How do IP video conferencing products typically handle fragmentation when sending a 1500 byte UDP video packet over a GRE data link? Do most IP video product vendors set the DF bit? If yes, is PMTUD typically supported by the IP video server?