network -- this is used to debug an application remotely, from Windows to Windows. You run an application on PC #1, while your CVI development environment is installed in, and controlled from PC #2. This can be useful to debug a problem that only happens in one particular computer, the setup of which is hard to duplicate in your primary development computer.
New RT target on the Network -- this is used to run (and debug) an application on a real-time target. This is, what I expect, you will want to do. The "application" in this case is a DLL that has an RTMain entry point.
New RT Target via LabVIEW on the Network -- this is used by developers whose application consists of a LabVIEW VI that calls CVI code via DLLs, and that will run on a real-time target. When you want to to debug your CVI code, you use this option to notify CVI that this code will be called from LabVIEW. CVI will then wait for LabVIEW to run, to download the VI to the target (you configure this part in LabVIEW) and waits until it can establish a connection with the target once the DLL is called. The DLL is a traditional DLL that does not require an RTMain.
You should also check out the following help topics in the CVI online help:
Programmer Reference
Creating and Downloading DLLS to Real-Time Targets
Creating and Downloading Real-Time DLLs from LabWindows/CVI
Using RTmain
Configuring a Real-Time Target
Manually Downloading Support Files
Luis