Hi Mohammed,
AFAIK, an ICT has no other limitations in terms of latency than a normal H.323 trunk would have. From that perspective, the ICT should work fine but what you might notice is a certain delay in call setup as the signaling has to traverse a long distance. There is not much you can do there apart from working on the network itself and try to get fastest links in terms of latency. Even putting a Gatekeeper will not help much as the packets still need to travel the network from one signaling point to the other.
Now, what I said above applies to signaling. For the communication itself (the RTP stream), IP packets do not have to traverse the CUCM nodes but rather are sent from IP endpoint to IP endpoint directly. And that's what you really need to care about. You need to make sure that the network path that connects the two communicating endpoints meets the requirements for latency and delay as specified in the ITU standard.
Now, from experience working in customers around the world, users are ok with call setup taking some time if they're calling the other side of the world. They can even bare a bit of a delay in the communication for the same reason although the more you can reduce that, the better it'll be. But again, users are very aware that if they're calling Australia from the UK, there might be a delay due to the geographical distance...
Cheers,
Jose
