I am planning to use QNET-over-IP in our product. Are there any problems with this way of using QNET? Is this fully 
supported configuration (e.g. not experimental)? And what disadvantages it would have in comparison with QNET-over-RAW (
"normal" QNET, connected directly to the Networking drivers)?

I assume, performance would be one of them. But, it is not big issue for us regarding this transport. The primary 
concern is reliability of message delivery (when IP links are re-established) and stability (failure recovery) of the 
configuration in the environment where failures (including communication paths) may happen randomly and often.

The operational runtime environment:
QNX-6.4.1 running on SMP mode (8 CPU cores) with two parallel networking stacks (only one of them uses QNET). The 
Networking Stacks operates under heavy load (UDP/TCP streams).

Thank you.