Re: New Internet Draft on automatic (end-user) tunneling for SSM

[Ross Finlayson <finlayson@live.com>]

>    o Also, justify why you're not using GRE.

Dino (et al),

Let me apologize for my somewhat flippant response to your earlier question 
about not using GRE (especially since at the time I had forgotten that you 
were one of GRE's authors :-) :-)

I do remember considering GRE when I first designed UMTP (back in 1997).  I 
took another look at the GRE spec just now, and I can now recall why I 
chose not to use GRE - the space overhead would have been too great.

 From my understanding of GRE (please correct me if I'm wrong), the 
overhead to tunnel an arbitrary UDP multicast packet using GRE would be 32 
bytes: 4 bytes (for the GRE header), plus 28 bytes (for the UDP/IP headers 
of the encapsulated packet).

In contrast, UMTP adds an overhead of just 16 bytes (when carrying a SSM 
packet - it'd be 12 bytes for ISM).  Furthermore, by making this a trailer, 
rather than a header, RTP/UDP/IP header compression - if applied - will 
continue to work on the entire packet.  (The space savings could be 
significant if the UMTP packets traverses a low-bandwidth link...)

(Also, the same 16-byte trailer turns out to be useful for sending control 
packets (in the reverse direction from data) - e.g., JOIN_GROUP (including 
periodic 'keep alive's) and LEAVE_GROUP messages.)

         Ross.